Serval
截至 2026-06-02,基于公开来源的 Serval 尽调
Serval 获得了极强投资人背书,也有可信的 AI 原生 ITSM 架构;但收入增长说法没有基数支撑,客户全是科技创业公司,团队不足 30 人,单位经济完全未披露,仅靠公开证据无法支撑 $1B 估值。
封面要素
公司概况
Serval 是一家 2024 年成立、总部位于旧金山的 AI 原生 ITSM 初创公司,部署两个协作的 AI 智能体——Help Desk Agent 和 Automation Agent——为企业 IT 团队自动处理工单解决、即时访问开通、员工入职和离职。2025 年 10 月,公司完成由 Redpoint Ventures 领投的 $47M Series A;仅七周后,Sequoia Capital 又领投了一轮先发制人的 $75M Series B,投后估值 $1B,累计披露融资达到 $127M。公司公开称 2025 年 8 月至 12 月收入增长 500%,但未披露绝对 ARR。已披露客户包括 Perplexity、Together AI、Mercor、Vercel、Verkada、Clay 和 Cribl,全部是技术原生公司。Series B 交割时员工约 30 人,公司称计划到 2026 年底超过 100 人。
- 成立时间
- 2024-01-01
- 创始人
- Jake Stauch, Alex McLeod
- 创立地点
- San Francisco, CA
- 总部
- San Francisco, CA
- 产品
- Serval 平台把五个模块——帮助台 / 工单、工作流自动化、访问管理、资产管理,以及处理升级工单的 AI copilot——放在双智能体架构下。Help Desk Agent 通过 Slack、Teams、电子邮件或网页门户处理员工请求,并调用预先批准的确定性工具;Automation Agent 让 IT 管理员用自然语言构建这些工具,并以版本化代码保存。第三个 Insights Agent 从工单模式中发现自动化机会。Serval 公开保证 50% 的 IT 工单自动化率,支持 60+ 原生集成,并提供云、混合和完全自托管部署,满足 SOC 2 Type II、HIPAA 和 GDPR 合规。
- 客户
- 面向有复杂 IT 自动化需求的技术原生公司和风险投资支持的扩张期企业;当前披露客户全部来自 AI / 软件领域。平台定位服务于中型市场到企业级买家,既可替换完整 ITSM,也可作为现有系统上的 AI 层。
- 商业模式
- 企业 SaaS,采用定制报价、仅询价定价和高接触入职模式(专属部署工程师、四阶段激活)。年订阅合同有暗示但未确认。50% 工单自动化保证支撑商业话术。公司未公开定价底线、毛利率或单位经济性。
- 阶段
- Series B private company
- 融资情况
- 2025 年 12 月完成 $75M 先发制人 Series B,投后估值 $1B,由 Sequoia Capital 领投,Redpoint Ventures、Meritech Capital 和 General Catalyst 继续参投。$47M Series A 于 2025 年 10 月完成,由 Redpoint 领投,并有广泛联合投资方参与。累计披露融资:$127M。
执行摘要
主要优势
- 企业 ITSM 史上最快跻身独角兽;Sequoia 抢先领投 Series B,并把早期客户信号类比 ServiceNow。
- 双智能体架构(工具构建者与工具执行者分离)给企业安全叙事留出防线,降低 AI 治理风险。
- 7 家具名、可核验的技术原生客户,且独立材料从第一天起就印证 50%+ 自动化率。
主要风险
- 2025 年 8–12 月 500% 收入增长说法完全没有锚点;公司未披露绝对 ARR 或 MRR,核验和同业对标都无从做起。
- CEO Jake Stauch 身上的关键人集中度极高;除 co-founder Alex McLeod 和一名战略项目员工外,没有具名 C-suite、董事会或工程负责人。
- 客户全是 AI 创业公司和技术原生公司;能否跨到大企业、满足受监管行业合规,以及已融资客户收缩时的流失风险,都还是严肃问题。
未决问题
- 需要绝对 ARR 或 MRR 来锚定 500% 增长说法,并评估收入质量、队列留存和可持续增长率。
- 单位经济——CAC、LTV、回本周期、毛利率和净收入留存——完全未公开。
- 在 EDGAR 中未找到 Serval, Inc. 的 SEC Form D 文件;按披露融资规模看,这一点反常。
- $1B Series B 价格下的优先股堆叠、清算瀑布和股权结构条款。
- 具名的大企业或非科技客户;当前具名名单全是 VC 支持的 AI 和软件公司。
目录
01公司概况
1.1 身份、创立与运营模式
Serval, Inc. 是一家 AI 原生 IT 服务管理(ITSM)公司,总部位于 CA 94104 旧金山。公司由 Jake Stauch(CEO)和 Alex McLeod 于 2024 年创立,使命是部署能够自主解决帮助台请求、开通访问、办理员工入职,并通过自然语言构建工作流自动化的 AI 智能体,替代传统工单和工作流工具。公司是一家在 Delaware 注册、获得风险投资支持的私营公司,对外将产品定位为「面向 IT 的 AI 智能体」。 Serval 将其产品描述为一个 AI 原生统一平台,组合帮助台、访问管理和自动化引擎。公司的差异化架构使用两个不同的 AI 智能体:Help Desk Agent 通过 Slack、Teams、电子邮件或网页门户与员工交互,并通过调用预先批准的确定性工具解决请求;Automation Agent 允许管理员用自然语言指令构建这些工具——CEO 称之为「IT 自动化的 vibe coding」。第三个 Insights Agent 分析工单模式,发现自动化机会和配置改进点。把工具构建和工具执行分开,是 Serval 企业安全姿态的核心:Help Desk Agent 只能调用预授权自动化,避免不受约束的智能体采取破坏性动作这类「失控 AI」事故。 截至报告日期,Serval 的商业模式是企业 SaaS,采用定制、按报价定价。公司提供云托管、混合和自托管部署选项,目标客户是技术公司以及有复杂 IT 自动化需求的企业。其 GTM 策略同时支持完整替换传统 ITSM 工具,以及为被长期合同锁定的客户在现有系统上叠加「AI 层」。Serval 的注册实体为 Serval, Inc.,产品条款使用的法律域名是 serval.com。[CO001, CO002, CO003, CO004, CO005, CO006]
| 指标 | 数值 / 状态 | 日期 | 置信度 | 缺口 / 备注 |
|---|---|---|---|---|
| 成立 | 2024 | 2024-01-01 | 中 | 具体成立月份未公开确认 |
| 总部 | San Francisco, CA 94104 | 2026-06-02 | 高 | |
| CEO | Jake Stauch | 2026-06-02 | 高 | |
| 联合创始人 | Alex McLeod | 2026-06-02 | 中 | 公开材料中的角色细节稀少 |
| 阶段 | Series B 独角兽 | 2025-12-11 | 高 | |
| 最新估值(USD) | $1 billion | 2025-12-11 | 高 | 无股权结构表或独立验证 |
| Series A(USD) | $47 million | 2025-10-21 | 高 | |
| Series B(USD) | $75 million | 2025-12-11 | 高 | |
| 累计融资(USD) | $127 million | 2025-12-11 | 高 | |
| 员工数 | 不到 30 人 | 2025-12-11 | 中 | 公司目标是 2026 年超过 100 人 |
| 收入增长(自 2025 年 8 月以来) | 500% | 2025-12-11 | 中 | 绝对收入未披露 |
| 自动化率主张 | 第一天处理 >50% 工单 | 2026-06-02 | 中 | 基于公司说法;无独立审计 |
| 合规认证 | 认证 / 合规:SOC 2 Type II、HIPAA、GDPR | 2026-06-02 | 中 | 公司声称取得 SOC 2 Type II;审计报告未公开 |
| 原生集成 | 60+ | 2026-06-02 | 中 | |
| 部署选项 | 云、混合、自托管 | 2026-06-02 | 高 |
估值和员工数来自 Reuters / US News 报道。绝对收入未披露;500% 增长数字为公司声称。合规认证由公司表述,未能从公开审计报告独立验证。
[CO001, CO002, CO016, CO017, CO018, CO019]1.2 创始人、领导层与关键人风险
Jake Stauch 担任 Serval CEO 兼联合创始人。他是公司最主要的公开面孔,出现在所有主要媒体采访中,也是 Serval 产品叙事核心「vibe coding」框架的提出者。Alex McLeod 是另一位具名联合创始人,Redpoint 投资组合页面将其列为联合创始人,但 McLeod 尚未在面向公众的报道中高频出现。截至报告日期,公开材料中没有其他具名 C-suite 高管、董事会成员或顾问,反映出公司仍处于极早期,员工数不足 30 人。 Redpoint 投资组合页面列出三位赞助 Series A 交易的 Redpoint 合伙人:Alex Bard、Patrick Chase 和 Jordan Segall。General Catalyst 的 Serval 投资组合页面列出参与的 GC 投资人为 Marc Bhargava、Vedant Suri 和 Kate Bender。Sequoia 的 Serval 页面将 Anas Biad 列为 Series B 的主导合伙人——Biad 公开表示,客户反馈之强可比 Sequoia 十六年前投资 ServiceNow 前听到的反馈;以 Series B 新闻周期看,这是一个异常直接的类比。 2026 年 5 月推出的 Serval Start 项目,是创始团队雄心的另一个信号。这个两年制项目招募技术型 builder 到 Serval 工作,并把它作为未来创办自己公司的路径;工作坊由 Stauch、Christine Kim(Head of Strategic Projects)、Bill Trenchard(First Round 合伙人)、Greg Rosen(BoxGroup 合伙人)、Anas Biad(Sequoia 合伙人)和 Patrick Chase(Redpoint 合伙人)主持。该项目既是人才获取和品牌建设动作,也显示 Serval 的高层领导梯队仍然很小——除 Stauch 外,Christine Kim 似乎是唯一具名的运营高管。Stauch 身上的关键人集中,是尽调的核心关注点。[CO009, CO010, CO011, CO012, CO013, CO014]
| 人物 | 角色 | 背景 | 创始人-市场匹配 / 职能 | 关键人物依赖 |
|---|---|---|---|---|
| Jake Stauch | CEO 兼联合创始人 | 企业软件和 AI;Serval 产品叙事与“vibe coding”定位的主要设计者 | 产品愿景、融资、客户关系、公开叙事 | 关键——唯一公开面孔;所有重要外部沟通都经由 Stauch |
| Alex McLeod | 联合创始人 | Redpoint 组合公司页面列名为联合创始人;其他公开信息有限 | 未知——可能是技术或获客端联合创始人 | 中等——已确认在公司,但具体职能和依赖度未建立 |
| Christine Kim | 战略项目负责人 | Serval Start 工作坊材料列名;未披露此前公司背景 | 战略运营和项目管理 | 低到中——具名高管,但公开材料中的职能范围有限 |
管理层表覆盖不完整;截至运行日期,公开材料中未出现董事会成员、CFO、CTO 或 VP 级高管姓名。Serval Start 工作坊把投资人合伙人(非 Serval 员工)与两名具名 Serval 员工并列展示。
[CO009, CO010, CO013, CO014, CO015]1.3 融资历史、估值与投资方地图
即便按 2025 年风险投资标准看,Serval 的融资节奏也很快。公司于 2025 年 10 月完成 $47 million Series A,由 Redpoint Ventures 领投,First Round Capital、General Catalyst、BoxGroup、Bessemer Venture Partners、Meritech Capital、Strike Capital、Sunflower Capital 和 Operator Partners 参投,另有一批天使投资人参与。Reuters 报道引用的 PitchBook 数据显示,Series A 对 Serval 的估值约为 $232 million。 仅三个月后,2025 年 12 月,Serval 完成由 Sequoia Capital 领投的 $75 million Series B——Anas Biad 公开评论称这是一轮先发制人的融资——Redpoint、Meritech 和 General Catalyst 也参投。Series B 使累计融资达到 $127 million,公司估值达到 $1 billion,Serval 因此成为独角兽。三个月内估值上调 4 倍(从约 $232M 到 $1B),同时伴随 Serval 声称自 2025 年 8 月以来收入增长 500%,但公司没有披露绝对收入数字。 Serval 计划把 Series B 资金用于 GTM 和工程岗位招聘,目标是从不足 30 名员工扩张到超过 100 人。公司的投资方包括具备深厚企业软件经验的机构:Sequoia 投资过 ServiceNow,而 ServiceNow 是 Serval 的主要 incumbent 竞争对手;Redpoint 主导了交易;General Catalyst、First Round 和 Bessemer 共同补齐蓝筹投资方阵容。截至报告日期,公开报道中没有二级流动性、债务或信贷额度证据。[CO016, CO017, CO018, CO019, CO020, CO021]
| 利益相关方 | 角色 / 轮次 | 经济 / 控制相关性 | 尽调要求 |
|---|---|---|---|
| Redpoint Ventures | Series A 领投 | Series A 领投方,可能有董事席位;Alex Bard、Patrick Chase、Jordan Segall 被列名 | 确认董事席位和治理权利 |
| Sequoia Capital | Series B 领投(抢先) | Series B 领投方;Anas Biad 为具名合伙人;组合中有强 ServiceNow 经验 | 确认董事席位;了解抢先投资条款和反稀释条款 |
| General Catalyst | Series A 和 B 参与方 | Marc Bhargava、Vedant Suri、Kate Bender 被列名;跨轮跟投强化确信度信号 | 确认持股规模和治理参与 |
| Meritech Capital | Series A 和 B 参与方 | 跨轮参与;企业 SaaS 专家 | 确认经济条款和董事会观察员权利 |
| First Round Capital | Series A 参与方 | Bill Trenchard 被列名;早期投资专家,参与 Serval Start 工作坊 | 确认持股,以及任何董事 / 观察员权利 |
| BoxGroup | Series A 参与方 | Greg Rosen 被列名;早期通用型投资人 | 确认持股规模 |
| Bessemer Venture Partners | Series A 参与方 | VCTavern 报道列名;企业 SaaS 专家 | 确认参与情况和持股规模 |
| Strike Capital / Sunflower Capital / Operator Partners(投资人) | Series A 参与方 | VCTavern 报道列名 | 确认角色和条款 |
| 天使投资人 | Series A 共同投资人 | 被描述为拥有构建和规模化企业软件的经验;身份未公开 | 索取名单并确认无利益冲突 |
持股规模、所有权比例和董事会构成未公开披露。Sequoia 抢先 Series B 暗示 Series B 投资人可能拥有强控制或优先权条款。表格仅覆盖已确认的公开提及;完整股权结构表为私有信息。
[CO016, CO017, CO018, CO019, CO020, CO021]从创立到 Series B 的融资和估值里程碑时间线。
2025 年 8 月 Series A 前基线为近似值;2024 年内具体创立日期未公开确认。
[CO001, CO016, CO017, CO018, CO019, CO020]1.4 产品、技术与合规
Serval 平台由五个模块组成:Resolve Requests(帮助台工单和 AI 解决)、Build Workflows(自然语言自动化构建器)、Transform Ticketing(AI 原生工单,并与 ServiceNow、Jira、Zendesk 双向同步)、Manage Access(JIT 访问开通与治理)和 Manage Assets(跨 MDM、采购、IdP 和 HRIS 的统一资产管理)。每个模块都连接到 Serval 的中央 AI 智能体基础设施。 底层技术架构值得关注:工作流自动化在无代码界面背后生成代码,让非技术 IT 经理和工程团队都能在 Git 中检查和管理自动化。这种「底层是代码」的做法,意在提供确定性代码工作流的可靠性和可审计性,同时移除传统 ITSM 构建器的拖拽摩擦。平台支持 60+ 原生集成,包括 Slack、Okta、Google Workspace、GitHub、Jira、ServiceNow、AWS、Confluence 和 Kandji。 Serval 将安全和合规作为核心产品属性来销售:公司声称拥有 SOC 2 Type II 认证、符合 HIPAA,并满足 GDPR。平台对传输中数据使用 TLS 1.3,对静态数据使用 AES-256,并提供基于角色的访问控制、审计日志、SAML/SCIM 支持和 SIEM 集成。公司提供灵活部署,包括通过 Terraform 和 Kubernetes 进行自托管。Serval 的 Acceptable Use Policy(AUP)公开可查,并明确禁止使用该平台与 Serval 竞争,这与标准企业 SaaS 条款一致。公司以引导式试点项目和专属部署工程师支持,保证 50% 自动化率。[CO025, CO026, CO027, CO028, CO029, CO030]
| 模块 | 功能 | 核心能力 | 集成面 |
|---|---|---|---|
| Help Desk Agent | 用对话式 AI 解决员工请求 | 处理 JIT 访问、回答知识库问题、执行自动化;必要时升级 | Slack、Teams、电子邮件、Web 门户 |
| Automation Agent | 用自然语言构建工作流自动化 | 生成确定性的代码型自动化;支持无代码和代码检查模式 | Git、CI/CD、API |
| Insights Agent | 分析工单模式并给出建议 | 建议新增自动化、知识库更新和配置改进 | 分析仪表盘、内部运营 |
| Access Management | JIT 访问开通与治理 | SCIM、API、基于 Terraform 的开通;限时且受策略门控的访问 | 集成系统:Okta、JumpCloud、Google Workspace、Rippling |
| Ticketing | AI 原生工单,带升级和双向同步 | 为升级工单提供 AI copilot;与第三方系统完整双向同步 | ServiceNow、Jira、Zendesk、Freshservice |
| Asset Management | 跨采购、MDM、IdP、HRIS 的统一资产跟踪 | 自动对账 MDM + 采购 + IdP + HRIS;去重 | Jamf、Kandji、MDM 供应商 |
功能描述来自 serval.com 首页和定价页。所有能力均为公司声称;没有第三方评测或独立产品审计公开可用。
[CO025, CO026, CO027, CO028]Serval 的 AI agents 如何与企业系统交互,以解决 IT 请求。
[CO005, CO006, CO007, CO026, CO027]1.5 客户牵引、GTM 与规模
Serval 公开确认的客户包括 Perplexity、Together AI、Mercor、Vercel、Verkada 和 Clay——均为企业买家谱系中偏 AI 原生的技术公司。公司称,其平台从客户使用第一天起就能自动处理超过 50% 的入站 IT 工单;官网案例研究引用了 IT 负责人 Vernon Man 的客户证言:「切换到 Serval 后,我们能自动完成超过 50% 的入站请求。」公司未公开具名任何 F500 企业客户,所有已确认客户都是技术前沿公司,可能无法代表完整的企业 IT 买家版图。 根据 Reuters 引用的公司说法,2025 年 8 月至 12 月收入增长 500%,但未披露绝对数字。对于一家 2025 年 10 月完成 Series A 的公司,2025 年 8 月的收入基数很可能仍处早期——因此 500% 作为牵引轨迹信号有意义,但缺少绝对分母时,不能单独作为增长指标。客户数量未公开披露。 GTM 主要是直销,配套结构化部署项目:Phase 1(Meet)、Phase 2(Build)、Phase 3(Deploy)、Phase 4(Optimize)。公司采用询价制、仅企业级定价;不提供自助服务或 SMB 层级。Serval Start 项目显示出更长期的招聘和品牌策略,活跃博客与案例研究发布则说明公司有内容驱动的需求生成动作。截至 2026 年 6 月,博客提到关于 Eesel 和 Siit 替代方案的对比文章,说明除旗舰企业动作外,公司也在瞄准特定中型市场 ITSM 买家细分。[CO032, CO033, CO034, CO035, CO036, CO037]
截至运行日期的关键财务和运营 KPI。
收入增长和自动化率均为公司声称数字,未获独立验证。客户数量仅反映公开点名的客户标识;客户总数未披露。
[CO019, CO021, CO022, CO023, CO028, CO032]1.6 里程碑时间线与反向检查
Serval 的公开记录覆盖从创立到成为独角兽的大约两年。截至报告日期,公开记录中未发现针对 Serval、其创始人或具名高管的监管行动、法律程序、制裁或执法行动。公开材料中未出现产品宕机、安全事件或数据泄露通知。任何公开来源中也未发现裁员或已报道的领导层离职。 主要反向尽调点是结构性的,而非事件驱动:公司披露 500% 增长时拒绝给出绝对收入;在员工不足 30 人、没有公开财务验证的情况下获得独角兽身份;确认客户名单局限于技术公司,没有公开的 F500 或受监管行业参考客户;其核心竞争论点(替换 ServiceNow 及类似在位厂商)将持续面对 ServiceNow 自身 AI 智能体产品发布的竞争。Gartner Peer Insights 的 AI ITSM 市场列出了既有厂商已验证的企业部署,Serval 缺席这一验证轨道,对受监管或大型企业买家构成实质企业信任缺口。 公开材料中未出现已知创始人利益冲突、未决诉讼或 FINRA/SEC 相邻问题。招聘页面将创始团队描述为「精英 builder 和销售团队」,符合一家扩张前、正在招聘的初创公司状态。公司融资节奏极快、独角兽叙事激进,这些高风险声誉承诺会放大任何执行失误的后果。[CO038, CO039, CO040, CO041, CO042]
| 日期 | 事件 | 类型 | 金额 / 估值 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2024 | Serval 由 Jake Stauch 和 Alex McLeod 创立 | 创立 | Jake Stauch、Alex McLeod | 确立公司身份和创始团队 | |
| 2024 | 产品开发和初始客户试点 | 产品 | 内部团队 | Pre-seed 开发阶段;该阶段无公开融资或媒体报道 | |
| 2025-08(约) | 收入基线确立;Series A 前增长参考点 | 规模化 | 用作 Series B 时 500% 增长主张的基线 | ||
| 2025-10-21 | Series A 完成公告 | 融资 | $47M,估值约 $232M | Redpoint(领投)、First Round、GC、BoxGroup、Bessemer、Meritech、Strike、Sunflower、Operator Partners、天使投资人 | 确立机构背书;TechCrunch 和其他一线媒体覆盖 |
| 2025-12-11 | Series B 完成公告(抢先) | 融资 | $75M,估值 $1B;累计融资 $127M | Sequoia(领投)、Redpoint、Meritech、General Catalyst | 独角兽状态;Sequoia 抢先显示投资人需求异常强 |
| 2025-12 | 公司员工数不到 30 人;计划 2026 年扩至 100+ 人 | 规模化 | 显示独角兽规模下团队极精简;招聘速度是近期执行风险 | ||
| 2026-05-18 | 面向有志创始人的 Serval Start 项目公告 | 产品 | Jake Stauch、Christine Kim、投资人合伙人 | 品牌建设与人才获取项目;首期申请已截止 | |
| 2026-06-01 | Serval 博客发布“Three Operating Principles” | 治理 | 首次公开明确文化 / 运营原则;释放成熟化信号 | ||
| 2026-06-02 | 公开记录未发现不利监管、法律或运营事件 | 反向 | 无不利事件是正面尽调信号,但也反映公司年轻 |
时间线由公开媒体、公司博客和投资人组合公司页面构建。2024 年内的具体成立日期未公开确认。收入基线和 Series A 前里程碑为推断;2025 年 10 月前没有公开媒体报道。
[CO001, CO016, CO017, CO018, CO019, CO020]1.7 附录
02市场分析
2.1 市场定义、范围与相邻领域
Serval 的产品切入 IT 服务管理软件市场。广义上,ITSM 是接收、路由、自动化并解决员工内部服务请求的软件层。IBM 将 ITSM 定义为规划、实施、管理和优化信息技术服务端到端交付的实践,以满足用户和业务目标,涵盖事件管理、问题管理、变更管理、服务请求管理和 IT 资产管理。这个边界已经成熟:它排除 IT 监控和可观测性(衡量基础设施健康状况,而不是工作流交付)、端点管理(没有请求层的设备开通)以及企业资源规划(掌管业务流程,而不是 IT 服务交付)等相邻类别。 在 ITSM 内部,Serval 竞争的是新兴 AI 原生子细分:用大语言模型和智能体自动化处理端到端请求解决,而不只是路由或分类。相比传统工单厂商,这一子细分仍处早期,但正吸引 incumbent(ServiceNow 的自主 AI specialist 项目、BMC 的 HelixGPT、Freshservice 的 Freddy AI)和风险投资支持的初创公司(Aisera、Moveworks、Atomicwork、Serval)大量投入。Serval 的可服务市场不包括纯可观测性平台(Datadog、PagerDuty)、DevOps 工具链(GitHub Actions、CircleCI)以及不包含帮助台层的独立身份治理工具(SailPoint、Saviynt)。 Enterprise Service Management(ESM)是一个重要相邻方向,它把 ITSM 模式延伸到 HR、Finance 和 Legal 工作流。Serval 平台已经支持 IT、Security、HR 和 Workplace 团队的多部门隔离,这意味着若公司推进,ESM 扩张是一个合理增长向量,并会扩大可服务市场。Serval SAM 纳入的支出包括 AI 原生 ITSM 平台席位、实施服务和自动化工作流额度,面向技术前沿行业中 50–10,000 名员工的组织。[CM004, CM005, CM006, CM007, CM013, CM015]
| 细分 / 类别 | 纳入支出 | 排除支出 | 买方 / 付款方 | 与 Serval 的相关性 |
|---|---|---|---|---|
| AI 原生 ITSM 平台 | 面向智能体 IT 帮助台、访问管理和入职自动化的按席位与按用量订阅 | 没有 AI 解决层的纯工单 SaaS;传统本地部署服务台软件 | IT 负责人、IT 主管、掌握 IT 运营预算的 CIO | 核心直接市场;Serval 的主要产品面 |
| 传统 ITSM 软件 | 云和本地部署服务台、事件管理、变更管理和 ITAM 授权 | 不包含请求层的基础设施硬件、端点管理和网络监控 | IT 主管、CIO,以及掌握多年期 IT 平台预算的采购 | 替换机会;Serval 必须替代或共存的在位玩家 |
| 企业服务管理(ESM) | 面向 HR、Finance、Legal 和 Workplace 团队的跨职能请求自动化 | 不含 IT 请求工作流的核心 ERP 模块、薪资系统和会计软件 | CHRO、CFO、总法律顾问,以及拥有职能预算线的部门负责人 | 相邻扩张市场;Serval 的多团队隔离支持 ESM 增长 |
| AIOps 和 IT 运营 | AI 驱动的事件关联、事件管理和智能告警平台 | 不含请求路由的网络监控、可观测性仪表盘和性能管理 | IT 运营经理和 SRE 团队 | 相邻;在自动修复工作流上重叠,但 Serval 不覆盖基础设施监控 |
| 身份与访问治理 | 智能体访问开通、just-in-time 访问和访问审查自动化 | 独立 PAM(特权访问管理)工具,以及不含请求工作流的身份治理 | IT 安全经理、CISO 和合规团队 | Serval 平台内的关键能力;也是独立治理软件市场 |
| IT 帮助台外包 | 第三方处理 IT 支持工单的托管服务台合同 | 人员补充、离岸 IT 支持团队和 IT 咨询项目 | COO、拥有外包预算的 IT 主管 | 间接竞争压力;Serval 自动化外包商靠人工配置的工作 |
市场边界基于 Serval 的产品面和 IBM 的 ITSM 定义。支出估计来自行业报告和公开定价推断;截至研究日期,没有单一权威市场规模覆盖 AI 原生 ITSM 子细分。
[CM004, CM006, CM007, CM010, CM013, CM037]ITSM 与 IT 运维 AI 市场顶部很宽,但 Serval 可触达的细分市场,是总盘子里聚焦于云端和 AI 原生部署的一小块。
这是一组语境镜头,不是严格的 TAM-SAM-SOM 级联;三项出版方数字采用不同市场定义,不能相加。SAM 行是自下而上的估算,没有一手来源,置信度低。
[CM001, CM002, CM003, CM040]2.2 TAM、SAM 与 SOM 测算视角
最宽口径的 ITSM TAM 来自 MarketsandMarkets,其预计全球 ITSM 市场到 2028 年将达到 $22.1 billion,以 2023 年为基准,复合年增长率为 15.9%。该数字覆盖所有垂直行业和组织规模的本地部署与云 ITSM 解决方案,包括服务台软件、变更和配置管理、运营与性能管理以及 IT 资产管理。同一数据源预计 AIOps 平台市场到 2028 年将达到 $32.4 billion,CAGR 为 22.7%,反映的是更广义的 AI-in-IT-operations 空间,包含纯 ITSM 之外的可观测性、事件关联和自动化工具。 更窄的视角是云 ITSM 子市场,它与 Serval 产品最直接相关。MarketsandMarkets 另行预计,云系统管理(包含 ITSM、ITOM 和 ITACM)将从 2020 年的 $10.6 billion 增长到 2025 年的 $31.4 billion,对应 CAGR 24.1%。这一增长率强化了 ITSM 向 SaaS 交付迁移的趋势,而 Serval 的云原生和自托管部署模式正面向这一趋势。Freshservice 公开披露的 74,000+ 企业客户数提供了需求侧交叉验证:它说明中型市场 ITSM 已经是成熟且广泛采用的类别,要替换 incumbent,必须有足够有说服力的差异化故事。 Serval 的 SAM 最适合通过把 ITSM 市场筛到 AI 原生、具备高自动化潜力的部署来估算。基于 Serval 当前客户画像(快速增长的技术公司、50–2,000 名员工),SAM 很可能只是整体 ITSM 市场的一部分,今天或许为 $2–4 billion,并随着 AI 原生 ITSM 替代以工作流为中心的前代产品而以 30%+ 增长。SOM 未见公开记录,也无法从现有证据中分离;本报告将其视为开放测算缺口。公开估计中的关键矛盾包括:ITSM 头部数字($22.1B,窄口径软件)与云系统管理数字($31.4B,定义更宽)之间的差距,以及 AIOps 数字($32.4B)包含 Serval 并不覆盖的可观测性和事件管理。[CM001, CM002, CM003, CM017, CM040, CM041]
| 发布方 | 预测年份 | 地理范围 | 市场价值 | CAGR | 方法 / 范围 | 置信度 | 对 Serval 的局限 |
|---|---|---|---|---|---|---|---|
| MarketsandMarkets | 2028 | 全球 | $22.1B | 15.9% | 仅 ITSM 软件;包括服务台、变更管理、运营管理、ITAM;本地部署和云 | 中 | 宽泛定义包含本地部署传统厂商;Serval 可触达份额只是小子集 |
| MarketsandMarkets | 2028 | 全球 | $32.4B | 22.7% | AIOps 平台市场;包括可观测性、事件关联、ITSM 和自动化工具 | 中 | 夸大 SAM;包含大型可观测性厂商(Dynatrace、Datadog),Serval 不与其竞争 |
| MarketsandMarkets | 2025 | 全球 | $31.4B | 24.1% (2020–2025) | 云系统管理;包括 ITSM、ITOM、ITACM;仅云部署 | 中 | 比纯 ITSM 更宽;可作为云优先市场视角;基准年为 2020 |
| Freshservice (Freshworks) | 2026 | 全球 | 74,000+ 客户 | 定价页披露客户数;中端市场 ITSM 需求侧直接数据点 | 高 | 客户数而非收入;不能隔离 AI 原生与传统 ITSM 采用 | |
| Serval(公司声称) | 2026 | 全球(美国主导) | 累计融资 $127M;估值 $1B | Series B 公告中的融资和估值;可作为投资人评估市场规模的代理 | 中 | 估值反映投资人预期,不是已实现收入;没有公开收入数字 | |
| 估计 SAM(推导) | 2026 | 全球 | $2B–4B(估计) | >30%(估计) | 自下而上估计:面向 50–10,000 人技术行业公司的 AI 原生 ITSM;由 ITSM TAM 和 AI 采用率推导;并非来自一手来源 | 低 | 作者推导估计;无一手来源;只能视为粗略数量级 |
所有美元数字均为 USD。MarketsandMarkets 数字来自 2024 年 1 月报告;CAGR 和预测年份按原文发布。SAM 行为推导估计,非一手发布方来源,不应作为事实引用。Freshservice 客户数为自披露且未经审计。
[CM001, CM002, CM003, CM017, CM028, CM040]AI 相邻的 IT 市场板块增长快于传统 ITSM;Serval 可触达的利基市场位于 CAGR 分布的高端。
MarketsandMarkets 各行的低值和高值,是作者围绕已发布中值给出的不确定性区间;并非单独发布的上下限。AI 原生 ITSM 这一行由作者推导估计,应视为示意,而非事实。
[CM001, CM002, CM003, CM041]2.3 买方、用户与付款方分层
ITSM 购买动作本质上由 IT 主导,中央 IT 预算掌握采购。在中型市场公司(100–2,000 名员工),通常由 Head of IT 或 IT Director 评估并选择平台;在更大型企业中,CIO 或 VP of IT Infrastructure 负责采购治理,并让安全和法务参与。Serval 的早期客户——Perplexity、Mercor、Together AI 和 Cribl——主要是快速增长的技术公司,Head of IT 管理精简团队,需要把大部分例行工作自动化,而不是继续加人。这一买方画像与 TechCrunch 对 Serval 价值主张的描述一致:公司把自动化做得足够轻松,以至于永久构建一个工作流比手工做一次还容易。 ITSM 平台用户包括广义员工(提交请求的所有员工)和处理升级工单的 IT 专员。Serval 的智能体架构意味着,大多数终端用户交互发生在 Slack、Microsoft Teams、电子邮件或网页门户中,由 AI 智能体处理解决;只有请求超出定义好的自动化范围时才升级给人工。这减少了一级支持所需的 IT「席位」数量,使 Serval 的 ROI 论证围绕自动化率,而不是单纯围绕单席成本。 付款方几乎完全是 IT 预算线;随着 ESM 用例被激活,潜在可扩张到 HR、Finance 或 Legal 预算负责人。Freshservice 的多产品结构(Freshservice for Business Teams、Freshservice for MSPs)显示,随着 ESM 成熟,市场已经接受跨部门预算碎片化。ServiceNow 模式主要面向大型企业,付款方是企业 IT 组织,合同金额可达数百万美元。Serval 保证 50% 自动化率,并采用专属部署工程模式,说明其销售动作会先从定义明确的试点中建立 ROI,再扩张到完整部署,这是典型的企业软件 land-and-expand 模式。[CM008, CM009, CM023, CM024, CM025, CM027]
| 细分 | 买方 | 用户 | 付款方 | 工作流语境 | 预算所有者 | 采用触发点 |
|---|---|---|---|---|---|---|
| 快速增长的技术初创公司(50–500 人) | IT 负责人或 IT 经理 | 全体员工;1–5 人 IT 团队 | CFO 或 CTO 控制的 IT 预算 | 帮助台被 SaaS 开通、入职 / 离职请求压垮 | CTO 或 COO | 团队翻倍;IT 不能线性扩张;自动化成为成本刚需 |
| 成长期技术公司(500–2,000 人) | IT 主管或 IT 副总裁 | 500–2,000 名员工加 IT 运营团队 | G&A 下的 IT 预算线;偶尔与安全团队共用 | SaaS 蔓延带来访问管理复杂度;SOC 2 合规审计要求 | CIO 或 CFO | SOC 2 或合规审计触发访问审查项目;自动化成为合规路径 |
| 企业级科技公司(2,000–10,000 名员工) | CIO 或 IT 运营副总裁 | 员工分布在多个地点,规模达数千人 | 中央 IT 预算;可能向业务部门交叉分摊 | 需要压低服务台人头,同时维持或提升 SLA 表现 | CIO 或 CFO(TCO 优化) | 降本要求;FTE 人头压力;ServiceNow 成本上升 |
| 受监管垂直行业企业(医疗、金融、法律) | CIO 或 IT 总监,安全 / 合规共同采购 | 员工需要访问敏感数据 | IT 预算叠加安全 / 合规预算 | HIPAA、GDPR 或金融服务监管要求数据访问与审计日志 | CISO 与 CIO 共同采购 | 监管审计发现或合规审查;SOC 2 Type II 要求 |
细分边界来自 Serval 披露的客户(Perplexity、Mercor、Together AI、Cribl)以及典型 ITSM 市场分层。预算归属与采用触发因素参考公开 ITSM 文献和竞品定价证据。公开资料没有 Serval 特定 ACV 数据。
[CM009, CM011, CM023, CM025, CM029, CM030]高速增长的科技公司同时具备高 AI 准备度、精简 IT 团队和强 ROI 紧迫感,因此是最容易切入的初始滩头阵地;受监管行业摩擦更高,但长期潜力更大。
分段边界和准备度评级是作者基于市场证据作出的判断;研究日期时,没有可用的一手调查数据能按细分市场展示 IT 买方 AI 准备度。
[CM011, CM023, CM029, CM030, CM031, CM036]2.4 推动 AI 原生 ITSM 采用的需求驱动因素
四股结构性力量正在压缩 AI 原生 ITSM 的采用时间线。第一,企业 SaaS 蔓延正在指数级增加访问开通和离职交接事件:公司每引入一个新的 SaaS 工具,IT 都要管理访问策略、凭证开通和定期访问审查。Serval 的 60+ 工具集成目录——覆盖 Okta、Google Workspace、GitHub、AWS、Rippling、Workday、Jira、ServiceNow 和 Zendesk——直接应对这种碎片化。第二,企业 AI 投资正在加速底层 LLM 能力成熟,而智能体 IT 工作流需要这些能力。ServiceNow 把自主 AI specialist 定位为核心平台功能(走向「zero-touch service」),Aisera 的企业证据(60–70% 自动解决率、某客户节省 $2.2M 支持成本)也验证技术已经可以进入企业生产部署,降低了 2023–2024 年普遍存在的买方怀疑。 第三,快速增长公司里的精简 IT 团队形成紧迫感:一家每 12 个月员工数翻倍的公司,无法线性扩大 IT 人员,因此自动化是成本必需品,而不是可选创新支出。第四,围绕最小权限访问(零信任、即时开通)的监管压力,让手工访问管理成为合规风险。Serval 的智能体开通、自动撤权和限时访问,直接对应 NIST 和 SOC 2 控制要求;过去这些要求需要 IT 手工介入。[CM010, CM011, CM014, CM016, CM030, CM033]
| 因素 | 方向 | 时点 | 对 Serval 的含义 | 尽调问题 |
|---|---|---|---|---|
| SaaS 蔓延推高 IT 工单量和访问复杂度 | 加速因素 | 当前;未来 1–3 年继续恶化 | 单个客户可覆盖的工作负载更大;需要更多集成才能覆盖完整技术栈 | Serval 每个客户平均使用多少个 SaaS 工具?集成覆盖如何演进? |
| 智能体 AI 成熟,让端到端工单解决无需人工介入 | 加速因素 | 当前;LLM 能力按季度提升 | 拉高自动化率;降低客户疑虑;同时逐步抬高质量预期 | Serval 已部署客户的自动化率中位数是多少?趋势如何? |
| 高增长公司精简 IT 团队的经济账 | 加速因素 | 当前;高增长公司存在结构性需求 | 形成拉动式需求,不必完全靠销售推动;客户会主动识别痛点 | Serval 管线中自然流入与主动拓客各占多少? |
| 最小权限访问与 JIT 开通的合规要求 | 加速因素 | 当前;NIST CSF 2.0 和 ISO/IEC 27001 修订后继续强化 | 访问自动化从可选项变成合规要求;预算责任人从 IT 扩展到更多职能 | Serval 是否已纳入客户 SOC 2 或 ISO 27001 控制映射?它能否作为访问日志的记录系统? |
| 企业 AI 投资预算转向运营自动化 | 加速因素 | 当前;2026 年预算周期把 AI 运营工具列为优先项 | AI 原生 ITSM 更容易获批预算;相较 2023 年缩短销售周期 | Serval 平均销售周期多长?过去 12 个月如何变化? |
| 围绕智能体 AI 作用范围的安全与治理顾虑 | 约束 | 当前;持续存在 | 需要确定性、带权限边界的架构;加重评估负担;拖慢采购 | 面向重视安全的买方,Serval 如何证明动作范围控制?它是否已为大型部署完成企业安全问卷? |
| 既有 ITSM 锁定(ServiceNow、Freshservice、Jira SM) | 约束 | 当前;多年合约周期 | 客户可能先把 Serval 作为补充层采用,再考虑完整替换;限制全面替换速度 | Serval 如何与现有 ITSM 平台共存?双向工单同步会不会侵蚀自身收入? |
| AI 平台的企业采购周期长 | 约束 | 当前;通常 3–12 个月 | 拖慢新增客户获取;更利于从现有客户先落地再扩张 | Serval 试点转付费率是多少?平均扩张时间线多长? |
| 受监管市场的数据驻留与主权要求 | 约束 | 当前;欧盟和美国受监管行业继续强化 | 自托管部署是进入受监管垂直行业的前提;支持起来运营复杂度更高 | Serval 管线中有多少比例要求自托管部署?适用哪些 SLA 和支持模型? |
| 新 AI-ITSM 预算项的定价与 ROI 不确定 | 约束 | 短期;随着市场成熟会减弱 | 初始交易需要强 ROI 锚点(自动化率保证、每张工单成本节省) | Serval 如何衡量并报告自动化率?50% 保证是否可审计且无条件? |
方向判断基于公开市场证据和 Serval 的产品定位;公开资料没有 Serval 内部管线、销售周期或赢 / 输率数据。时点来自作者对市场证据的判断。
[CM010, CM011, CM013, CM014, CM016, CM025]2.5 采用摩擦、合规约束与采购障碍
尽管需求顺风强劲,几股力量仍会拖慢 AI 原生 ITSM 采用。围绕 AI 动作范围的企业安全和治理要求,是最显著的摩擦点。TechCrunch 关于 Serval 的报道点出了核心企业顾虑:组织对失控 AI 智能体采取意外动作的风险高度敏感,因此 Serval 架构把工具构建与工具执行分离,并强制采用确定性、经授权的动作,而不是单一全能智能体。这是真正的架构差异化,但也意味着潜在客户购买前必须理解并信任一个以 AI 为中心的安全模型——这不是轻量评估障碍。 ITSM 平台的采购周期很长,因为它有大量集成要求。Serval 的入职模式包含四个阶段(meet、build、deploy、optimize)和一名专属部署工程师,这意味着潜在客户在评估自动化率保证前,可能要经历数周到数月的企业销售周期。ServiceNow、BMC 和 Freshservice 等传统 ITSM 平台拥有广泛合作伙伴生态和专业服务关系,形成切换成本和惯性。合规认证要求也抬高门槛:SOC 2 Type II、HIPAA 和 GDPR 是企业采购的标准要求;Serval 文档称三者均满足,但截至研究截止时,独立审计证据并未公开。 对受监管行业(医疗、金融服务、政府)而言,数据驻留和主权是实质约束,因为 ITSM 数据包含员工 PII、访问日志和敏感工作流历史。Serval 的云托管、混合和自托管部署选项原则上应对数据驻留,但自托管支持和 SLA 保证细节没有公开记录。最后,定价模型依赖专属部署工程师,意味着很小的公司(50 人以下)可能得不到 Serval 当前 GTM 动作的充分服务,从而限制 SAM 低端。[CM011, CM012, CM025, CM026, CM029, CM030]
企业采用 AI-ITSM 会按阶段推进:认知、试点、集成,再到全面部署;每个阶段都有明显流失,主要由安全审查、集成复杂度和 ROI 验证要求驱动。
漏斗百分比是用于可视化的序数权重,代表作者对各阶段相对流失的判断;它们不是 Serval 发布的转化指标。Serval 没有公开可用的管线转化数据。
[CM009, CM012, CM025, CM026, CM046, CM047]2.6 附录
03竞争格局
3.1 格局与玩家类型
2026 年,AI 原生 IT 服务管理市场围绕四类玩家展开:主导企业合同的传统在位平台(ServiceNow、BMC Helix),以自动化深度挑战在位厂商的 AI 原生点解决方案(Serval、Moveworks、Aisera),把开发者工具延伸到 ITSM 的工作流自动化混合体(Atlassian Jira Service Management),以及把 AI 功能加到现有工单产品中的传统 SaaS 帮助台厂商(Freshservice、Zendesk)。Serval 同时与四类玩家竞争:它分别对标 ServiceNow 的复杂度和成本、Moveworks 的自动化主张、Freshservice 的 SMB 价格点,以及 Atomicwork 的从初创到企业叙事。截至 2026 年 Q1,ServiceNow 仍是主导力量,约渗透 85% 的 Fortune 500,全球拥有 7,700 多家企业客户,但正受到资本密集型 AI 原生替代方案挑战。Freshservice 以公开的按席位定价服务 74,000 多家企业,从 SMB 到企业级,锚定市场可进入端。Jira Service Management 以 ITIL v4 对齐和 PinkVERIFY 认证瞄准 DevOps 相邻团队,并借助更广的 Atlassian 生态实现跨产品锁定。Moveworks 以面向 HR、IT 和设施等员工服务的企业 AI copilot 切入市场,支持 100 多种语言,并用 Reasoning Engine 处理多步骤智能体工作流。Atomicwork 和 Aisera 攻击的也是与 Serval 相同的 AI 自主叙事:Aisera 入选 Gartner Magic Quadrant for AI Applications in ITSM,并声称自动解决率为 65% 至 90%。BMC Helix 被评为 Forrester Wave Leader for Enterprise Service Management Q4 2025,代表在位厂商投入 AI 的投资逻辑。Serval 的差异化建立在三点上:合同 SLA 保证 50% 自动化、500% ARR 增长验证快速市场牵引,以及多数 AI 原生同业缺少的完整部署灵活性(云、混合、自托管)。[CP001, CP002, CP003, CP004, CP005, CP006]
| 厂商 | 类别 | 成立时间 | 员工数(估算) | 总融资额 | 目标客群 | AI 深度(1–4) | 关键认证 |
|---|---|---|---|---|---|---|---|
| Serval AI | AI 原生 ITSM | 2024 | ~30 | $127M | 高速增长科技公司 | 4 | SOC2、HIPAA、GDPR |
| ServiceNow | 传统平台 | 2004 | ~23,000 | 上市公司(NOW) | Fortune 500 企业 | 3 | FedRAMP、SOC2、ISO 27001 |
| Freshservice | SaaS 帮助台 | 2010 | ~5,000 | $250M+ | SMB 至中端市场 | 2 | SOC2、ISO 27001 |
| Jira Service Mgmt | DevOps 混合型 | 2002 | ~12,000 | 上市公司(TEAM) | 开发相邻团队 | 2 | PinkVERIFY、SOC2、FedRAMP |
| Moveworks | 企业 AI 副驾驶 | 2016 | ~500 | $315M+ | 大型企业 | 3 | SOC2、ISO 27001 |
| Atomicwork | AI 原生 ITSM | 2022 | ~100 | 未披露 | 科技优先的中端市场 | 3 | SOC2、ISO 27001 |
| Aisera | 对话式 AI ITSM | 2017 | ~400 | $100M+ | 企业 IT / HR | 3 | SOC2、HIPAA、GDPR |
| BMC Helix | 传统平台 | 1980 | ~6,000 | 上市公司(BMC Software) | 大型企业 | 3 | FedRAMP、SOC2、ISO 27001 |
融资金额为截至 2026 年 6 月已披露轮次的累计值;员工数为 LinkedIn 和公司页面的近似估计;AI 深度采用 1–4 分,4 代表完全自主且保证解决。
[CP002, CP003, CP004, CP005, CP006, CP007]3.2 能力与产品对比
Serval 的核心能力是端到端自主工单解决:其 AI 智能体接收请求,识别所需动作(通过 Okta 开通访问、关闭 Jira issue、更新 Slack channel),通过 150 多个原生集成执行工作流,并在无人介入下关闭工单。其他被评估厂商没有在 SLA 中合同化保证 50% 自动化下限。ServiceNow 的 NowAssist AI agents 原生运行在 Now Platform 上,具备 ML 模型编排、CMDB 耦合和深度变更工作流管理,但需要大量配置投入,也没有公开自动化保证。Freshservice 的 Freddy AI 提供 AI 辅助工单分流和坐席辅助建议,标准层级尚未达到完全自主解决。Jira Service Management 包含 Atlassian Intelligence,用于请求分类和虚拟智能体支持,但 AI 深度明显低于 Serval 的自主层级。Moveworks 的 Reasoning Engine 支持多步骤智能体工作流,并声称需要人工介入的工单减少 40%——这是有意义的主张,但它是相对减少,不是绝对下限保证。Atomicwork 的 Atom AI copilot 处理自然语言路由,无需手工分类;Aisera 的对话式 AI 和 RPA 集成支撑其声称的 65% 至 90% 自动解决区间。Serval 的部署灵活性使其区别于多数 AI 原生竞争对手:云、混合和自托管配置让它能在数据驻留要求排除纯 SaaS 的受监管行业竞争。ServiceNow 高价提供本地部署;Freshservice、Moveworks 和 Atomicwork 在标准层级都是纯云。Serval 持有 SOC 2 Type II、HIPAA 和 GDPR 认证,使其能够追逐纯云竞争对手无法进入的受监管行业交易。[CP011, CP012, CP013, CP014, CP015, CP016]
| 能力 | Serval | ServiceNow | Freshservice | JSM | Moveworks | Atomicwork | Aisera |
|---|---|---|---|---|---|---|---|
| 自主端到端解决 | 是 | 部分 | 否 | 否 | 部分 | 部分 | 部分 |
| 有保证的自动化 SLA | 是(50%) | 否 | 否 | 否 | 否 | 否 | 否 |
| ITIL v4 对齐 | 部分 | 是 | 是 | 是 | 部分 | 部分 | 部分 |
| 变更与发布管理 | 部分 | 是 | 是 | 是 | 否 | 部分 | 否 |
| 自托管 / 本地部署选项 | 是 | 是(高级版) | 否 | 否 | 否 | 否 | 否 |
| 150+ 原生集成 | 是 | 是(450+) | 是(1,000+) | 是(3,000+) | 部分 | 部分 | 部分 |
| SOC2 / HIPAA / GDPR | 是 | 是 | 部分 | 是 | 是 | 部分 | 是 |
能力评估基于截至 2026 年 6 月的公开产品页面;「部分」表示能力有限或需加购;「否」表示标准套餐不包含。
[CP011, CP012, CP013, CP014, CP015, CP016]3.3 定价、定位与 GTM
ITSM 市场定价跨度很大,从公开的按席位 SaaS 层级,到完全不透明的企业合同。Freshservice 最透明:四个公开层级从每坐席每月 $19 到 $99,支持自下而上的 PLG 采用。Atlassian JSM 为最多三名坐席提供免费层级,Premium 扩展到每坐席每月 $44.27。Zendesk Suite 以每坐席每月约 $55 锚定中型市场入口。Atomicwork 跳出按席位定价,Professional 方案为每员工每年 $90——这是面向技术团队的总拥有成本话术,因为所有员工都可能是服务请求者。ServiceNow 和 Moveworks 不公布标价;定制企业合同估计分别从每年 $200,000 和 $250,000 以上起。Aisera 定制报价,最低承诺 12 个月。Serval 不公布定价;其 GTM 瞄准 Perplexity、Together.ai、Mercor 和 Cribl 等高速技术公司——这些 AI 优先组织看重自动化率保证,超过功能广度。TechCrunch 2025 年 10 月的 Series A 报道明确把 Serval 描述为 ServiceNow 替代品,这会提高 ServiceNow 企业销售团队的竞争关注。ServiceNow 的分销依赖 7,500 个合作伙伴生态和认证实施伙伴——这是 Serval 以 30 名员工无法复制的结构性渠道优势,在 5,000 人以上企业细分中形成不对称分销挑战。[CP021, CP022, CP023, CP024, CP025, CP026]
| 厂商 | 入门层级 | 入门价格(USD) | 最高层级价格 | 定价模型 | 最低合约 |
|---|---|---|---|---|---|
| Serval AI | 定制 | 未披露 | 未披露 | 按合约定制 | 年付 |
| ServiceNow | ITSM Standard | 定制(估算每年 $200K+) | 定制 | 定制企业方案 | 多年期 |
| Freshservice | Starter | $19 / 坐席 / 月 | $99 / 坐席 / 月(Enterprise) | 按坐席 SaaS | 月付或年付 |
| Jira Service Mgmt | 免费 | 最多 3 个坐席 $0 | $44.27 / 坐席 / 月(Premium) | 按坐席 SaaS | 月付或年付 |
| Moveworks | 定制 | 定制(估算每年 $250K+) | 定制 | 定制企业方案 | 年付 |
| Atomicwork | Starter | 定制 | $90 / 员工 / 年(Professional) | 按员工年费 | 年付 |
| Aisera | 定制 | 定制(12 个月最低) | 定制 | 定制企业方案 | 年付 |
| Zendesk Suite | Suite Team | ~$55 / 坐席 / 月 | 定制(Enterprise) | 按坐席 SaaS | 月付或年付 |
公开定价截至 2026 年 6 月;需要定制报价的厂商显示分析师或市场估算;Serval 未公开定价,记为定制报价。
[CP021, CP022, CP023, CP024, CP025, CP026]3.4 切换成本、护城河与竞争耐久性
2026 年的 ITSM 竞争格局,一边是在位厂商的耐久锁定机制,另一边是 AI 原生玩家正在形成的数据飞轮护城河。ServiceNow 的切换成本位居企业软件最高之列:客户要面对 CMDB 迁移、自定义工作流重构,以及 ITIL 认证管理员再培训——这是一个多年、数百万美元级过程。Atlassian 通过跨产品依赖加深锁定:同时使用 Jira Software、Confluence 和 JSM 的团队,退出成本会叠加。Serval 的护城河尚早,但结构上成立:其 150 多个原生集成形成数据和工作流耦合,自动化数据集随着每个已解决工单累积,随时间提升模型准确性;没有同等部署历史的新进入者很难复制。Moveworks 的多租户 AI 模型基于客户通信历史训练,部署 12 到 18 个月后会产生专有学习锁定。$19-billion-plus 的 ITSM 市场足够大,短期内多个 AI 原生玩家和在位厂商可以共存,而不会进入赢家通吃。Gartner 截至 2025 年将 AI Applications in ITSM 市场归为高流失、快速新进入者子细分,说明商品化压力会持续存在。Atomicwork 获得客户证言,被称为面向技术团队的 ServiceNow 替代方案,直接重叠 Serval 的替代叙事,并带来渠道冲突风险。BMC Helix 的 Forrester Leader 认定证明在位厂商仍在大举投入 AI,威胁 Serval 的差异化窗口。Serval 的 SOC 2、HIPAA 和 GDPR 认证在受监管行业提供护城河,因为 Atomicwork 和部分 AI 原生同业缺少同等资质。快速资本注入——Moveworks $200M Series C、Serval 累计 $127M——加剧了同一可寻址 TAM 上的竞争。相对于 ServiceNow 的 23,000 多名员工,Serval 30 人团队在交付能力和合作伙伴支持上存在集中风险。[CP031, CP032, CP033, CP034, CP035, CP036]
| 风险因素 | 严重度 | 证据 | 缓释因素(Serval) |
|---|---|---|---|
| ServiceNow CMDB 锁定 | 高 | 7,700+ 家企业客户深度绑定工作流;多年合约很常见 | 瞄准尚未嵌入 ServiceNow 的绿地科技公司 |
| Atlassian 生态交叉锁定 | 中 | JSM 加 Jira Software 加 Confluence 的捆绑,让 DevOps 团队退出成本叠加 | 定位为 DevOps 原生团队的覆盖层或直接替代品 |
| AI 商品化压力 | 高 | Gartner 提示新进入者流失率高;Atomicwork 和 Aisera 瞄准同一 AI-ITSM 买方 | 合同约定的 50% 自动化 SLA 与不断累积的自动化数据集,构成难以复制的底线 |
| 既有厂商 AI 反向投资 | 高 | ServiceNow NowAssist、BMC Helix 为 Forrester 领导者;大型既有厂商大力投入 AI | 相比配置繁重的既有厂商实施,Serval 强调部署速度和有保证的结果 |
| TAM 内资本竞争 | 中 | Moveworks C 轮 $200M;Serval 总融资 $127M;多家 AI 原生厂商同步融资 | $1B 估值和 Sequoia 背书,传递一线支持与竞争续航能力 |
| 团队与规模集中风险 | 中 | Serval 只有 30 名员工,而 ServiceNow 超过 23,000 名;伙伴基础设施有限 | 加速招聘,并借助 Sequoia 网络引入大型企业 |
严重度按高 / 中 / 低评估;证据引用截至 2026 年 6 月的公开来源;缓释因素反映 Serval 当前能力和定位。
[CP031, CP032, CP033, CP034, CP038, CP039]04财务情况
4.1 收入模型与定价可见度
Serval 在公开定价页上呈现「一个平台费,没有意外收费」定价模型,但未公布实际价格卡。所有购买都需要预约销售演示,所有部署都从带专属部署工程师的引导式试点开始。四阶段入职序列——Meet、Build、Deploy、Optimize——意味着高接触、服务辅助的激活模式,更像中型市场和企业软件厂商,而不是产品驱动增长公司。Serval 公开保证 50% 的 IT 工单自动化率,这是一个异常大胆的商业承诺,也说明其销售动作以绩效为锚,而不是纯使用量或席位模式。 产品集成 60 多种企业工具,包括 Okta、Google Workspace、GitHub、Jira、Slack、Freshservice 和 ServiceNow。这一广度意味着先落地再扩张模型:初始部署覆盖一个工作流,随后随时间扩展到更大的 IT 表面。双智能体架构——一个智能体理解 IT 请求,一个智能体执行动作——是自动化保证的技术基础,但该架构如何转化为确认收入,商业层面并不公开可见。年订阅合同有暗示但未确认。未披露按席位、按坐席或按用量定价层级。没有价格卡、毛利率或合同条款,定价权仍只是从投资人信心中推断,而不是可衡量的商业事实。 [CI001, CI002, CI003, CI004, CI005, CI006]
| 收入流 | 定价机制 | 公开证据 | 收入质量判断 | 尽调问题 |
|---|---|---|---|---|
| 平台订阅 | 一个平台费,仅报价,无公开价目表 | 定价页面确认「一个平台费,没有意外收费」表述 | 可能是经常性年度合约;收入确认政策未知 | 披露合约期限、ACV 区间和收入确认方法。 |
| 引导式试点 / 上线服务 | 包含在平台费内,或另收实施费;未披露 | 定价页面描述 4 阶段上线,并配备专属部署工程师 | 可能是单独专业服务收入,也可能完全打包 | 说明实施是否单独收费,还是由平台费吸收。 |
| 自动化保证履约 | 50% 自动化率保证;未说明罚则或抵扣机制 | 定价页面把 50% 自动化率保证列为产品承诺 | 如果保证触发抵扣,会影响确认收入和毛利率 | 说明 SLA 抵扣机制、历史索赔率和 P&L 处理。 |
| 扩张 / 坐席或工作流加购 | 未公开披露;可能绑定更多工作流或智能体部署 | 集成页面列出 60+ 个工具,意味着扩张集成面很宽 | 结构上可行但尚未验证;NRR 和扩张经济性未知 | 提供净收入留存率,以及扩张收入占总收入比例。 |
所有收入流假设均来自公开营销材料推断。公司未公开价格卡、合同模板或收入确认政策。绝对 ARR 未知。
[CI001, CI002, CI003, CI004, CI005, CI006]Serval 从创立到成为独角兽大约只用了 18 个月,两轮披露的风险融资相隔七周完成。即便按 AI 初创公司的标准,这条时间线也很罕见。
[CI010, CI012, CI013, CI015, CI016, CI021]4.2 融资时间线与资本位置
Serval 的资本历史以速度见长。公司成立于 2024 年,到 2025 年 12 月已通过两轮主要融资累计获得 $127M 披露风险投资。$47M Series A 由 Redpoint Ventures 领投,于 2025 年 10 月完成,First Round Capital、General Catalyst、BoxGroup、Bessemer Venture Partners、Chemistry VC、Strike Capital、Sunflower Capital 和 Operator Partners 参投。仅七周后,Sequoia Capital 领投了投后估值 $1B 的 $75M Series B,使 Serval 成为企业软件史上最快进入独角兽行列的公司之一。 估值轨迹尤其醒目。Reuters 文章引用的 PitchBook 数字显示,Serval 在 2025 年 8 月的隐含估值为 $232M——同一个夏天,公司首次披露 500% 收入增长说法。到 2025 年 12 月,估值达到 $1B,约三个月内上调 4.3 倍。公司未披露债务额度、可转债或项目融资义务。据报道,资本部署目标是工程人员和 GTM 招聘。公司计划从 Series B 时约 30 名员工扩张到 2026 年底超过 100 人,这意味着烧钱速度会显著加快,也使执行高度依赖于维持吸引 Sequoia 领投的增长率。EDGAR 系统中未找到 Serval, Inc. 的 SEC Form D 备案;对于一家接受如此规模豁免证券发行的美国公司,这并不常见。 [CI010, CI011, CI012, CI013, CI014, CI015]
| 轮次 | 金额 | 交割日期 | 领投方 | 投后估值 | 主要参与方 |
|---|---|---|---|---|---|
| 种子轮(估算) | ~$5M(推算) | 2024(未披露) | 未披露 | 未披露 | 未披露;由 $127M 总融资额减去 Series A 和 B 金额推算。 |
| Series A | $47M | October 2025 | Redpoint Ventures | $232M(2025 年 8 月 PitchBook 引用) | 投资人:First Round Capital、General Catalyst、BoxGroup、Bessemer VP、Chemistry VC、Strike Capital、Sunflower Capital、Operator Partners。 |
| Series B | $75M | December 2025 | Sequoia Capital | $1.0B | 现有投资者;Reuters 报道引用了 Sequoia 合伙人 Anas Biad 的说法。 |
| 已披露总融资 | $127M | 截至 2025 年 12 月 | — | $1.0B(Series B 投后) | VCTavern 与 Reuters/US News 交叉印证。 |
| 规划员工数目标 | 2026 年底超过 100 名员工 | 前瞻性 | — | — | Series B 交割时约 30 名员工;资金投向工程和 GTM 招聘。 |
Reuters 文章引用 PitchBook 数据,称 $232M 估值对应 2025 年 8 月。EDGAR 未见 SEC Form D 备案。种子轮金额为推算值($127M 总融资 - $47M Series A - $75M Series B = $5M),若存在过桥轮或其他工具,仍可能修订。
[CI010, CI011, CI012, CI013, CI014, CI015]Serval 的商业模式从企业 IT 问题流向 AI 自动化解决,收入在平台层捕获。高接触导入推高成本;自动化保证是差异点,但其财务机制未披露。
[CI001, CI002, CI003, CI004, CI005, CI006]4.3 牵引信号与单位经济性
Serval 最强的公开商业证据来自已披露客户名单和收入增长主张。公司公开列出 Perplexity、Together AI、Mercor、Vercel、Verkada、Clay 和 Cribl 为客户。这些都是技术原生公司,其中大多数本身也是有风投支持的初创公司或扩张期企业。Sequoia 投资组合页面以相同措辞独立佐证了这一组客户。技术原生客户集中有助于证明早期采用者牵引,但也提出企业跨越问题——Serval 能否卖给受监管行业、大型金融服务公司或政府机构;这些客户的合规要求、变更管理和采购周期严苛得多。 Serval 在 Reuters/US News 文章中报告自 2025 年 8 月以来收入增长 500%。该主张未同时披露 ARR 或 MRR 数字,因此无法把增长率锚定到起始基数。一家公司从 $100K ARR 增长到 $600K ARR 也是 500% 增长,但其风险画像与从 $5M 增长到 $30M 的公司完全不同。单位经济性——获客成本、客户终身价值、回本周期、毛利率和净收入留存——全部未公开。定价模型为询价制,没有披露底价或上限。自动化保证(50% 工单)是商业承诺,但公开材料没有描述罚金或 credit 机制。SOC 2 Type II 认证和 99.9% uptime SLA 提供企业可信度信号,但不能替代披露的财务指标。 [CI019, CI020, CI021, CI022, CI023, CI024]
| 厂商 | 定价模式 | 入门价格(公开) | AI / 自动化主张 | 目标客群 | 对 Serval 的竞争风险 |
|---|---|---|---|---|---|
| Serval | 平台费,仅报价 | 未披露 | 保证 50% 工单自动化 | 科技原生公司;向企业级扩张 | 本分析对象。 |
| ServiceNow(AI / Now Platform 平台) | 订阅制,企业协商价 | 非公开;通常 $10K-$200K+ ACV | AI agent 原生嵌入 Now Platform | 大型企业、IT 运维团队 | 高——存量厂商可向 7,000+ 企业客户无增量成本打包 AI。 |
| Freshservice(Freshworks) | 按 agent 订阅 | 标价:$15/agent/month(Growth),$40+(Pro) | Pro 及以上套餐包含 Freddy AI assistant | SMB 至中端市场 | 中——透明分层定价形成价格锚的替代方案。 |
| Atlassian Jira Service Management | 按 agent 订阅 | 有免费层;$17.65/agent/month(Standard) | Cloud 套餐含 AI 功能 | 开发者中心团队;向 ITSM 扩张 | 中——开发者生态锁定叠加已知品牌。 |
| Moveworks | 企业合同,仅报价 | 未披露 | 面向 IT 和员工服务的对话式 AI | 企业,1000+ 名员工 | 高——AI ITSM 定位直接重叠;进入市场更久。 |
ServiceNow 和 Moveworks 定价基于分析师估算和公开描述的交易区间,不是官方价格卡。Freshservice 和 Atlassian 定价来自各自公开定价页。
[CI027, CI028, CI029, CI030, CI031, CI032]整体 ITSM 市场给 Serval 留下很大的长期机会,但 AI ITSM 子板块仍只是总市场支出的一小部分。ServiceNow 和老牌厂商目前掌握多数企业席位。
所有数字均为 MarketsandMarkets 和 Grand View Research 的分析师估计。AI ITSM 子板块估计为推断值;没有供应商发布过经独立验证的 AI 专属 ITSM 市场规模。
[CI028]4.4 竞争性财务格局
ITSM 市场规模大且在增长。MarketsandMarkets 和 Grand View Research 的分析师估计显示,全球 ITSM 市场 2024 年约为 $11B,预计到 2030 年将超过 $30B。AI 赋能 ITSM 子细分增长更快,因为大型在位平台正竞相把智能体自动化与专门的 AI 原生厂商一起整合进产品。ServiceNow 2024 年收入约 $10.98B,市值超过 $150B,已经在 Now Platform 上直接推出自有生成式 AI 和智能体 AI 能力。这一竞争压力是 Serval 最实质的近期战略风险:资源充足的在位厂商可以向现有客户捆绑同等功能且不额外收费,从而压缩 Serval 的商业窗口。 Freshservice 和 Atlassian Jira Service Management 都公布透明定价层级,便于直接对标。Moveworks 和 Aisera 是 AI 优先 ITSM 替代方案,融资历史更早,但 ARR 同样未披露。Atomicwork 将自己定位为带免费入口的 AI ITSM 玩家。BMC Helix 代表正在用 AI 附加功能现代化的传统 ITSM 在位厂商细分。Serval 的差异化——双智能体架构、保证自动化率、引导式试点——作为技术护城河可信,但尚未接受 ServiceNow 资源深度的压力测试。缺少 Fortune 500 企业品牌客户标志,意味着 Serval 超越技术原生细分的跨界吸引力仍未证实,这会实质影响其长期 TAM 捕获潜力。 [CI027, CI028, CI029, CI030, CI031, CI032]
| 募资用途类别 | 公开信号 | 估计优先级 | 风险评级 |
|---|---|---|---|
| 工程人员扩张 | Reuters 文章称计划从 30 人增至 100+ 人;技术岗位可能占主导 | 高 | 中——若 AI 工程师招聘市场持续竞争激烈,执行存在风险 |
| GTM 与销售 | Careers 页面使用“Elite team of builders and sellers”措辞;仅 demo 定价印证销售主导打法 | 高 | 高——企业销售周期长且昂贵;客户数尚未披露 |
| 基础设施与安全 | SOC 2 Type II 认证和 99.9% SLA 承诺需要持续投入;docs.serval.com 提到云托管和自托管选项 | 中 | 中——多部署模型增加基础设施成本复杂度 |
| 客户成功与上线 | 每个试点配专属部署工程师是公司明确的产品承诺;若员工数扩至 3 倍,该模式会推高成本 | 中 | 高——若高接触上线已压缩毛利,扩招会放大这个问题 |
所有资金分配估算均由公开材料和管理层表述推断。公司未披露官方募资用途拆分。烧钱速度和 runway 均无公开信息。
[CI021, CI022, CI024, CI037]4.5 财务风险因素与结论
Serval 面临三层财务风险。第一层是收入不透明:500% 增长由公司自报,未经审计,也没有绝对金额锚点。没有投资人或独立分析师发布过 ARR 估计。烧钱速度、现金跑道和现金头寸完全未公开,资本是否充足只能从 Series B 的融资时间推断。Series B 完成时公司约 30 人,成本基数很低;但到 2026 年底扩至 100 人以上,意味着员工数扩大 3-4 倍,在收入加速兑现之前,月度现金消耗很可能增至三到四倍。 第二层风险是智能体 AI 安全。Acronis 的独立安全研究人员在 Dark Reading 发表的研究显示,智能体 AI 系统——即便构建在现代 LLM 架构上——仍会继承传统软件漏洞,包括提示词注入、权限提升和认证绕过。企业 IT 环境处理敏感凭证、配置状态,以及生产系统的特权访问。Serval 部署一旦发生安全事件,可能严重损害客户信任并带来修复成本,而公开记录中没有量化这些影响。 第三层风险是独角兽时间线被压缩。Serval 估值在三个月内从 $232M 跳到 $1B,是 2025 年企业软件中最快的跃升之一。早期收入阶段的激进升值会在后续轮次制造瀑布分配复杂性、员工股权稀释压力,以及投资人预期对齐难题。没有 SEC Form D 文件,对这一资本规模的美国公司而言虽属小异常,但值得注意。财务结论对增长势头持建设性态度,但在严谨承销估值之前,需要公司私下披露 ARR、毛利率、烧钱速度和客户数。 [CI034, CI035, CI036, CI037, CI038, CI039]
| 指标 | 披露状态 | 来源 | 置信度 | 尽调优先级 |
|---|---|---|---|---|
| 年度经常性收入(ARR) | 未披露;仅有 500% 增长说法 | 公司通过 Reuters 发布的表述 | 很低——缺少基准的相对数字 | 关键——要求提供当前和前期 ARR 及收入确认政策 |
| 毛利率 | 未披露 | 无公开来源 | 无——必须估算 | 关键——高接触上线可能把毛利压到典型 SaaS 水平以下 |
| 烧钱速度与 runway | 未披露 | 无公开来源 | 无——由 Series B 时间接近推断 | 关键——员工数翻三倍计划意味着烧钱会快速加速 |
| 客户数 | 未披露;仅有 7 个具名客户 | 官方网站和 Sequoia portfolio 页面 | 低——具名 logo 由公司选择 | 高——要求提供总 logo 数、中位 ACV 和前 10 大客户集中度 |
| 净收入留存 | 未披露 | 无公开来源 | None | 高——扩张经济性是长期毛利潜力的关键指标 |
| 估值倍数 | 隐含——$1B / 未披露 ARR | Reuters、VCTavern 的融资披露 | 很低——分子已知,分母未知 | 高——未披露 ARR 时无法给倍数找可比基准 |
在分析师记录中的 $1B 估值公司里,Serval 的财务证据缺口属于最完整的一类。所有关键财务指标都需要私下尽调。SEC EDGAR 搜索未发现 Serval, Inc. 的 Form D 备案;作为一家已接受约 $127M 证券融资的美国公司,这也是额外证据缺口。
[CI023, CI026, CI034, CI036, CI037, CI038]Serval 的估值从 August 2025 的 $232M 升至 December 2025 的 $1B,不到四个月跃升 4.3x。即便按 2025 年 AI 标准,这也是异常压缩的独角兽时间线。
August 2025 估值来自 Reuters/US News 文章引用的 PitchBook;Serval 未确认。Series B 的 $1B 估值已由多家独立新闻来源确认。
[CI015, CI016, CI038]4.6 图表
05产品与技术
5.1 产品套件与核心模块
Serval 将自己定位为统一的 AI 原生 ITSM 平台,组合三类功能界面:Help Desk、Access Management 模块和 Automation Engine。公司产品首页把该产品描述为部署在企业 IT 内部的「AI 智能体劳动力」,可处理帮助台请求、发放有时限的访问权限、完成员工入职和离职流程,并用自然语言搭建工作流自动化。公司营销中列出五个明确能力区:Resolve Requests、Build Workflows、Transform Ticketing、Manage Access 和 Manage Assets。 Help Desk 模块是面向用户的界面。员工通过 Slack、Microsoft Teams、电子邮件或网页门户交互。Serval 的 AI 会根据请求内容和上下文路由工单,检索已连接知识库并带来源信息回答问题,执行自动化以在无需人工介入的情况下关闭常见请求,并在需要人工处理时携带完整上下文升级。公司提供与第三方系统的双向工单同步——具体包括 ServiceNow、Jira 和 Zendesk——让企业迁移期间可将 Serval 与既有系统并行运行。 Access Management 模块处理即时(JIT)权限发放、自动撤权、审批工作流定制、访问配置限制,以及与身份提供商的 SCIM 集成。根据定价页面,访问策略可细化到最小权限规则、多审批人关卡和有时限授权。Serval 也列出 SCIM 权限发放,以及基于 Terraform/API 的自定义发放能力。 Analytics and Insights 界面提供请求量、AI 解决率、解决时长和 SLA 合规性的仪表盘。Insights Agent 还会分析工单模式,挖出自动化机会,并建议知识库更新——Serval 文档称之为「把自动化本身自动化」。 第五个界面 Asset Management 在首页作为命名能力出现,但公开文档没有给出实质性描述,其成熟度仍不清楚。[CE001, CE002, CE003, CE004, CE005, CE006]
| 模块 / 资产 | 主要用户 | 状态 / 成熟度 | 关键差异点 | 尽调缺口 |
|---|---|---|---|---|
| Help Desk Agent | 员工(终端用户) | GA——已在具名客户生产环境运行 | 通过 Slack/Teams/email/portal 处理请求;限定使用预先批准的工具 | 未披露解决准确率、幻觉频率、SLA 违约率 |
| Automation Agent | IT 管理员 | GA——自然语言到代码的工作流生成 | 从自然语言生成确定性代码工作流;无拖拽 | 未披露代码质量、错误率和 LLM 提供商身份 |
| Insights Agent | IT 管理员 | GA——持续模式分析 | 自动浮现自动化建议和 KB 更新建议 | 未披露建议采纳率、模型在工单模式分类上的准确度 |
| Access Management(JIT) | IT / 安全团队 | GA——支持 SCIM 的 JIT 开通 | 限时授权、自动回收权限、最小权限配置 | 特权访问 / PAM 用例覆盖不清;无 PAM 厂商引用 |
| Knowledge Base | 员工 / IT 团队 | GA——基于 RAG 的搜索与同步 | 从 Notion、Google Drive、Confluence 同步;基于带来源的上下文回答问题 | 检索准确度、切分策略、引用可靠性未经外部测试 |
| Analytics / Insights Dashboard | IT 负责人 / 管理者 | GA——请求量、解决率、SLA 仪表盘 | 实时展示 AI 解决率和工单吞吐 | 未对标行业标准 ITSM KPI;无同行评审 SLA 数据 |
| Asset Management | IT 管理员 | 首页提及;文档无细节 | 未知——文档或定价页没有产品描述 | 成熟度、范围和 GA 状态完全不清;重大缺口 |
模块状态基于截至 2026-06-02 的官方产品文档和定价页。“GA”表示根据产品描述和具名客户部署判断为普遍可用;成熟度深度仅为公司自称。尽调缺口反映公开技术基准或独立审计缺失。
[CE001, CE002, CE003, CE004, CE005, CE006]| 员工任务 / 请求 | 传统工作流 | Serval 方案 | 宣称收益 | 限制 / 注意事项 |
|---|---|---|---|---|
| 软件访问请求 | 发邮件给 IT → 手动 Okta 开通 → 等待 24–48 小时 | Help Desk Agent 接收 Slack 请求;触发 JIT 访问工作流;自动回收权限 | 即时开通;不积累长期访问权限 | 未展示特权访问审批路由复杂度 |
| 员工入职 | IT 团队用数天手动开通 10–15 个系统 | Automation Agent 运行由 HRIS 事件触发的多步骤入职工作流 | 声称入职任务自动化率 50%+ | 端到端入职覆盖取决于集成广度;复杂组织场景未经验证 |
| 密码重置 | IT 工单 → 手动验证 → 重置邮件;30–60 分钟 | Help Desk Agent 通过现有工具验证并以程序方式重置 | 即时自助解决 | 需要管理员预先搭好工具;初始设置投入未量化 |
| 知识库问题 | 员工手动搜索 Confluence/Notion 或在 Slack 提问 | Help Desk Agent 查询同步后的 KB;返回带来源的答案 | 更快回答,减少工单量 | 企业文档上 RAG 检索准确度未经独立验证 |
| 事件路由 / 升级 | IT 负责人手动分诊;工单夜间堆积 | Agent 带完整上下文分类、路由并升级 | 更快分派;降低分诊积压 | 路由准确度取决于模型质量;误路由可能延误关键事件 |
| 离职 | 跨 10+ 系统手动清单;访问权限残留风险 | Automation 工作流在 HRIS 触发后跨集成系统撤销访问 | 员工离职后立即回收访问权限 | 覆盖仅限已集成系统;未集成 SaaS 应用留下残余风险 |
用例描述来自 Serval 官方产品和定价页。宣称收益为公司自称或由产品描述推断;多数工作流没有第三方案例研究数据。50% 自动化率来自首页客户证言(Vernon Man,Head of IT)和定价页保证;没有受控研究。
[CE007, CE008, CE018, CE019]Serval 的平台分层从顶部的员工交互渠道,到智能体智能、工作流运行时、集成连接器,再到底部基础设施。
分层结构根据官方产品文档、定价页和 AUP 推断。LLM 提供商和底层云基础设施供应商未公开披露,因此省略。
[CE010, CE011, CE018, CE019, CE020]5.2 双智能体架构与工作流引擎
Serval 核心架构差异点在于,它有意把 AI 智能体拆成权限有边界的专业角色。该架构包含三个功能不同的智能体。Automation Agent(供 IT 管理员使用)接收对目标自动化的自然语言描述,在幕后生成确定性的代码工作流,并在无代码 UI 中展示,供审查和管理。技术团队也可以直接检查代码,并在 Git 中管理。Help Desk Agent(供终端员工使用)只能调用 Automation Agent 已搭建并预先批准的自动化来解决请求;它不能在权限集合之外临时行动。Insights Agent 自主运行,分析模式、提出自动化建议,并推荐知识库内容更新。 CEO Jake Stauch 在 2025 年 10 月向 TechCrunch 解释安全逻辑:「你不会希望有人进 Slack 说,‘嘿,我想删除公司所有数据’,然后一个特别热心的 AI 智能体回答,‘好,我来删除所有数据。’它应该说,‘嘿,我没有删除公司所有数据的工具。’」Help Desk Agent 只能调用管理员预先授权的工具,从而把潜在 AI 错误或对抗性提示词的爆炸半径限制在既有授权范围内。 工作流引擎从自然语言生成代码,而不是拖拽式逻辑块。Serval 将其描述为「面向 IT 自动化的 vibe coding」——这是有意的产品叙事,把它类比到更广泛的 vibe-coding 潮流。工作流以代码形式存储,可接入版本控制和 CI/CD 流水线;无代码 UI 则提供非工程师也能理解的呈现。公司称,该工作流引擎支持确定性的多步骤执行,输出可预测、可审计。 公开资料中没有技术白皮书、开源仓库或独立基准测试说明支撑智能体的 LLM 提供商、知识库检索使用的 RAG 流水线、工作流执行延迟特征,或代码生成层的可靠性。对一个主张企业级可靠性和合规性的产品而言,这些都是实质性缺口。NIST AI 100-2(2025)将提示词注入和对抗性输入列为基于 LLM 的智能体系统关键风险——Serval 借助确定性工具门控在架构上缓解这些风险,但无法在 LLM 推理层彻底消除。[CE010, CE011, CE012, CE013, CE014, CE015]
| 层 / 组件 | 作用 | 关键依赖 | 风险 |
|---|---|---|---|
| LLM 推理层 | 支撑三个 agent 的自然语言理解、代码生成和 KB 检索 | 第三方 LLM API 提供商(身份未公开披露) | 提供商宕机、弃用、涨价或安全政策变化会削弱平台 |
| 工作流代码生成引擎 | 将自然语言指令转成确定性代码工作流;以可版本化代码存储 | LLM 代码生成质量和 Serval 专有脚手架层 | 生成代码错误可能引入访问控制 bug;无公开测试套件 |
| Help Desk Agent 运行时 | 只执行预先批准的确定性工具;处理员工请求 | IT 管理员通过 Automation Agent 构建的自动化库 | 自动化库完整度限制解决率;未自动化请求仍会升级 |
| 知识库 / RAG 管道 | 从同步的企业文档中检索并综合答案 | Notion、Google Drive、Confluence 连接器;切分和索引质量 | 文档结构差或陈旧时,检索准确度会下降 |
| 集成层 | 通过 API 连接器接入 60+ 个第三方系统 | 第三方 API 稳定性、速率限制和连接器维护 | 任一连接器故障都会削弱自动化覆盖;无公开连接器 SLA |
| 部署基础设施(云) | 完全托管 SaaS;99.9% uptime SLA | Serval 的云提供商和内部 SRE 能力(不足 30 人团队) | 小团队管理多租户 SaaS,集中度风险上升 |
| 部署基础设施(自托管) | 客户通过 Kubernetes 和 Terraform 管理私有云 | 客户自有基础设施以及 Serval 的 Helm/Terraform 模块 | 客户承担更新、安全补丁和 DR 责任 |
| 审计与日志层 | 全面记录所有 agent 和自动化动作 | Serval 内部日志存储和留存政策 | 留存期和日志导出格式未公开说明 |
架构细节来自官方文档、产品页面和 AUP。LLM 提供商身份和基础设施供应商未公开披露。风险评估由产品描述和通用 agentic-AI 风险文献(NIST AI 100-2,2025)推断。
[CE010, CE011, CE012, CE013, CE014, CE015]员工 IT 请求如何从初次提交,经 Serval Help Desk Agent,走向自动解决或人工升级。
流程根据 docs.serval.com 官方文档、产品主页和 TechCrunch Series A 文章拼出。具体分类逻辑和升级阈值未公开记录。
[CE002, CE003, CE004, CE005, CE007]Serval 交付产品时依赖的关键外部环节,展示各类供应商以及每条依赖关系的风险水平。
依赖图根据集成页、文档、AUP 和产品描述推断。LLM 提供方和云基础设施提供方身份属于未核验假设:关键依赖存在,但身份未公开披露。
[CE013, CE015, CE018, CE019, CE023]5.3 集成、部署选项与 API
Serval 集成页面列出 60+ 个连接器,覆盖身份提供商、HR 系统、云平台、工单工具、设备管理和开发者工具。集成页面可见的已验证命名集成包括:Okta、JumpCloud、Microsoft Graph、Google Workspace、Slack、Jira、Linear、ServiceNow、GitHub、AWS、Rippling、BambooHR、Confluence、Notion、CrowdStrike、Jamf、Kandji、PagerDuty、Opsgenie、incident.io、Rootly、Ramp、Navan、HubSpot、Asana、ClickUp、Figma、Calendly、Ashby、Ironclad、PandaDoc、Persona、Envoy、Sigma Computing 和 Fresh Service,另有自定义应用选项。这些横跨四类功能:Access Management、Knowledge Base、Ticket Syncing 和 Workflows。对一家不到 30 人的公司而言,广度具备竞争力,但深度(每个集成的动作 / 触发器数量)未披露。 部署提供两种模式。Cloud-hosted 由 Serval 全托管,客户无需承担基础设施负担,自动更新,并提供 99.9% 正常运行时间 SLA。Self-hosted 使用 Terraform 和 Kubernetes 部署在客户私有云中,文档称可在数小时内完成资源发放;该选项面向有数据驻留或隐私要求的企业。未找到公开状态页,无法独立验证 99.9% SLA 说法。 技术 API 访问已有文档:Serval 为所有集成提供完整 API 访问,并支持通过 API 对接任意系统的自定义集成。定价页面提到 CLI。Terraform 支持基础设施发放和自定义访问模式。SCIM 被列为身份提供商集成能力。这些能力面向希望把 Serval 嵌入现有 CI/CD 和 IaC 流水线的 DevOps 与平台工程团队。 Serval AUP 明确禁止对服务进行逆向工程、反汇编或推导源代码;禁止使用输出构建竞争性 AI 产品;并要求在依赖 AI 生成输出作出重大决策前进行人工审查——这表明公司承认 AI 的可靠性边界。[CE018, CE019, CE020, CE021, CE022, CE023]
5.4 信任、安全与合规
Serval 文档列出的企业安全姿态包括:加密(传输中 TLS 1.3、静态 AES-256)、组织和团队层面的基于角色访问控制、对所有操作的全面审计日志、SOC 2 Type II 认证,以及定期渗透测试。公司称设有 trust.serval.com Trust Center,但研究时该 URL 返回 404,说明页面可能仍在建设中,或 URL 已变更。AUP 将 trust.serval.com 指定为 AI subprocessors 清单的权威位置。 SOC 2 Type II 对企业销售是有意义的信任信号。它要求独立审计在一个监测期内审查安全性、可用性、处理完整性、保密性和隐私控制;一家不足 30 人的公司取得该认证,说明其很早就优先投入合规基础设施。不过,公司没有公开认证函、范围定义或审计机构名称,SOC 2 覆盖范围无法验证。 AUP 包含具体 AI 透明度承诺:AI 生成输出标记为「Serval AI」,以区别于人工撰写内容;Serval 维护 AI subprocessors 和模型提供商清单;未经明确授权,客户数据不会用于训练通用 AI 或外部机器学习模型;可识别客户材料不会用于跨客户模型训练。这些承诺体现出公司意识到企业采购对 AI 数据治理的要求,但它们是合同承诺,不是独立审计控制。 AUP 还要求对重大决策进行有意义的人工监督(第 4.2 节):明确禁止「在没有有意义人工监督的情况下,对个人作出具有法律或类似重大影响的完全自动化决策」。这是产品层面的护栏,但不构成经认证的 AI 安全机制。NIST AI 100-2(2025)将对抗性机器学习风险——包括提示词注入和数据投毒——列为智能体 AI 系统威胁;Serval 的确定性工具门控方法缓解了动作执行层风险,但没有完全解决推理层的对抗性风险。 公开资料中没有 bug-bounty 项目、已发布 CVE 披露或安全事件历史。对早期初创公司而言这很常见,但对企业安全评估者来说仍是尽调缺口。[CE025, CE026, CE027, CE028, CE029, CE030]
| 控制 / 认证 | 状态 | 范围 | 缺口 / 注意事项 |
|---|---|---|---|
| 传输中加密(TLS 1.3) | 已实施(公司自称) | 客户端与 Serval 服务之间传输的所有数据 | 未引用独立审计;仅见于文档主张 |
| 静态加密(AES-256) | 已实施(公司自称) | Serval 云环境中的静态客户数据 | 自托管加密范围由客户管理;未经独立审计 |
| SOC 2 Type II | 已认证(公司自称) | AICPA 针对安全性、可用性和保密性的 Trust Service Criteria | 未披露公开鉴证函、审计期间或审计机构名称 |
| 基于角色的访问控制(RBAC) | 已实施(文档说明) | Serval 平台内组织级和团队级权限范围 | 无外部评估验证 RBAC 实现稳健性 |
| 全面审计日志 | 已实施(文档说明) | 所有 agent 动作、自动化执行和管理员变更 | 未说明留存期、导出格式和 SIEM 集成 |
| 定期渗透测试 | 持续进行(公司自称) | Serval 委托测试;范围和频率未披露 | 无 bug-bounty 计划;未发现公开 CVE 披露 |
| AI 输出标注(透明度) | 已实施(AUP 承诺) | 平台中 AI 生成回复标注为“Serval AI” | 仅属合同承诺;无第三方审计验证标注完整性 |
| AI 子处理方清单 | 维护于 trust.serval.com(Trust Center) | 使用的 LLM 提供商和 AI 子处理方清单 | trust.serval.com 在研究日期返回 404;清单内容无法核验 |
| 客户数据不用于模型训练 | AUP 承诺 | 未经授权,Customer Materials 不用于训练外部 AI | 仅属合同承诺;无独立审计验证数据管道隔离 |
| GDPR / CCPA 合规 | AUP 提及合规义务 | 按 AUP 第 3.2 条,禁止用途包括违反数据保护法 | 未公开 DPA 条款、数据驻留地图或子处理方协议 |
所有安全控制均基于官方文档和 AUP。SOC 2 Type II 状态为公司自称,未公开鉴证。“已实施”表示官方文档如此主张;没有任何一行代表经独立验证的控制。
[CE025, CE026, CE027, CE028, CE029, CE030]5.5 差异化、风险与产品缺口
Serval 所称差异化建立在四个支柱上。第一,原生基于代码生成工作流,区别于 ServiceNow Flow Designer 和 Freshservice Workflow Automator 等拖拽式前代产品;确定性代码工作流可以审计、版本化,并兼容 CI/CD,低代码逻辑块做不到这些。第二,双智能体安全架构明确把 Help Desk Agent 限制在预先授权的工具集合内,为企业对无约束 AI 的担忧给出结构性答案。第三,自然语言界面大幅降低搭建新自动化的边际成本,Serval 认为这正是 ITSM 自动化采用的瓶颈。第四,部署灵活性(云端或 Kubernetes 自托管)加上广泛集成覆盖(60+)降低切换成本,并支持「AI layer」叠加策略。 与这些优势相对,产品承担实质性技术风险。LLM 提供商依赖最深:Serval 的智能体依赖一个或多个第三方 LLM API(身份未披露)。任何提供商宕机、定价变化、模型弃用或安全政策调整,都可能实质性削弱产品。Automation Agent 的代码生成质量无法从公开来源验证——生成的工作流代码若出错,可能引入访问控制漏洞或数据暴露风险,这比聊天机器人答错问题后果更重。99.9% 正常运行时间 SLA 已被陈述,但无法独立验证。没有公开状态页或历史可用性数据。 Serval 与开发者社区的互动非常有限。没有找到 Serval 集成或 SDK 的公开 GitHub 仓库。来自 serval.com 的 Hacker News 投稿没有归档公开讨论,HN Algolia 搜索「Serval ITSM」也没有返回工程讨论。这种低开发者社区足迹符合纯企业销售动作,但也意味着产品架构或可靠性没有独立技术审视。 The Register 和 VentureBeat 在 2026 年初发布文章,指出企业 IT 部署智能体 AI 的一般风险,包括工作流生成中的幻觉、大规模集成失败,以及企业对 AI 可靠性保证的怀疑。这些风险适用于 Serval,不过目前没有具体事件或产品故障被公开归因于 Serval 本身。路线图信号很少:Serval 博客提到会扩展到 HR、法务和财务工作流(Reuters 报道中提及),但没有公开产品路线图、发布变更日志或开发者 API 版本历史。[CE032, CE033, CE034, CE035, CE036, CE037]
| 日期 / 信号 | 功能 / 里程碑 | 状态 | 含义 | 来源 |
|---|---|---|---|---|
| Oct 2025 | Series A 交割($47M,Redpoint);Help Desk Agent 和 Automation Agent GA | 已完成 | 核心产品 GA 已确认;提及具名客户 Perplexity、Together AI、Mercor | TechCrunch Series A 公告 |
| Dec 2025 | Series B 完成($75M,Sequoia);进入独角兽行列;声称收入增长 500% | 已完成 | 融资节奏和估值显示投资人信心很强;收入基数仍未披露 | Reuters、BusinessWire Series B 公告 |
| Dec 2025 (Reuters) | 宣布扩展至 HR、法律和财务工作流自动化 | 进行中 / 早期阶段 | Serval 走出核心 IT 用例;集成复杂度和竞争范围同步上升 | Reuters 2025 年 12 月报道 |
| May 2026 | Serval Start 项目启动;首期 12 人技术构建者队列 | 已完成(队列曾开放,现已关闭) | 显示公司重视招聘;也释放开发者品牌信号;未披露产品路线图 | 网址:serval.com/start |
| Jun 2026(运行日期) | 主页将 Asset Management 列为产品能力 | 无文档;成熟度未验证 | 可能是早期功能,也可能是营销占位;带来客户预期风险 | serval.com 主页 |
| 未披露 | 除基础 CLI 参考外,公开 API 版本、更新日志或开发者文档 | 不可用 | 没有公开发布历史,削弱开发者社区可见度和信任 | 文档:docs.serval.com |
| 未披露 | 漏洞赏金计划或公开渗透测试范围 | 未发现 | 缺口会影响企业采购的安全信号;在当前阶段仍属正常 | 研究(未见证据) |
路线图信号基于截至 2026-06-02 的新闻稿、产品页和新闻报道推断。公司尚未发布官方产品路线图。日期来自新闻稿时间戳和报道日期。「未披露」行代表公开信息中已确认的缺口,不等于公司失败。
[CE032, CE033, CE034, CE035]对 Serval 核心能力领域的成熟度和证据质量作估算,对比公司声称状态与现有外部验证。
成熟度评级和证据质量是分析师基于截至 2026-06-02 可得公开信息作出的判断。「Company-Claimed」对应官方产品描述;「External Evidence Quality」和「Diligence Confidence」基于可独立核验的公开证据,而非公司陈述。
[CE001, CE002, CE009, CE032, CE037]5.6 图表
06客户
6.1 客户基础画像
Serval 披露的客户名单数量不多,但经过战略筛选,在 AI 和开发者工具社区中能最大化释放可信度信号。公司首页页脚列出四家公司——Perplexity、Together.ai、Mercor 和 Cribl——Sequoia 的 Serval 组合页面则独立扩展至 Vercel 和 Verkada。TechCrunch 的 Series A 文章以及 Reuters/USNews 的 Series B 报道又为 Perplexity、Mercor 和 Together AI 提供独立第三方印证。七家客户都是有风投支持的技术公司,员工规模从约 50 人到数百人不等;没有上市公司、Fortune 500、受监管金融机构、政府机构或传统企业客户。客户名单覆盖多个子赛道:AI 推理平台(Perplexity、Together AI)、开发者平台(Vercel)、上市营销数据工具(Mercor、Clay)、可观测性与数据路由(Cribl),以及实体安全技术(Verkada)。Perplexity 和 Together AI 是当前 AI 基础设施浪潮中最受关注的名字之一,为 Serval 在风投支持生态中提供强社会证明。Vercel 和 Verkada 将可信度延伸到纯 LLM 空间之外,说明平台可能处理硬件相邻和基础设施规模公司的 IT 工作流。公司没有披露来自受监管垂直领域——医疗、金融服务、保险、政府——的客户,这限制了企业风险买家在合规密集环境中评估完整 ITSM 替换时可依赖的证据基础。 [CU001, CU002, CU003, CU004, CU010, CU011]
| 客户 | 细分领域 | 购买方 / 用户 / 付款方画像 | 主要用例 | 规模(估计) | 收入 / 战略价值 | 证据缺口 |
|---|---|---|---|---|---|---|
| Perplexity | AI 推理平台 | IT 负责人 / IT 团队 / 工程主导运营 | 服务台自动化、访问权限开通 | ~200–500 名员工(估计) | 高(旗舰 AI 品牌,Sequoia 被投) | 无案例研究;主页 404 |
| Together AI | AI 推理 / 训练云 | IT 负责人 / 运营团队 | 工单自动化、入职 / 离职流程 | ~100–300 名员工(估计) | 高(Reuters、TC、Serval 主页均点名) | 无案例研究;主页 404 |
| Mercor | 招聘自动化平台 | IT / 运营团队 | 服务台、访问管理 | ~50–150 名员工(估计) | 中(TC、Redpoint 被投列表确认) | 无公开成效数据 |
| Vercel | 开发者部署平台 | IT 团队 | 服务台、访问权限开通、集成 | ~300–600 名员工(估计) | 高(企业级开发者工具;Sequoia) | 无公开成效数据 |
| Verkada | 物理安全技术 | IT / 安全运营 | 访问管理、入职流程 | ~1000+ 名员工(估计) | 高(硬件 + 软件混合;规模更大) | 无公开成效数据 |
| Clay | GTM 数据与增强平台 | IT / 运营 | 工作流自动化、访问请求 | ~50–200 名员工(估计) | 中(Sequoia 被投列表) | 无公开成效数据 |
| Cribl | 可观测性 / 数据路由 | IT / SecOps 团队 | 服务台自动化、数据工作流 | ~300–600 名员工(估计) | 高(列在主页页脚;成熟 SecOps 品牌) | 无公开成效数据 |
员工数估计来自 LinkedIn / Crunchbase 公开信号,仅为近似值;所有客户均未披露官方员工数。收入 / 战略价值来自分析师基于投资人影响力和品牌认知度的推断。
[CU001, CU002, CU003, CU004, CU031, CU032]| 客户 | 细分领域 | 部署 / 用例 | 生产环境 vs. 试点 | 披露成效 | 证据质量 | 局限 |
|---|---|---|---|---|---|---|
| Perplexity | AI 平台 | 服务台自动化、访问权限开通 | 生产环境(由主页和 Reuters 暗示) | 纳入第 1 天自动化 50% 声称的汇总口径 | 中(TC、Reuters、Sequoia — 3 个来源;无专门案例研究) | 案例研究 URL 返回 404;未发现 Perplexity 直接表态 |
| Together AI | AI 云 | 工单自动化、入职 / 离职流程 | 生产环境(由 Reuters 暗示) | 纳入第 1 天自动化 50% 声称的汇总口径 | 中(TC、Reuters、主页 — 3 个来源;无专门案例研究) | 案例研究 URL 返回 404;未发现 Together AI 直接表态 |
| Mercor | 招聘自动化 | 服务台、访问管理 | 生产环境(由 TC 和 Redpoint 被投列表暗示) | 未单独披露;纳入汇总声称 | 中(TC、Redpoint — 2 个来源) | 无案例研究;无客户自行发布的确认 |
| Vercel | 开发者平台 | 服务台、访问权限开通 | 生产环境(由 Sequoia 被投列表暗示) | 未单独披露 | 中(Sequoia、Serval 主页集成列表 — 2 个页面) | Sequoia 并非独立来源;无客户自行发布的确认 |
| Verkada | 物理安全 | 访问管理、入职流程 | 生产环境(由 Sequoia 被投列表暗示) | 未单独披露 | 中(Sequoia 列表 — 1 个独立来源) | 佐证最弱;只有 Sequoia 列表;无新闻报道 |
| Clay | GTM 数据平台 | 工作流自动化、访问请求 | 生产环境(由 Sequoia 被投列表暗示) | 未单独披露 | 中(Sequoia 列表 — 1 个独立来源) | 佐证最弱;只有 Sequoia 列表;无新闻报道 |
| Cribl | 可观测性 / 数据路由 | 服务台自动化 | 生产环境(由主页页脚暗示) | 未单独披露 | 低–中(仅主页页脚;没有针对 Cribl 的第三方确认) | 无第三方来源单独确认 Cribl;仅页脚列名 |
证据质量从独立媒体佐证、投资人被投页面列名、官网展示和客户自行发布内容几方面评估。生产环境 vs. 试点状态基于上下文推断;合同条款或上线日期均未公开披露。
[CU001, CU002, CU003, CU004, CU009, CU031]Serval 的客户旅程沿公司定义的四个阶段推进——Meet、Build、Deploy、Optimize;部署后的增长闭环来自扩展到更多部门、更深地自动化工作流。
旅程阶段来自 Serval 定价页和入职文档。摩擦点根据平台架构和竞争语境推断,并非来自已披露客户访谈。
[CU022, CU023, CU035]6.2 采用轨迹与部署证据
最醒目的商业信号是 Serval 在 2025 年 8 月至 12 月之间收入增长 500%,该数字由 Reuters 直接报道并经公司确认。$1B 估值仅在 $232M 估值三个月后出现,意味着投资人看到的不只是增长率,还有正在加速的轨迹。Sequoia 合伙人 Anas Biad 的评论——Serval 带来的客户反馈,是 Sequoia 自 16 年前投资 ServiceNow 以来听过最强的——是记录中最有力的第三方背书,尽管发言方是有直接经济利益的投资人。50% 首日工单自动化率出现在三个独立界面:Serval 首页、Sequoia 组合页面和 Reuters 文章。官方、合作伙伴证明和独立新闻来源的三重引用,提高了对该指标存在性的信心,但并不验证其方法,也不说明它是合同保证还是平均观察值。 Serval 的上市路径有两条:完全替换既有 ITSM 系统,以及作为 AI layer 叠加在旧平台之上,服务受多年合同锁定的客户。这种灵活性在商业上重要,因为它降低了无法立即退出 ServiceNow 或 Freshservice 的潜在客户的初始采购阻力。四阶段客户 onboarding(Meet、Build、Deploy、Optimize)记录在公司定价页面中,意味着其激活模式是结构化、服务辅助的,而非自助试用。从 Series A 到 Series B 不到 90 天,本身也是间接采用信号:投资人根据期间看到的客户牵引数据加速下注。 [CU005, CU006, CU007, CU008, CU009, CU016]
| 指标 | 数值 | 日期 / 期间 | 来源 | 置信度 | 含义 | 缺失分母 |
|---|---|---|---|---|---|---|
| 收入增长率 | 2025 年 8 月以来增长 500% | Aug–Dec 2025 | Reuters / USNews | 中 | 独立新闻确认早期商业牵引力很快 | 未披露绝对 ARR 或 MRR 数字 |
| 第 1 天工单自动化率 | >50% 的 IT 工单 | 截至 Dec 2025 | Reuters;Sequoia;Serval 主页 | 中 | 核心商业证明点;三处引用,但未披露方法 | 未定义工单类型、数量或错误率 |
| 估值抬升 | $232M(Aug 2025)→ $1B(Dec 2025),~3 个月内增长 4.3x | Aug–Dec 2025 | Reuters / PitchBook | 高 | 投资人信心增长快于已披露收入证据 | 未披露 ARR 倍数;估值依据未知 |
| Series A 到 Series B 间隔 | <90 天(Oct → Dec 2025) | Oct–Dec 2025 | TechCrunch;Reuters | 高 | 异常快速的后续融资意味着客户验证信号很强 | 两轮公告均未披露客户数或 NRR |
所有指标均由公司或投资人披露;未发现任何数字经过独立验证或第三方审计。收入增长百分比基于 Serval 向 Reuters 自述的说法;公司未公布基期美元金额。
[CU005, CU006, CU008, CU016, CU030]从初始认知到生产部署再到部门扩展的采用流程,并标出每个阶段的证据质量评级。
漏斗阶段是结构性阶段,基于已记录流程,而非量化数据。公开资料没有转化率或阶段量数据;该流程代表已记录的客户入职顺序。
[CU021, CU006, CU030]6.3 客户证明质量与留存缺口
尽管采用信号很强,按企业尽调标准,Serval 的客户证明质量仍偏薄。公司为 Perplexity 和 Together AI 设置的案例研究页面均返回 HTTP 404,/customers 页面也无法访问,意味着没有公开的、由独立方撰写的案例研究内容。记录中唯一一条推荐语——IT 负责人 Vernon Man 描述 50% 以上自动化——没有标明其代表的公司,无法交叉验证。Gartner Peer Insights 在「AI Applications in IT Service Management」类别下没有列出 Serval 评论,也未找到 G2 评论;这与一家 2024 年成立、尚未优先经营第三方评论平台的公司相符。公开来源中没有披露净收入留存、总收入留存、流失率、客户数、年度合同价值或客户终身价值。 安全维度为潜在买家再加一层运营风险。Dark Reading 引用的 Acronis 研究人员证明,智能体 AI 平台——包括采用 Serval 同类架构的「vibe coded programs」——在非确定性 LLM 层与确定性工具层交界处会带入传统软件漏洞。认证绕过、访问控制不足和提示词注入,仍是部署智能体 IT 自动化的组织面临的现实问题。Serval 借助双智能体架构和基于角色的访问控制缓解部分风险,但这项研究也强调,客户安全姿态是当前证据基础尚未充分覆盖的关键尽调维度。 [CU012, CU013, CU014, CU015, CU017, CU026]
| 指标 | 数值 / 状态 | 客群 | 置信度 | 尽调要求 |
|---|---|---|---|---|
| 净留存率(NRR) | 未披露 | 所有客户 | Unknown | 向 Series C 数据室索取;对标 ITSM SaaS 110–130% 的常见水平 |
| 总留存率(GRR) | 未披露 | 所有客户 | Unknown | 与 NRR 一并索取;区分 logo 与收入流失的指标 |
| 客户数(总账户) | 未披露 | 所有客户 | Unknown | 在 Series C 披露总 logo 数;当前未给出下限或上限 |
| 已发布第三方评价(G2 / Gartner) | 未发现评价(两个平台均无) | 所有客户 | 高(截至 2026-06-02 已检查) | 鼓励设计合作伙伴发布匿名评价;年轻平台出现滞后符合预期 |
Serval 当前发展阶段下,所有留存和满意度指标要么未披露,要么尚不适用。G2 和 Gartner 上没有评价,与公司 2025 年前才成立的时间线相符,但相对企业买方尽调惯例仍是缺口。
[CU014, CU015, CU026]| 供应商 | 已披露具名客户 | 已发布案例研究 | G2 / Gartner 评价 | 已披露客户数 | 净新增 Fortune 500 客户(已披露) |
|---|---|---|---|---|---|
| Serval | 7 个具名客户(Perplexity、Together AI、Mercor、Vercel、Verkada、Clay、Cribl) | 无(案例研究页面返回 404) | 未发现评价(截至 Jun 2026) | 未披露 | 披露为零 |
| Moveworks | 公开点名 Broadcom、Palo Alto Networks、DocuSign 等 | moveworks.com 发布多篇案例研究 | G2 上有评价(企业级) | "800+"(据 Moveworks 网站) | 多家 Fortune 500 / 大型企业 |
| Atomicwork | 网站列出若干企业名称 | 可用案例研究有限 | G2 评价有限(早期阶段) | 未披露 | 披露很少 |
| Freshservice (Freshworks) | 披露数千家客户 | 案例研究库丰富 | >4000 条 G2 评价(截至 2026) | ">60,000"(据 Freshworks IR) | 多家 Fortune 500 |
Serval 对比反映截至 2026 年 6 月的研究发现。Moveworks 和 Freshservice 数据来自其公开网站和分析师报告;精确数字可能变化。本表用于呈现 Serval 相对证明成熟度的背景,不是完整竞争分析。
[CU012, CU014, CU015, CU026]从四个维度评估每个具名客户的证据质量:独立媒体确认、投资人组合佐证、专门案例研究、客户自撰公开声明。
[CU002, CU003, CU004, CU012, CU017]6.4 扩张动力与集中度风险
Serval 的 land-and-expand 逻辑有两条可见路径:从 IT 部门扩展到 HR、法务和财务;以及加深与 60+ 企业工具的集成。Reuters 明确报道了部门扩张,Serval 集成页面也确认了受支持工作流界面的广度。四阶段 onboarding 旨在随时间加深部署,Insights Agent 模块会明确给出配置建议,推动新增工作流——这是扩张收入的结构性机制。不过,没有客户公开披露跨部门使用,也没有 upsell 或扩张 ARR 指标。集中度风险显著:七家命名客户均为风投支持公司,并且与 Serval 自身一个或多个投资方共享投资人关系。Perplexity 是 Sequoia 组合公司,而 Sequoia 正是 Serval Series B 领投方。领投 Serval Series A 的 Redpoint,也与 Serval 投资人基础共同投资了多家命名客户。这种共同投资重叠在紧密的 AI 初创生态中很常见,但会形成推荐网络动态;当 Serval 需要从投资人组合之外获取客户时,这种动态未必能持续。没有披露任何 Fortune 500、中端市场或受监管行业客户,是商业画像中最重要的单一缺口;在补上之前,Serval 的增长故事依赖于在一个相对同质、与 VC 相连的早期采用者队列中持续保持速度。 [CU018, CU019, CU020, CU022, CU023, CU024]
| 扩张驱动 | 集中度风险 | 严重性 | 尽调路径 |
|---|---|---|---|
| 部门扩张(IT → HR、法律、财务) | 已披露客户均为单团队部署;多部门收入未知 | 中 | 索取多部门 ARR 拆分;要求提供 HR / 法律部署的客户背书 |
| 借工作流自动化深度先落地再扩张 | 未发布增购或扩张 ARR 指标;扩张仅从架构建模 | 中 | Series C 时按队列索取过去 12 个月净扩张收入 |
| VC 共同投资网络获客 | 7 个具名客户中有 6 个与 Serval 至少共享一位投资人;转介绍网络有饱和风险 | 高 | 跟踪投资人网络之外的 pipeline 占比;询问首个主动入站付费客户 |
| 叠加 AI 层的部署路径 | 已签 ServiceNow / Freshservice 合同的客户,额外 ITSM 预算可能有限 | 中 | 澄清 AI 层客户与完全替换客户的平均合同金额和合同期限 |
风险严重性反映尽调团队基于现有证据的评估。鉴于七个具名客户均被确认是一个或多个 Serval 投资人的被投公司,VC 网络集中度风险评为高。
[CU018, CU019, CU020, CU021, CU024]6.5 图表
07风险
7.1 智能体 AI 可靠性与安全风险
Serval 的核心产品——一个可自主发放访问权限、解决 IT 工单并编排企业工作流的 AI 智能体——带来的风险画像与传统 SaaS 工具有质的不同。当 AI 智能体能够以用户身份认证、创建访问授权并修改审计日志时,故障爆炸半径远高于基于表单的工作流工具。安全研究人员已经给出具体案例:一名研究人员发现,一个权限过宽的 ServiceNow 聊天机器人只受出厂默认凭证保护,只要提供任意用户的电子邮件地址即可完成认证,从而获得在任何公司 ServiceNow 实例中创建强大 AI 智能体的权限。另一个演示显示,通过支持工单投递的漏洞利用——类似 IT 帮助台请求——大约 47 秒内就能提升权限、访问客户记录、外传数据,并修改审计日志掩盖痕迹。这些事件直接适用于 Serval,因为支持工单正是其主要输入通道。 Acronis 研究人员发现,智能体 AI 漏洞主要不是由 AI 模型缺陷造成,而是由连接 LLM 与企业工具的确定性软件层中的经典软件错误造成——缺少输入清洗、硬编码凭证和访问控制不足。NIST AI 100-2 E2025 将这一威胁面编入对抗性机器学习攻击分类,覆盖提示词注入、数据投毒、模型规避和模型提取。Serval 的双智能体架构把 Help Desk Agent 限制为只能使用预先授权工具,这是有意义的设计层控制,但无法消除授权范围配置错误或集成被攻破的风险。公司 150+ 原生集成连接带特权的身份系统(Okta、Google Workspace、GitHub、AWS),意味着攻击面会随采用扩大。CISA 在 2026 年针对企业环境中谨慎采用智能体 AI 服务发布具体指南,说明监管共同体已将这一风险类别识别为实质性风险。[CR001, CR002, CR003, CR004, CR007, CR008]
| 风险类别 | 描述 | 可能性 | 影响 | 当前控制 |
|---|---|---|---|---|
| 通过支持工单触发的 Agentic AI 提示注入 | 恶意载荷嵌入 IT 支持工单,经 Help Desk Agent 的输入解析层触发未经授权的 AI agent 动作 | 中 | 严重 | 预授权工具约束限制影响范围;输入净化尚未经过独立审计 |
| 50% 自动化 SLA 违约 | 集成缺口、AI 回归,或训练分布之外的高复杂度工单类型,使 AI agent 未能达到承诺自动化率 | 中 | 高 | SLA 设计;违约补救路径未公开说明 |
| 访问开通过程中的 AI 幻觉 | LLM 非确定性导致授予错误访问级别,或把访问权限授给错误资源或人员 | 低-中 | 高 | AUP 要求人审;未披露公开事故日志或监控看板 |
| 集成供应链漏洞 | 第三方集成(Okta、GitHub、Slack)被攻破或配置错误,把攻击向量带入 Serval 的自动化执行层 | 中 | 高 | SOC 2 Type II 覆盖数据处理;第三方集成安全态势未见 Serval 公开披露的独立审计 |
| 影子 AI 部署风险(客户) | 企业客户在敏感 IT 工作流中启用 Serval 自动化,却缺少足够变更管理;这与 Token Security 记录的 OpenClaw 快速失控部署模式相似(22% 的 AI-forward 公司在数天内出现) | 中 | 中 | 入职和实施流程;预授权工具约束降低爆炸半径 |
可能性和影响评级为作者判断,依据 Dark Reading 和 NIST AI 100-2 的可比系统事故报告;目前没有公开的 Serval 特定事故数据。空值单元格表示没有证据。
[CR001, CR002, CR003, CR007, CR008, CR010]Serval 七类主要风险的内在可能性和影响分数,并列出控制成熟度和剩余风险评估。
可能性和影响分数(1–5)是作者基于行业可比公司和已记录事件作出的判断;控制成熟度只反映公开披露的控制措施。
[CR002, CR007, CR013, CR016, CR031, CR033]7.2 监管、合规与法律风险
Serval 位于企业软件与重大自动化决策的交汇处,而这一组合正受到越来越多监管。欧盟 AI Act 于 2024 年 7 月 12 日刊登在 Official Journal,对作出或影响重大决策的 AI 系统施加有约束力的透明度和人工监督要求。自动化 IT 访问发放和员工离职决策,在具体部署语境和欧盟监管解释下,可能落入 Annex III 就业类别的高风险应用。GDPR Article 22 也独立限制对个人产生法律或类似重大影响的完全自动化决策,要求有意义的人工介入。Serval 自身 AUP 通过禁止「在没有有意义人工监督的情况下,对个人作出具有法律或类似重大影响的完全自动化决策」,并要求在依赖 AI 生成建议作出重大决策前进行「合格人工审查和批准」,承认了这一约束;这是自我限制条款,可部分缓解但不能消除监管风险。 研究中出现一个实质性治理异常:尽管 Serval 披露约 $127M 风险融资($47M Series A、$75M Series B),但在 2025 年至运行日期间,EDGAR 中未找到 Serval 的 SEC Form D 文件。根据 Regulation D,进行私人证券发行的公司通常必须在首次销售后 15 天内提交 Form D。没有任何文件,要么是需要纠正的疏漏,要么说明这些发行使用了其他豁免结构;无论哪种情况,尽调中都应具体询问。CISA 2026 年关于谨慎采用智能体 AI 服务的指南,也为关键基础设施行业中可能成为 Serval 客户的企业买家增加监管压力层,从而给销售流程带来间接合规开销。[CR005, CR006, CR011, CR013, CR014, CR015]
| 监管领域 | 具体要求 | Serval 风险暴露 | 严重性 | 已部署缓解措施 |
|---|---|---|---|---|
| EU AI Act(透明度 / GPAI) | 第 52 条规定与人交互的 AI 系统承担透明度义务;自动化、就业相邻决策可能被归入 Annex III 高风险分类 | 在欧盟部署自动化 IT 访问和 HR 工作流自动化,可能需要透明度披露;公司未发布 EU AI Act 就绪声明 | 高 | AUP 含人工监督条款;无正式合规声明 |
| GDPR 第 22 条 | 禁止在缺少有意义人工干预时,作出具有法律效力或重大影响的纯自动化决策 | 自动化 IT 访问开通和员工离职流程具备就业相邻影响,可能触发欧盟客户的第 22 条义务 | 高 | AUP 要求对有后果影响的决策进行合格人工复核 |
| 美国证券法(Regulation D / Form D) | 豁免证券发行首次销售后 15 天内,必须向 SEC 提交 Form D | 截至运行日期,未发现 Serval 针对 Series A 和 Series B 已披露约 $127M 风险融资提交 SEC Form D | 中 | 未披露补救措施;缺口需要直接询问 |
| CCPA / 美国州隐私法 | 数据主体权利、自动化决策通知和退出要求 | Serval 在企业 IT 工作流中处理员工 PII;义务取决于客户所在地和各 MSA 的数据处理范围 | 中 | 声称具备 SOC 2 Type II;AUP 声明符合 GDPR/CCPA |
| CISA 智能体 AI 指引(2026) | CISA 指引要求关键基础设施运营方和企业 IT 环境谨慎采用智能体 AI 服务 | 关键基础设施行业的 Serval 客户面临 CISA 指引下的采购和安全审查要求,会给 Serval 销售带来间接合规负担 | 中低 | 双智能体预授权架构;SOC 2 Type II;声称符合 HIPAA |
覆盖范围并不完整;行业特定监管义务(HIPAA、FedRAMP、EU NIS2)需要按客户逐一分析。EU AI Act 分类是作者判断,仍待监管机构对 ITSM AI 自动化给出正式指引。SEC Form D 缺失是基于 EDGAR API 搜索的事实发现;缺失并不确认违规,但值得追问。
[CR005, CR006, CR009, CR013, CR014, CR015]7.3 竞争与生态依赖风险
对 Serval 最具战略意义的竞争威胁是 ServiceNow 本身。ServiceNow 平台覆盖 85% 的 Fortune 500,正在部署具备自主多步骤工作流能力的 NowAssist AI agents,并拥有 Serval 缺少的既有续约周期、企业信任和捆绑杠杆。如果 ServiceNow 复制 Serval 的 50% 自动化 SLA 保证,或以无增量成本把该用例吸收到自身平台,Serval 的主要差异化就会消失。Moveworks、Aisera 和 Atomicwork 是已有客户基础且架构相近的 AI 原生 ITSM 竞争者,压缩了 Serval 建立品类领导地位的窗口。 生态依赖风险很高。Serval 的 Help Desk Agent 完全通过 Slack、Microsoft Teams、电子邮件和网页门户运行——这些都是第三方沟通渠道。其核心自动化能力依赖 Okta、Google Workspace、GitHub、Jira、AWS 和 Workday 集成;这些提供商的任何 API 政策变化、宕机或信任撤销,都会直接削弱 Serval 产品。最关键的是,Serval 没有披露哪家或哪些 LLM 提供商支撑其 AI 层。单一提供商 LLM 依赖会引入模型可用性、定价和能力风险,而公开披露不足以评估这些风险。早期采用技术初创客户集中——虽适合初始牵引——但若一批规模相近的客户因共同宏观风险同步流失,收入会变得脆弱。[CR023, CR024, CR025, CR026, CR027, CR028]
| 依赖项 | 类型 | 集中度风险 | 可切断性 | 缓释措施 |
|---|---|---|---|---|
| Sequoia Capital(Series B 领投) | 资本提供方 | 高 — Sequoia 领投 Series B 的背书支撑 Series C 叙事;若 Series C 前失去 Sequoia 支持,下一轮融资可选项会显著受损 | 低 | Redpoint 和 General Catalyst 作为共同投资方提供部分兜底 |
| Okta / 身份提供商 | 基础设施集成 | 高 — 身份与访问管理自动化是 Serval 的核心功能;Okta API 弃用、价格调整或宕机会直接削弱产品能力 | 中 | 多 IdP 支持(Google Workspace、Azure AD、Okta)降低单一供应商依赖 |
| Slack / Microsoft Teams | 通信渠道 | 高 — Help Desk Agent 主要通过 Slack 和 Teams 运行;任一渠道中断都会关闭面向员工的主要界面 | 中 | Web 门户和邮件兜底渠道提供连续性路径 |
| 底层 LLM 提供商(未披露) | AI 模型提供商 | 严重 — Serval 的 AI 质量、延迟和成本完全取决于底层 LLM 提供商;提供商身份未公开披露 | Unknown | 未披露;LLM 提供商集中度是核心尽调问题 |
| ServiceNow(既有竞争对手和相邻威胁) | 竞争生态 | 中 — 许多 Serval 潜在客户已是 ServiceNow 客户;ServiceNow NowAssist 与既有 ITSM 续约捆绑,威胁 Serval 的相邻销售策略 | 不适用 | 覆盖式和集成层销售动作;Serval 一边定位为互补产品,一边也竞争完整平台替换 |
LLM 提供商身份未公开,是从没有任何公开披露中推断得出;这是实质性缺口。ServiceNow 行反映竞争风险,不是依赖项。集中度水平为作者基于产品架构审查作出的判断。
[CR023, CR024, CR025, CR026, CR027, CR028]Serval 的依赖图:公司位于中心,周围是关键基础设施、资本、集成和竞争依赖;边的方向表示依赖流向。
[CR023, CR024, CR025, CR026, CR027, CR028]7.4 执行、人员与财务风险
按任何标准看,Serval 的执行风险画像都极端。2025 年 12 月 Series B 时公司不到 30 名员工,却背负 $1B post-money 估值、合同化 50% 自动化 SLA,以及到 2026 年底扩至 100+ 人的计划——12 个月内员工数扩大 3 倍。按旧金山薪酬水平,一家 100 人 AI 公司每年全负担员工成本约 $18M–$25M;公司披露的 $127M 资本提供了现金跑道,但以这种速度同时做深产品、建立企业销售能力和安全审计深度,运营要求很高。 截至运行日期,Jake Stauch 是所有公开 Serval 传播中唯一被点名的 C-suite 高管。公司没有公开宣布 CTO、CFO、CPO 或首席安全官。Alex McLeod 在 Redpoint 组合页面被列为联合创始人,但在任何媒体报道中都没有公开露面。对一家已融资 $127M、估值 $1B 的公司而言,这种关键人物集中并不典型;如果 Stauch 在 Series C 前离任或丧失履职能力,将实质性影响投资人信心、客户关系和产品路线图。 在收入未披露的情况下,Serval 的估值倍数很激进。公司称自 2025 年 8 月以来收入增长 500%,但没有披露 ARR 或客户数。如果 Series B 时 ARR 为 $2M–$5M 区间(与这一员工规模下非常早期的企业 AI 公司相符),$1B 估值意味着 200x–500x 收入倍数——只有在增长持续保持或超过当前速度直至 Series C 时,这一水平才可持续。增长放缓、关键客户流失,或在与 ServiceNow 竞争中失利,都可能迫使公司下轮降价。[CR031, CR032, CR033, CR034, CR035, CR036]
| 风险 | 描述 | 严重性 | 触发信号 | 缓释措施 |
|---|---|---|---|---|
| CEO 关键人物依赖(Jake Stauch) | Stauch 是所有公开 Serval 沟通中唯一具名的 C-suite 高管;若他在 Series C 前离开,投资人信心、客户关系和产品叙事都会失稳 | 高 | CEO 离任公告;异常投资人董事会动作;长期媒体沉默 | 未披露继任计划;Serval Start 项目搭建人才管线 |
| 人员规模扩张执行(12 个月从 30 人到 100+ 人) | 12 个月人员扩张 3 倍,会迅速拉高管理容量、文化和产品质量风险;工程团队以这种速度入职,容易把缺陷带入安全关键的访问自动化工作流 | 高 | 招聘里程碑未达成;Glassdoor / Blind 负面信号;客户升级 | Series B 资金已投入;Sequoia 和 Redpoint 网络支持;Serval Start 管线 |
| 联合创始人与领导层覆盖缺口 | Alex McLeod(联合创始人)公开资料很少;公司尚未宣布 CTO、CFO 或 CISO;在这一融资阶段和规模下,单一可见创始人风险并不典型 | 中 | McLeod 离开;Series B 关闭后仍未出现具名领导层招聘 | 投资人董事会监督;Sequoia 和 Redpoint 运营支持团队 |
| 规模化后的工程技术债 | 随着 100+ 企业集成扩张,vibe-coded 自动化架构可能累积技术债;安全关键的访问开通需要持续工程严谨性,而高速扩员下很难守住 | 中 | 客户升级;集成脆弱性报告;SLA 未达标模式 | SOC 2 审计纪律;Series B 计划中优先招聘工程人才 |
| 竞争性人才获取 | ServiceNow、Moveworks、Glean 和高融资 AI 初创公司争抢同一批旧金山 AI/ML 工程与企业销售人才 | 中-低 | offer 接受率下降;入职后 6–12 个月窗口内流失率上升 | $1B 估值带来的股权上行;Serval Start 品牌项目;投资人网络招聘 |
严重性评级为作者判断。人数数据来自 Series B 关闭时的公开新闻报道(2025 年 12 月)。工程技术债风险从产品架构描述推断;没有缺陷模式的直接公开证据。
[CR031, CR032, CR033, CR034, CR035, CR036]展示 Serval 的主要风险事件如何传导为投资论点层面的后果;节点代表风险事件,边代表因果连接。
[CR002, CR005, CR012, CR016, CR031, CR033]7.5 缓解框架与放弃标准
Serval 最强的结构性缓解来自架构:双智能体设计将工具构建(Automation Agent)与工具执行(Help Desk Agent)分离,避免无约束 AI 智能体自主采取破坏性动作这一最危险故障模式。预授权工具范围把 Help Desk Agent 限定在明确批准的工作流内,SOC 2 Type II 合规说法也表明公司在数据处理和变更管理上有审计纪律。NIST AI Risk Management Framework 提供了独立风险治理模板,而企业买家正越来越多地要求 AI 供应商采用这类模板。 破坏投资逻辑的触发点可以识别。生产安全事件——AI 大规模错误发放访问权限、提示词注入攻击,或通过集成外传数据——很可能立刻引发合规敏感客户流失,并吸引监管审查。持续 SLA 违约率超过 50% 自动化保证,会形成公司尚未公开说明的合同责任。鉴于当前高管集中,CEO 在 Series C 前离任几乎必然破坏投资逻辑。降价轮或过桥融资会释放投资人信心恶化信号,并放大竞争和留存风险。投资人应把生产 SLA 数据、安全披露、员工数轨迹和 ServiceNow 竞争赢 / 输率作为主要前瞻指标,建立持续监控。[CR041, CR042]
| 风险领域 | 主要缓释措施 | 早期预警信号 | 打破投资论点的触发项 | 尽调问题 |
|---|---|---|---|---|
| Agentic AI 安全 | 双 agent 预授权架构;SOC 2 Type II;AUP 人工监督条款;CISA 指引合规路径 | 未披露安全事件;Serval 特定漏洞被分配 CVE;AI 错误开通的公开报告 | 生产环境 AI 错误开通造成数据泄露、监管行动或客户合同终止 | 索取渗透测试结果、AI 红队报告和事故响应手册 |
| 监管与法律合规 | GDPR/CCPA AUP 条款;HIPAA 合规声明;SOC 2 Type II;欧盟市场进入推迟至 AI Act 合规确认后 | 欧盟客户询问暴露 AI Act 缺口;隐私监管机构询问;Form D 通知 | 针对 Serval 或引用客户的监管执法行动(GDPR、CCPA、SEC),并点名 Serval 是记录系统 | 索取 DPA 模板、EU AI Act 自评,以及 SEC Form D 状态与理由 |
| 竞争替代 | 50% 自动化 SLA 保证作为护城河;相较 ServiceNow NowAssist 更快产生价值;面向 ServiceNow 锁定账户的覆盖式销售动作 | ServiceNow 宣布等效 SLA 保证;Moveworks 或 Glean 为 ITSM 扩张融资 $200M+;Serval 对 ServiceNow 的竞争胜率跌破 40% | ServiceNow 在企业续约时向既有 ITSM 客户免费捆绑 AI 自动化;Serval 管线连续两个季度下滑超过 30% | 验证企业交易中针对 ServiceNow NowAssist 的竞争赢 / 输分析 |
| 执行与人员 | Series B 资金($127M);Sequoia 和 Redpoint 董事会监督;Serval Start 人才管线;强投资人品牌帮助招聘 | CEO 离任公告;人数里程碑未达成;客户升级中可见产品质量恶化 | CEO 在 Series C 关闭前离开;到 2026 年 Q3 人数未达到 60;60 天内第二位具名高管离开 | 背调所有具名高管;验证人数计划和 2026 年招聘管线 |
| 财务与估值 | 500% 收入增长声明;强投资人财团;报价制保留毛利弹性;$127M 资金按当前 burn 可提供 24-36 个月 runway | 未能以 $1B 或更高估值完成 Series C;客户 cohort 流失率超过 10%;Series C 材料出现 ARR 增长减速信号 | 达到 $10M ARR 前需要 down round 或过桥融资;Series C 披露 ARR 低于 $3M;Sequoia 拒绝领投 Series C | 要求在 Series C 材料中披露 ARR、NRR、CAC 回本期和毛利率 |
触发项是作者基于 Series B AI 基础设施公司行业规范设定的阈值;Serval 尚未发布自身 SLA 违约或流失阈值。财务触发值为分析估计。
[CR039, CR040, CR041, CR042]7.6 图表
08估值
8.1 估值结构与融资背景
Serval 于 2025 年 12 月完成 Sequoia Capital 领投的 $75M Series B 后,达到 $1B post-money 估值。这次升值很醒目:PitchBook 引用 2025 年 8 月的隐含估值为 $232M,同月公司披露收入增长 500%。到 2025 年 12 月——大约四个月后——Sequoia 领投的轮次按 $1B 定价,不到一个财季估值上升 4.3 倍。已披露累计融资为 $127M(2025 年 10 月 Redpoint Ventures 领投 $47M Series A,加上 $75M Series B)。Sequoia 合伙人 Anas Biad 公开表示,公司带来的客户反馈是 Sequoia 自 16 年前投资 ServiceNow 以来见过最强的,这为独角兽估值提供了最可信的公开背书。 但融资背景暴露出结构性尽调缺口。在 EDGAR 全文搜索系统中,没有找到 Serval, Inc. 的 SEC Form D 豁免通知;对一家已接受约 $127M 豁免证券发行的美国公司而言,这并不寻常。它究竟反映申报延迟、不同申报实体,还是离岸控股结构,仍未解决。公司 Series B 完成时约 30 名员工——以 $1B 估值看,运营基数极其精简——并计划到 2026 年底超过 100 人,意味着员工数扩张 3 倍以上,会在收入同比例放大前显著加快烧钱速度。公司没有公开披露现金头寸、现金跑道或烧钱速度。 [CV001, CV002, CV003, CV004, CV007, CV008]
从 Series B 进入点($1B)到 2027–2028 年流动性事件,Serval 的三情景退出估值区间。牛市情形产生温和正回报;基准情形持平至为负;熊市情形深度为负。
所有情景 ARR 数字均为估计。退出倍数参考 2024–2025 年上市 SaaS 可比公司交易。回报为扣除管理费、carry 和优先权结构前的总回报——LP 净回报会更低。
[CV037, CV038, CV039]8.2 隐含倍数与可比公司基准
在 ARR 未披露的情况下,Serval 的隐含估值倍数只能通过区间情景推断。如果 2025 年 8 月收入基数约为 $2M ARR,500% 增长意味着 Series B 完成时约 $12M ARR,则 $1B 估值约为 83x 过去十二个月 ARR 倍数。如果基数为 $5M 并增长到 $30M,倍数压缩至约 33x。两种情形都处于 2025 年早期 SaaS 融资的前十分位,只有在市场预期其增长轨迹异常强劲时才说得通。2025 年前四分位早期 AI SaaS 公司获得 15–40x 预期 ARR 倍数;三位数倍数说明要么增长极强,要么收入基数被压缩且无法验证。没有独立分析师发布过 Serval 的具体 ARR 估计。 可比上市公司组锚定上限。ServiceNow 是主导性 ITSM 既有厂商,2024 年全年收入约 $10.98B,其市值隐含 13–15x 过去十二个月收入倍数——远低于 Serval 隐含区间,反映出成熟度和现金流生成能力。Atlassian 的 Jira Service Management 业务因增长放缓,交易倍数约为 9–10x 收入。私有 AI ITSM 同业 Moveworks 在 ARR 同样未披露的情况下,以 $2.1B 估值融资约 $300M,说明投资人群体能接受成长阶段 AI ITSM 的溢价倍数。PitchBook 在该轮后独立将 Serval 标记为 $1B 公司,为 post-money 标记提供第三方印证。ITSM 市场基于订阅的经常性收入结构支持更高长期倍数,但 Serval 的倍数需要一条目前无法从公开数据验证的增长轨迹。 [CV012, CV013, CV017, CV018, CV019, CV020]
| 公司 | 2024 收入 / ARR | 估值或市值 | 收入倍数 | 阶段 / 相关性 | Serval 可比限制 |
|---|---|---|---|---|---|
| ServiceNow | $10.98B(FY2024 订阅收入) | ~$150B 市值 | ~13–15x 历史收入 | 成熟上市既有厂商;Now Assist AI 所在的直接竞争 ITSM 平台 | ServiceNow 的成熟度和现金生成能力压低其倍数,相较成长阶段 AI-native 同行并不适合作为 Serval 入场价格可比 |
| Atlassian(Jira SM) | >$4.4B(FY2024 总收入) | ~$40–50B 市值 | ~9–11x 历史收入 | 中端市场和企业 ITSM;Jira Service Management 是直接产品竞争对手 | Atlassian 较慢的云 ITSM 增速进一步压低倍数;Serval 更高的增长值得溢价,但基础 ARR 未知 |
| Moveworks | 未披露(2022–2023 年以 $2.1B 估值融资 ~$300M) | $2.1B 上轮估值 | 未披露(估计 20–40x ARR) | 后期成长 AI ITSM 私有可比公司;2016 年成立;市场位置可直接比较 | 未公开 ARR,倍数无法验证;更早年份公司意味着增长预期低于 Serval |
| Aisera | 未披露(融资 >$100M,估值未披露) | 未披露 | 未披露 | 早期成长 AI ITSM 私有可比公司;2017 年成立;产品范围可比 | 估值和 ARR 均未披露;仅用于定性确认竞争格局 |
| Freshworks(Freshservice) | ~$700M(Freshservice 分部按 Freshworks 总 ARR $700M+ 估算) | ~$3.5B 市值(2024) | ~5x ARR | 中端市场 ITSM 上市公司;Freshservice 是相关产品线 | Freshworks 自身也面临倍数压缩;增长轨迹低于 Serval 声称的速度;相比 Serval 的企业动作更偏 SMB |
所有私有公司估值均为披露轮次的最后已知标记。私有公司收入倍数基于融资水平和市场定位下的合理 ARR 区间估算。本表只部分列举相关可比公司;并非所有 AI-native ITSM 厂商都有公开知名度或足够披露财务数据可纳入。
[CV012, CV013, CV024, CV025, CV036]在六种 ARR 情景下,用固定 33x ARR 倍数反推 $1B Series B 标记对应的隐含估值。敏感性显示,无法验证的收入基数会让隐含倍数大幅摆动。
所有 ARR 数字都是推断情景;Serval 未公开披露 ARR。y 轴代表 $1B Series B 投后估值下的隐含收入倍数。若要达到 25x 倍数,需要约 $40M ARR,符合牛市情形,但公开数据尚无法验证。
[CV017, CV018, CV019]8.3 乐观、基准与悲观情景分析
乐观情景要成立,Serval 需要在 2026 年底把 ARR 做到 $40–60M,净留存率(NRR)超过 150%,并从技术原生客户跨进至少两个企业垂直行业(金融服务或医疗健康),同时在规模化后守住 50% 工单自动化保证。在这些条件下,若 2027–2028 年退出时按 25–30x 远期 ARR 估值,对应 $1.25–1.8B 估值,相比 Series B 入场价产生 1.3–1.8x 回报。Sequoia 的高确信度、Perplexity 和 Verkada 的客户验证,以及 IT 自动化转向 AI 原生的长期趋势,让乐观情景具备可能性。但它要求公司在获客、产品和安全上持续极强执行,同时 ServiceNow 不能以零增量成本把 AI ITSM 打包进现有产品来反击。 基准情景假设 2026 年底 ARR 为 $20–35M,NRR 超过 120%,技术原生客户继续扩张,并在 2028 年按 20–25x ARR 退出。这意味着估值区间为 $500M–$875M——大致持平或略低于 Series B 入场价——后期 Series B 投资人的回报生成有限。悲观情景假设 ServiceNow 到 2026 年 Q3 激进打包 Now Assist,且不收增量费用,把 Serval 的商业窗口压缩到 500 人以下技术公司。在这种情景下,Serval 到 2026 年底只能做到 $10–15M ARR,市场倍数压缩到 12–15x,2028 年退出对应 $150–$225M 估值——相较 $1B Series B 入场价损失明显,也可能触发下轮降估或困境退出。Axios 2026 年 1 月报道确认,LP 对收入基础不可验证的 AI 初创估值审查已经收紧,因此悲观情景的尾部比标题看起来更厚。 [CV026, CV037, CV038, CV039, CV030, CV032]
| 场景 | 关键假设 | 隐含 2026 ARR | 适用退出倍数 | 隐含退出估值 | 概率信号 |
|---|---|---|---|---|---|
| 牛市 | 2026 年 12 月达到 $40–60M ARR;150%+ NRR;扩展企业垂直行业;ServiceNow 不免费捆绑 AI | $50M(中点) | 25–30x 远期 ARR | $1.25B–$1.8B(2027–2028 年退出) | 低-中(需要完美执行;尚无公开证据显示已扩展到科技之外垂直行业) |
| 基准 | 2026 年 12 月达到 $20–35M ARR;120% NRR;tech-native 扩张持续;市场倍数漂移到 20x | $27M(中点) | 20x 历史 ARR | $540M–$700M(估值持平或低于 Series B) | 中(与当前客户画像和增长叙事一致) |
| 熊市 | ServiceNow 到 2026 年 Q3 零成本捆绑 AI ITSM;增长放缓至 $10–15M ARR;倍数压缩至 12x | $12M(中点) | 12x 历史 ARR | $150M–$225M(Series B 入场价值损失 75–85%) | 低-中(Axios 2026 年 1 月审视和 ServiceNow AI 平台重叠,让熊市尾部并非小概率) |
所有 ARR 数字均为推断估计;Serval 未公开披露 ARR。退出倍数参考 2024–2025 年相似增长阶段 SaaS 可比公司。概率信号为定性判断;没有私有收入数据,无法支撑量化概率模型。
[CV037, CV038, CV039, CV026, CV032]建议链条从市场和证据信号出发,经过风险和估值过滤,落到 Track 决策。Sequoia 信念和市场规模是正向因素;收入不透明、监管异常和 ServiceNow 捆绑风险阻止给出完整买入建议。
[CV001, CV017, CV026, CV037]8.4 监管异常与反向信号
三个反向信号需要升级尽调。第一,截至分析日,在 EDGAR 全文检索 Serval, Inc. 的 Form D 文件没有结果。美国 Regulation D 要求发行人在 Regulation D 发行首次出售证券后 15 天内提交 Form D 通知。一家公司至少两轮已接受约 $127M 融资,却没有该文件,这是异常。可能原因包括:Serval 由非美国控股实体架构、投资经由免于美国 Form D 要求的境外载体进入;文件以不同法定名称提交;或合规遗漏。任何一种解释都需要法律尽调,也会提高尽调画像里的监管风险。 第二,Axios 2026 年 1 月发布分析,记录 LP 对 AI 初创估值的审查升温,尤其是收入未经验证、增长率由公司自报、隐含倍数超过 50x 的公司。Serval 的画像正落在 Axios 所称容易吸引下轮降估风险的标准上:未经审计的增长主张、压缩到独角兽的时间线、没有公开绝对收入数字。第三,Dark Reading 报道称,部署在企业 IT 场景的智能体 AI 系统——正是 Serval 的部署语境——会继承传统软件漏洞,包括提示注入和权限升级。Serval 的企业部署若发生安全事件,可能造成实质客户信任损害和整改成本;公开来源都没有量化这两项。 [CV010, CV011, CV026, CV033, CV042]
| 维度 | 牛市论点 | 熊市反论点 | 证据质量 |
|---|---|---|---|
| 市场机会 | ITSM 市场到 2029–2032 年增长至 $22–29B;AI-native 子赛道最快,CAGR 超过 30% | ServiceNow 既有地位覆盖 7,700+ 企业客户;零增量成本捆绑 AI 会压缩可触达窗口 | 中(分析师预测;ServiceNow 策略由公司公告确认) |
| 产品差异化 | 双 agent 架构和 50% 自动化保证在运营上具备差异化;tech-native 客户认可产品 | 企业规模下保证能否持续尚未验证;ServiceNow 和 Moveworks 的工程人数多得多 | 低–中(技术架构公开;企业可扩展性未证实) |
| 投资人质量 | Sequoia、Redpoint 和 General Catalyst 参与,提供一线信号和网络优势 | 压缩到独角兽的时间线制造后续融资复杂性;激进估值跳升提高 down round 风险 | 高(投资人身份由多家独立来源确认) |
| 收入轨迹 | 2025 年 8 月以来 500% 增长,暗示强产品市场契合;Sequoia 背书客户反馈 | 未披露 ARR 基数;从 $100K 增长 500% 不等于从 $5M 增长 500%;倍数无法验证 | 低(增长率为自报且未经审计;没有独立佐证) |
每个论点都按公开可得证据校准。私有 ARR 和 NRR 数据会显著改变收入轨迹一行的置信度。
[CV014, CV015, CV016, CV021, CV026, CV032]| 触发项 | 条件 / 阈值 | 对投资论点的传导 | 建议动作 |
|---|---|---|---|
| ServiceNow AI 捆绑 | ServiceNow 在 2026 年 Q3 前宣布 Now Assist 对既有 ITSM 客户零增量成本包含 | Serval 商业窗口缩小到尚未签 ServiceNow 合同的公司;熊市情景启动 | 立即重估;暂停追加资本承诺;要求管理层给出应对计划 |
| 安全事件 | Serval 企业部署中发生公开归因的 agentic AI 安全事件(提示注入、权限升级或数据外泄) | 企业 IT 采购信任流失;潜在客户流失和监管审视;转入熊市或回避情景 | 升级为回避;开展根因尽调;监控客户留存影响 |
| Down-round 或持平 Series C | Serval 在 Series B 后 18 个月内以 $1B 或以下估值融资 Series C | 确认增长减速;暗示 ARR 轨迹进入基准或熊市;此前独角兽标记失效 | 转为回避;不建议追加资本部署 |
| 关键客户流失 | Perplexity、Vercel 或 Verkada 在 2026 年底前公开离开 Serval 平台 | 削弱具名客户证明;引发对自动化保证交付和竞争胜率的质疑 | 调查原因;索取 NRR 数据;进一步承诺前跟踪新 logo 速度 |
触发项是基于公开可观察条件的定性评估。阈值定义为指示性,可能需要根据私有 data room 发现调整。所有动作含义都假设 Series B 或后期投资人视角。
[CV032, CV033, CV030]8.5 尽调要求与投资逻辑失效触发点
公开证据不足以支撑该估值承销。五项非公开披露可以补上尽调缺口:(1)当前 ARR,以及自 2025 年 8 月以来的队列级推进,用来锚定 500% 增长主张;(2)软件订阅与专业服务入职实施的毛利率拆分,因为高接触试点模式可能把毛利率压到 80% 以下;(3)按客户批次拆分的净留存率,验证先落地后扩张逻辑;(4)Series B 后烧钱速度与现金跑道,确认资本足以支撑计划中的员工数翻三倍;(5)法律实体架构,以及所有 Regulation D 豁免通知或等效境外证券文件副本。 投资逻辑主要在两个条件下失效。第一,ServiceNow 打包事件:12 个月内 Now Assist 对现有 ITSM 客户改为零增量成本,这会立刻把 Serval 可触达的商业窗口收窄到尚未签 ServiceNow 合同的公司。第二,客户环境发生安全事件,并被公开归因于 Serval 的智能体 AI 架构;在企业 IT 采购里,信任损害会远大于任何财务整改成本。次级失效条件包括:Series C 持平或下轮估值(确认增长放缓)、失去 Perplexity 或 Verkada 等关键具名客户,以及创始团队在 2026 年关键扩张期离开。TechCrunch Startups Weekly 2026 年 5 月把 Serval 与 Atomicwork、Moveworks 并列为三家值得关注的 AI 原生 ITSM 厂商,进一步说明竞争强度很高,执行纪律不可妥协。 [CV022, CV023, CV031, CV040, CV041]
| 维度 | 评估 | 置信度 | 依据 |
|---|---|---|---|
| 建议 | 跟踪(暂不买入) | 中 | 公开证据不足以支撑 $1B 估值;Sequoia 的信念是必要条件,但不充分。 |
| 风险评级 | 高 | 中 | 收入不透明、EDGAR 异常、ServiceNow 捆绑威胁,以及刚起步的人数扩张共同抬高风险。 |
| 估值立场 | 不披露 ARR 则偏贵;只有牛市情景下才站得住 | 低 | 33–83x 隐含 ARR 倍数需要异常强劲的增长轨迹,但公开数据无法验证。 |
| 目标回报(牛市情景) | 2–3 年 1.3–1.8x(退出估值 $1.25B–$1.8B) | 低 | 取决于 $40–60M ARR、150%+ NRR,以及成功扩展到 tech-native 客户之外的垂直行业。 |
| 决策含义 | 要求开放私有 data room;ARR 披露后重新审视建议 | 高 | 在追加资本前,把 2026 年 5–12 月增长指标对照牛市里程碑跟踪。 |
建议仅基于公开可得证据。若披露私有 ARR、毛利率和 NRR,牛市情景下建议可能转为买入,熊市情景下可能转为回避。置信度评级反映公开证据质量,不代表公司本身质量。
[CV001, CV003, CV017, CV018, CV037]| 主题 | 缺失证据 | 重要性 | 尽调路径 |
|---|---|---|---|
| ARR 与收入质量 | 当前 ARR、上一季度 ARR,以及 2025 年 8 月以来逐月收入进展 | 500% 增长声明没有公开美元锚点;估值倍数无法计算或对标 | 要求 data room — ARR 瀑布、收入确认政策、按客户签约日期的 cohort 进展 |
| 毛利率与单位经济 | 软件与专业服务毛利拆分;按 cohort 的 CAC、LTV、NRR 和回本期 | 高接触入职模式可能把毛利率压到 80% 以下;自动化 SLA 信用敞口未知 | 索取财务报表;带服务成本分摊的详细 P&L;NRR 调节表 |
| 现金状况与 burn | 2025 年 12 月 Series B 后的现金余额和月度 burn rate;基准和熊市场景下的 runway 预测 | 到 2026 年底人数扩至 100+ 会显著加速 burn;必须验证资本充足性 | 索取资金报告;带季度现金预测的经营模型;CFO 对 runway 签字确认 |
| 法人实体与监管申报 | 实体结构(美国 Delaware vs. 离岸控股公司);所有 Regulation D 豁免通知或等效文件副本 | EDGAR 未显示 Serval, Inc. 的 Form D;按美国证券法,这是尽调升级项 | 聘请法律顾问;索取公司章程;所有投资人认购协议和豁免通知 |
所有尽调问题都来自公开识别的证据缺口;每一项都需要私有 data room 权限才能解决。优先级列反映投资决策的阻断顺序:ARR 必须先解决。
[CV009, CV010, CV011, CV042]截至 2026 年 6 月,Serval 在六个投资维度上的 IC-ready 打分。市场和投资人质量得分高;证据质量和估值纪律得分低,原因是私有证据缺口。
[CV001, CV014, CV017, CV021, CV029]免责声明
基于截至 2026-06-02 的公开来源编制。这是一份分析性尽调材料,不构成投资建议;结论受制于私营公司披露限制。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Serval, Inc. is headquartered in San Francisco, CA 94104, as stated on the company website. | 中 | SO001 |
| CO002 | Serval was founded in 2024 by Jake Stauch and Alex McLeod. | 高 | SO003, SO004, SO007, SO008 |
| CO003 | Serval's stated mission is to deploy AI agents to automate help desk requests, just-in-time access provisioning, onboarding, and offboarding workflows. | 中 | SO001 |
| CO004 | Serval markets its product under the domain serval.com and operates as Serval, Inc. | 高 | SO001, SO012 |
| CO005 | Serval's platform uses a two-agent architecture: a Help Desk Agent that resolves requests by calling pre-approved tools, and an Automation Agent that builds those tools using natural language, generating deterministic code workflows. | 高 | SO002, SO001 |
| CO006 | Serval CEO Jake Stauch describes the automation-building process as "vibe coding for IT automation," allowing IT admins to describe workflows in natural language. | 中 | SO002 |
| CO007 | Serval's Help Desk Agent is explicitly limited to pre-authorized automations to prevent rogue AI actions; it cannot take actions not pre-approved by administrators. | 高 | SO002, SO001 |
| CO008 | Serval's business model is enterprise SaaS with quote-based pricing; no self-serve or SMB tier is offered as of the run date. | 中 | SO009 |
| CO009 | Jake Stauch is the CEO and co-founder of Serval, serving as the primary public face across all press, investor, and product communications. | 高 | SO002, SO003, SO007, SO008 |
| CO010 | Alex McLeod is the co-founder of Serval, confirmed by Redpoint's portfolio page; his specific operational role is not disclosed in public materials. | 中 | SO008 |
| CO011 | Christine Kim holds the title Head of Strategic Projects at Serval and is listed as a workshop leader in the Serval Start program materials. | 中 | SO013 |
| CO012 | No named board members, CFO, CTO, or VP-level executives appear in any public Serval materials as of June 2026. | 中 | SO001, SO005, SO006 |
| CO013 | Redpoint Ventures assigned three named partners to the Serval Series A deal: Alex Bard, Patrick Chase, and Jordan Segall. | 中 | SO008 |
| CO014 | General Catalyst's Serval investment is attributed to Marc Bhargava, Vedant Suri, and Kate Bender, who have been involved since 2024. | 中 | SO014 |
| CO015 | Anas Biad of Sequoia Capital led the Series B investment and publicly compared the customer feedback for Serval to what Sequoia heard before partnering with ServiceNow sixteen years ago. | 高 | SO003, SO004 |
| CO016 | Serval closed a $47 million Series A round in October 2025, led by Redpoint Ventures. | 高 | SO002, SO003, SO004 |
| CO017 | The Series A investors included First Round Capital, General Catalyst, BoxGroup, Bessemer Venture Partners, Meritech Capital, Strike Capital, Sunflower Capital, and Operator Partners. | 高 | SO002, SO004 |
| CO018 | Serval closed a $75 million Series B round in December 2025, led by Sequoia Capital in a preemptive round. | 高 | SO003, SO004, SO007 |
| CO019 | Serval's total capital raised reached $127 million after the Series B close. | 高 | SO003, SO004 |
| CO020 | The Series B was described as preemptive by Sequoia partner Anas Biad, meaning Sequoia approached Serval without a competitive fundraising process. | 中 | SO003 |
| CO021 | Serval's post-money Series B valuation was $1 billion, making it a unicorn as of December 2025. | 高 | SO003, SO004, SO007 |
| CO022 | Serval claimed 500% revenue growth since August 2025, per the December 2025 Reuters report, without disclosing absolute revenue figures. | 中 | SO003 |
| CO023 | Serval had fewer than 30 employees at the time of its Series B close in December 2025. | 中 | SO003 |
| CO024 | Serval plans to expand its headcount from under 30 to more than 100 employees in 2026, with hiring focused on go-to-market and engineering. | 中 | SO003 |
| CO025 | Serval's platform includes five product modules: help desk/ticketing, workflow automation, access management, asset management, and AI copilot for escalated tickets. | 高 | SO001, SO009 |
| CO026 | Serval's automations are generated as code and stored with a no-code UI overlay, allowing technical teams to inspect and manage automations in Git. | 高 | SO001, SO010 |
| CO027 | Serval supports over 60 native integrations as of the run date, including Slack, Okta, Google Workspace, GitHub, Jira, ServiceNow, AWS, Confluence, Kandji, and others. | 高 | SO011, SO010, SO001 |
| CO028 | Serval claims SOC 2 Type II certification, HIPAA compliance, and GDPR conformance, with TLS 1.3 in transit and AES-256 at rest, per company documentation. | 中 | SO010, SO001 |
| CO029 | Serval offers cloud-hosted, hybrid, and fully self-hosted deployment options; the self-hosted option uses Terraform and Kubernetes. | 高 | SO009, SO010, SO001 |
| CO030 | Serval's Acceptable Use Policy prohibits resale, competitive benchmarking, and use of the platform to build a competing product, consistent with standard enterprise SaaS terms. | 中 | SO012 |
| CO031 | Serval guarantees a 50% automation rate with its guided pilot program and provides a dedicated deployment engineer for customer onboarding. | 中 | SO009 |
| CO032 | Confirmed Serval customers include Perplexity, Together AI, Mercor, Vercel, Verkada, and Clay, as of the run date. | 高 | SO007, SO002, SO003 |
| CO033 | Serval claims its platform automates more than 50% of incoming IT tickets from day one for customers, supported by a homepage case study testimonial. | 中 | SO001, SO007 |
| CO034 | The Sequoia portfolio page confirms that Serval's AI agents are used by customers to automate help desk tickets, JIT access requests, onboarding, and offboarding flows. | 中 | SO007 |
| CO035 | Serval's revenue baseline as of August 2025 was very early-stage; the 500% growth figure cannot be assessed without the absolute denominator, which was not disclosed. | 中 | SO003 |
| CO036 | Serval's confirmed customer list is limited to technology companies; no regulated-industry or F500 enterprise customers are publicly named as of the run date. | 中 | SO007, SO002 |
| CO037 | Serval's go-to-market strategy supports both full platform replacement of legacy ITSM tools and an "AI layer" overlay approach for customers in long-term contracts. | 中 | SO003 |
| CO038 | No regulatory actions, lawsuits, enforcement actions, or sanctions against Serval, Jake Stauch, or Alex McLeod appear in publicly available records as of June 2026. | 中 | SO001, SO002, SO003 |
| CO039 | No product outages, security incidents, or data breach notifications involving Serval appear in any public source as of the run date. | 中 | SO001, SO010 |
| CO040 | No layoffs, leadership departures, or organizational restructuring events appear in public materials as of the run date. | 中 | SO003, SO006 |
| CO041 | Serval has not disclosed any debt facilities, credit lines, or secondary liquidity transactions in public materials. | 中 | SO003, SO004 |
| CO042 | Serval launched the Serval Start program in May 2026, a two-year program for aspiring founders to work at Serval before starting their own companies. | 高 | SO005, SO013 |
| CO043 | Serval's Series A valuation was approximately $232 million in August 2025, per PitchBook data cited in Reuters reporting. | 中 | SO003 |
| CO044 | Serval's valuation increased approximately 4x in three months, from ~$232 million at Series A to $1 billion at Series B. | 中 | SO003, SO004 |
| CO045 | ServiceNow and Moveworks are established AI-native ITSM players that represent incumbents Serval must displace or supplement in the enterprise market. | 中 | SO017, SO019 |
| CO046 | Gartner Peer Insights for AI Applications in IT Service Management lists verified enterprise user reviews; Serval's absence from verified reviewer ratings represents a trust barrier for enterprise buyers who rely on Gartner for vendor validation. | 中 | SO016 |
| CM001 | MarketsandMarkets projects the global ITSM software market to reach $22.1 billion by 2028 at a CAGR of 15.9%. | 中 | SM004 |
| CM002 | MarketsandMarkets projects the global AIOps platform market to reach $32.4 billion by 2028 at a CAGR of 22.7%. | 中 | SM004, SM025 |
| CM003 | MarketsandMarkets projected the Cloud System Management market to grow from $10.6 billion in 2020 to $31.4 billion by 2025, at a CAGR of 24.1%. | 中 | SM004, SM025 |
| CM004 | IBM defines ITSM as the practice of planning, implementing, managing, and optimizing end-to-end delivery of IT services to meet user and business goals. | 中 | SM015 |
| CM005 | ITIL (IT Infrastructure Library) is the most widely adopted ITSM framework; ITIL 4 was released in 2019 with a focus on digital transformation. | 中 | SM015 |
| CM006 | ITSM covers incident management, problem management, change management, service request management, and IT asset management as core process domains. | 高 | SM015, SM006 |
| CM007 | Serval describes itself as an AI-native IT service management platform that automates help desk requests, just-in-time access, and onboarding and offboarding. | 中 | SM001 |
| CM008 | Serval's homepage announces a $75M Series B at a $1B valuation led by Sequoia. | 高 | SM001, SM003 |
| CM009 | Serval's pricing page prominently markets a guaranteed 50% automation rate alongside a dedicated deployment engineering model. | 中 | SM016 |
| CM010 | Serval's integrations page lists 60+ enterprise tools including Okta, Google Workspace, GitHub, AWS, Rippling, Workday, Jira, ServiceNow, and Zendesk. | 中 | SM017 |
| CM011 | Serval's documentation states SOC 2 Type II certification, HIPAA compliance, and GDPR compliance as standard security and compliance posture. | 中 | SM018 |
| CM012 | Serval offers cloud-hosted, hybrid, and fully self-hosted deployment options, with the self-hosted option using Terraform and Kubernetes. | 中 | SM018 |
| CM013 | ServiceNow positions its ITSM product as targeting zero-touch service delivery using autonomous AI specialists to handle routine requests. | 高 | SM006, SM007 |
| CM014 | ServiceNow's AI Platform claims access to data from 450+ systems including SAP and Salesforce, enabling AI context across enterprise. | 中 | SM007 |
| CM015 | BMC offers HelixGPT agentic AI for ITSM, covering incident correlation, problem management, knowledge surfacing, and change risk analysis. | 中 | SM005 |
| CM016 | Aisera reports enterprise customers have auto-resolved 60-70% of IT tickets and one customer saved $2.2M in support costs using its AI agent platform. | 中 | SM009 |
| CM017 | Freshservice is trusted by 74,000+ businesses worldwide, with pricing ranging from $19/agent/month (Starter) to $99/agent/month (Pro), with custom enterprise pricing. | 中 | SM010, SM026 |
| CM018 | Moveworks provides an enterprise AI assistant platform that combines search and workflow automation across IT, HR, Finance, Sales, and other domains. | 高 | SM012, SM013 |
| CM019 | Atomicwork markets itself as an AI workforce platform for modern IT teams, targeting the same AI-native ITSM segment as Serval. | 中 | SM014 |
| CM020 | Atlassian Jira Service Management includes asset management and AI features in premium plans, targeting enterprise ITSM alongside its project management user base. | 高 | SM011, SM019, SM024 |
| CM021 | Freshservice case studies cite a 23% ticket deflection rate, 81% reduction in resolution times, and 60% IT cost reduction at individual customer deployments. | 中 | SM010 |
| CM022 | Aisera is listed in the Gartner Magic Quadrant for Artificial Intelligence Applications in IT Service Management, indicating the AI-ITSM market is mature enough for Gartner coverage. | 中 | SM009 |
| CM023 | TechCrunch reported Serval's Series A customers include major AI companies Perplexity, Mercor, and Together AI, representing fast-growing tech companies with lean IT teams. | 中 | SM002 |
| CM024 | Serval CEO Jake Stauch articulated the company's mission as making automation effortless enough that building an automation forever is easier than doing the task manually once. | 中 | SM002 |
| CM025 | Enterprise clients are keenly aware of the risks of rogue AI systems, driving the requirement for deterministic, permissioned automation frameworks in ITSM. | 中 | SM002 |
| CM026 | Serval's two-agent architecture separates tool-building (Automation Agent) from tool-execution (Help Desk Agent) to limit the action scope of any single agent. | 高 | SM001, SM002 |
| CM027 | VCTavern reports Serval's platform has expanded from IT teams to HR, legal, and finance for internal request automation beyond traditional ITSM. | 中 | SM003 |
| CM028 | VCTavern reports Serval's total funding reached $127 million across the Series A ($47M) and Series B ($75M) rounds. | 中 | SM003 |
| CM029 | ITSM platform procurement is driven by IT leadership—IT Director, CIO, or VP IT Operations—using centralized IT operations budget, typically part of G&A. | 中 | SM015, SM006 |
| CM030 | SOC 2 Type II, HIPAA, and GDPR compliance are standard enterprise procurement requirements for ITSM platforms handling employee data and access logs. | 高 | SM018, SM010 |
| CM031 | ITSM platforms handle sensitive data including employee PII, access request histories, and security audit logs, raising data privacy and residency requirements. | 高 | SM015, SM018 |
| CM032 | Moveworks' platform supports over 100 languages, enabling global enterprise deployments across multi-regional organizations. | 中 | SM013 |
| CM033 | Aisera reports one customer (Lifescan) auto-resolved 65% of incoming support requests, and another (BDO Canada) increased productivity by 72% through IT support automation. | 中 | SM009 |
| CM034 | Serval's Automation Agent enables users to describe workflows in natural language and generates them as executable code with a no-code UI representation. | 高 | SM001, SM018 |
| CM035 | Serval's Insights Agent analyzes ticket patterns to surface automation opportunities, suggest new workflows, and update knowledge base content automatically. | 高 | SM001, SM018 |
| CM036 | Enterprise ITSM deployments require integration with identity providers (Okta, JumpCloud, Microsoft Entra) for SCIM and SAML-based user provisioning and access control. | 高 | SM017, SM018 |
| CM037 | Serval's platform supports team segregation for IT, Security, HR, and Workplace teams with distinct ticket queues, integrations, and AI configurations. | 中 | SM001 |
| CM038 | Redpoint Ventures led Serval's $47M Series A, with participation from First Round Capital, General Catalyst, BoxGroup, Bessemer, Chemistry VC, and others. | 高 | SM003, SM020 |
| CM039 | Sequoia Capital led Serval's $75M Series B round, marking one of the most prominent VC endorsements for an AI-native ITSM startup. | 高 | SM003, SM021 |
| CM040 | The ITSM market includes on-premise and cloud solutions and is fragmented across service desk, change and configuration management, IT operations management, and ITAM modules. | 中 | SM004, SM015 |
| CM041 | Key incumbent ITSM vendors include ServiceNow, BMC Software, Broadcom, Ivanti, ManageEngine, SolarWinds, Atlassian, OpenText, and Zendesk as identified by MarketsandMarkets. | 中 | SM004, SM005, SM006 |
| CM042 | The ITSM market is undergoing rapid AI augmentation as all major vendors (ServiceNow, BMC, Freshservice) have launched generative AI and agentic features since 2023. | 高 | SM005, SM006, SM008, SM010 |
| CM043 | SaaS sprawl at fast-growing companies creates exponentially more IT access provisioning and offboarding events, making automation a cost necessity rather than discretionary spend. | 中 | SM001, SM017, SM002 |
| CM044 | ServiceNow's AI Platform positions autonomous agents as the core product architecture, moving from AI-assisted service delivery to fully autonomous IT operations. | 高 | SM006, SM007 |
| CM045 | Serval's homepage lists Perplexity, Together.ai, Mercor, and Cribl as named customers using the platform for IT automation. | 高 | SM001, SM022 |
| CM046 | Serval's onboarding process involves four phases (Meet, Build, Deploy, Optimize) with a dedicated deployment engineer, implying a multi-week enterprise sales and onboarding cycle. | 中 | SM016 |
| CM047 | Legacy ITSM incumbents like ServiceNow, Freshservice, and Jira Service Management have established partner ecosystems and multi-year contracts that create switching cost and inertia for new entrants. | 中 | SM006, SM010, SM019 |
| CM048 | Very small organizations (under 50 employees) are likely underserved by Serval's current go-to-market model given the dedicated deployment engineering requirement. | 中 | SM016 |
| CP001 | The AI-native ITSM market in 2026 organises around four player classes: legacy platform incumbents, AI-native point solutions, workflow automation hybrids, and traditional SaaS helpdesk vendors adding AI capabilities. | 中 | SP026, SP005 |
| CP002 | ServiceNow holds approximately 85% of Fortune 500 companies as ITSM customers and operates a 7,500-strong certified implementation partner ecosystem as of 2026. | 中 | SP005, SP007 |
| CP003 | ServiceNow serves 7,700-plus enterprise customers globally as of Q1 2026, maintaining its position as the dominant ITSM platform for large enterprises. | 高 | SP005, SP007 |
| CP004 | Freshservice serves 74,000-plus businesses across SMB to enterprise segments on a cloud-native SaaS ITSM model with published per-seat pricing. | 中 | SP009, SP010 |
| CP005 | Atlassian Jira Service Management holds PinkVERIFY certification and targets DevOps-adjacent teams with ITIL v4 alignment, leveraging the broader Atlassian ecosystem. | 中 | SP012, SP013 |
| CP006 | Moveworks positions as an enterprise AI copilot for employee services, supporting 100-plus languages across HR, IT, and facilities workflows with its Reasoning Engine. | 中 | SP015, SP016 |
| CP007 | Atomicwork markets its agentic service management platform as displacing ServiceNow for high-growth tech companies that find ServiceNow overly complex. | 中 | SP019, SP020 |
| CP008 | Aisera was included in the Gartner Magic Quadrant for AI Applications in ITSM and claims 65 to 90 percent auto-resolution rates across IT domains. | 中 | SP022, SP024, SP026 |
| CP009 | BMC Helix ITSM was named a Forrester Wave Leader for Enterprise Service Management Q4 2025, confirming incumbents continue investing in AI capabilities. | 中 | SP027 |
| CP010 | Serval achieved 500% ARR growth between August and December 2025 and reached a $1B unicorn valuation in its $75M Series B led by Sequoia Capital in December 2025. | 高 | SP028, SP029 |
| CP011 | Serval's AI agent autonomously resolves tickets end-to-end by integrating with Okta, GitHub, Slack, Jira, and 150-plus tools without requiring human escalation. | 中 | SP001, SP003 |
| CP012 | ServiceNow NowAssist AI Agents run on the Now Platform with native ML model orchestration, CMDB coupling, and deep change workflow management. | 中 | SP006, SP005 |
| CP013 | Freshservice Freddy AI provides AI-assisted ticket deflection and agent suggestions; fully autonomous resolution is not available in its standard tiers. | 中 | SP009, SP010 |
| CP014 | Jira Service Management includes Atlassian Intelligence for request categorisation and virtual agent support, but its AI depth is substantially below Serval's autonomous resolution tier. | 中 | SP012, SP013 |
| CP015 | Moveworks' Reasoning Engine supports multi-step agentic workflows and claims 40% fewer tickets requiring human intervention relative to baseline. | 中 | SP015, SP016 |
| CP016 | Atomicwork integrates an AI copilot called Atom that handles natural-language IT requests and auto-routes without manual classification. | 中 | SP019, SP020 |
| CP017 | Aisera's AI Service Desk uses conversational AI and RPA integrations to achieve claimed 65 to 90 percent resolution auto-rates across IT domains. | 中 | SP022, SP023 |
| CP018 | Serval guarantees a 50% ticket automation rate in its contractual SLA — a commitment no evaluated incumbent or AI-native competitor publicly matches. | 中 | SP001, SP002 |
| CP019 | ServiceNow publishes no list pricing for its ITSM products; all enterprise contracts are custom-negotiated with multi-year commitments typically exceeding $200,000 annually. | 中 | SP005, SP008 |
| CP020 | Serval supports cloud, hybrid, and self-hosted deployment options, enabling competition in regulated industries where data residency mandates preclude pure SaaS solutions. | 中 | SP001, SP003 |
| CP021 | Freshservice publishes four pricing tiers from $19 to $99 per agent per month (Starter to Enterprise), enabling bottom-up PLG adoption. | 中 | SP011 |
| CP022 | Atomicwork's Professional tier is priced at $90 per employee per year, published on its pricing page and targeting cost-conscious tech teams. | 中 | SP021 |
| CP023 | Moveworks pricing is custom-quoted; analyst estimates suggest enterprise contracts start above $250,000 annually, consistent with its $200M Series C enterprise positioning. | 低 | SP017, SP026 |
| CP024 | Atlassian JSM offers a free tier for up to three agents and scales to $44.27 per agent per month on its Premium tier. | 中 | SP014 |
| CP025 | Aisera's pricing is custom-quoted with a minimum 12-month commitment; no per-seat or tier list pricing is published on its website. | 中 | SP022 |
| CP026 | Zendesk Suite ticketing starts at approximately $55 per agent per month for mid-market plans, anchoring the lower end of the competitive pricing spectrum. | 中 | SP025 |
| CP027 | Serval's GTM focuses on high-velocity tech companies including Perplexity, Together.ai, Mercor, and Cribl — AI-first organisations that outgrow manual helpdesks rapidly. | 中 | SP004, SP029 |
| CP028 | ServiceNow's distribution relies on a 7,500-partner ecosystem and certified implementation partners across Deloitte, Accenture, and KPMG, creating a high barrier for new enterprise sales. | 中 | SP005, SP007 |
| CP029 | Freshworks uses a PLG motion with free trials and published pricing, enabling bottom-up adoption that bypasses IT procurement — contrasting with ServiceNow's top-down enterprise sales model. | 中 | SP009, SP011 |
| CP030 | TechCrunch's October 2025 Series A coverage explicitly framed Serval as targeting ServiceNow replacement, accelerating competitive attention from ServiceNow's enterprise sales organisation. | 中 | SP029, SP028 |
| CP031 | ServiceNow's switching costs include CMDB migration, custom workflow re-engineering, and retraining of ITIL-certified administrators — a multi-year, multi-million-dollar process for large enterprises. | 中 | SP005, SP006, SP026 |
| CP032 | Atlassian's Jira ecosystem (Jira Software, Confluence, Jira Service Management) creates cross-product lock-in that raises switching costs substantially for DevOps teams. | 中 | SP012, SP013 |
| CP033 | Serval's 150-plus native integrations create data and workflow coupling that raises switching costs as customers deepen automation coverage across their tool stacks. | 中 | SP003, SP001 |
| CP034 | Moveworks' multi-tenant AI models trained on customer communication history create proprietary learning lock-in that strengthens after 12 to 18 months of deployment. | 中 | SP016, SP015 |
| CP035 | The global ITSM market projected above $19 billion for 2026 is large enough to sustain multiple AI-native and incumbent players without near-term winner-take-all dynamics. | 中 | SP026, SP028 |
| CP036 | Gartner classifies the AI Applications in ITSM market as a distinct sub-segment with high vendor churn and rapid new-entrant rates as of 2025, signalling persistent commoditization pressure. | 中 | SP026 |
| CP037 | Serval's moat depends on a compounding automation dataset: each autonomously resolved ticket trains its models, widening the performance gap over time in a way new entrants cannot easily replicate. | 中 | SP001, SP004 |
| CP038 | Atomicwork received customer testimonials in 2025 to 2026 describing it as a ServiceNow replacement for high-growth tech teams, creating a direct channel conflict with Serval's displacement narrative. | 中 | SP019, SP021 |
| CP039 | BMC Helix's Forrester Wave Leader position for Enterprise Service Management Q4 2025 demonstrates incumbents continue to invest in AI and retain enterprise trust, posing a durable threat to AI-native challengers. | 中 | SP027, SP026 |
| CP040 | Serval holds SOC 2 Type II, HIPAA, and GDPR certifications, enabling regulated-sector deals in healthcare and financial services where some AI-native peers lack equivalent accreditation. | 中 | SP001, SP003 |
| CP041 | Rapid capital competition across the AI-ITSM segment — Moveworks $200M Series C, Serval $127M total — intensifies pressure on the same addressable market and may accelerate feature commoditization. | 中 | SP015, SP029 |
| CP042 | Serval's 30-person team relative to ServiceNow's 23,000-plus employees poses a delivery concentration risk for large enterprise contracts requiring deep implementation and partner support. | 中 | SP002, SP028 |
| CI001 | Serval positions itself as an AI-native IT service management platform, not a traditional ticketing tool, targeting companies that want AI agents to handle IT requests end-to-end. | 高 | SI001, SI011 |
| CI002 | Serval's pricing page describes a "one platform fee, no surprises" model, indicating a subscription rather than per-transaction or consumption-based billing structure. | 高 | SI002, SI001 |
| CI003 | Serval does not publish a price card; all purchasing requires a sales demo, making it a sales-led, quote-only motion with no self-serve or PLG entry point. | 高 | SI002, SI001 |
| CI004 | Serval's onboarding follows four named phases — Meet, Build, Deploy, Optimize — and includes a dedicated deployment engineer for each guided pilot, indicating a high-touch professional services element. | 中 | SI002 |
| CI005 | Serval guarantees a 50% automation rate for customer IT tickets, making this the primary commercial anchor of its sales motion; no penalty or credit mechanism for failing to meet this guarantee is publicly described. | 中 | SI002 |
| CI006 | Serval integrates with 60-plus enterprise tools including Okta, Google Workspace, GitHub, Jira, Slack, Freshservice, and ServiceNow, supporting a land-and-expand deployment model across the IT stack. | 高 | SI003, SI007 |
| CI007 | Serval's product is built on a two-agent architecture: one agent classifies and understands IT requests, a second agent executes resolutions through integrations with enterprise systems. | 高 | SI008, SI009 |
| CI008 | Serval's legal entity is Serval, Inc., incorporated and headquartered in San Francisco, CA 94104, per the Acceptable Use Policy page. | 中 | SI006 |
| CI009 | Serval deploys only via guided pilot; there is no self-serve trial, freemium tier, or publicly accessible product sandbox, which limits top-of-funnel volume but concentrates sales effort. | 高 | SI002, SI004 |
| CI010 | Serval raised $47M in a Series A round led by Redpoint Ventures, which closed on or around October 21, 2025. | 高 | SI008, SI009 |
| CI011 | Serval's Series A investor group included First Round Capital, General Catalyst, BoxGroup, Bessemer Venture Partners, Chemistry VC, Strike Capital, Sunflower Capital, and Operator Partners, alongside unnamed angels. | 高 | SI010, SI008 |
| CI012 | Serval raised $75M in a Series B round led by Sequoia Capital, which closed in December 2025. | 高 | SI009, SI010 |
| CI013 | The Series B round valued Serval at $1B post-money, making Serval one of the fastest-emerging unicorns in enterprise software history given the company's 2024 founding date. | 高 | SI009, SI010 |
| CI014 | Serval's total disclosed venture funding as of December 2025 is $127M, confirmed by Reuters, US News, and VCTavern citing multiple independent sources. | 高 | SI009, SI010 |
| CI015 | A PitchBook figure cited in the Reuters article placed Serval's implied valuation at approximately $232M in August 2025, which represented the pre-Series B reference point for the company. | 中 | SI009 |
| CI016 | Serval's valuation stepped up 4.3x from approximately $232M to $1B in roughly three months between August and December 2025, one of the most compressed unicorn creation timelines in the enterprise software sector. | 中 | SI009 |
| CI017 | Sequoia Capital partner Anas Biad is quoted in the Reuters coverage of the Series B, providing named endorsement from a senior Sequoia partner. | 中 | SI009 |
| CI018 | Redpoint Ventures partners Alex Bard, Patrick Chase, and Jordan Segall are listed on the Redpoint portfolio page as the Series A investment team for Serval. | 高 | SI012, SI008 |
| CI019 | Serval reported 500% revenue growth since August 2025, per the Reuters/US News article; no dollar figure was disclosed alongside this relative growth claim. | 中 | SI009 |
| CI020 | Serval publicly names Perplexity, Together AI, Mercor, Vercel, Verkada, Clay, and Cribl as customers on its homepage; these are all technology-native companies, most of which are themselves venture-backed. | 高 | SI001, SI011 |
| CI021 | Serval had approximately 30 employees at the time of its Series B close in December 2025, per the Reuters/US News article. | 中 | SI009 |
| CI022 | Serval announced plans to grow headcount to more than 100 employees by end of 2026, implying a 3-plus-times headcount expansion using Series B capital. | 中 | SI009 |
| CI023 | Serval has disclosed no ARR, MRR, or absolute revenue figure; the 500% growth claim cannot be anchored to a starting base, making it impossible to construct a bottom-up financial model from public data. | 高 | SI009, SI010 |
| CI024 | Serval is SOC 2 Type II certified and offers a 99.9% uptime SLA per its documentation, providing enterprise-credibility signals but not substituting for financial metric disclosure. | 中 | SI007 |
| CI025 | All seven named Serval customers are technology-native companies; no Fortune 500, financial services, healthcare, or government customers have been publicly named, leaving Serval's enterprise crossover appeal unproven. | 中 | SI001, SI011 |
| CI026 | Serval's implied revenue multiple at the $1B valuation cannot be benchmarked because both the numerator (ARR) and the growth trajectory's absolute scale are undisclosed. | 高 | SI009, SI010 |
| CI027 | ServiceNow reported approximately $10.98B in full-year 2024 subscription and professional services revenue, and is the dominant incumbent in enterprise ITSM with more than 7,700 enterprise customers. | 高 | SI013, SI027 |
| CI028 | The global ITSM market is estimated at approximately $11B in 2024 and projected to exceed $30B by 2030, per MarketsandMarkets and Grand View Research analyst reports. | 中 | SI021, SI022, SI023, SI028 |
| CI029 | Moveworks and Aisera are direct AI ITSM competitors with earlier venture histories, but both companies have also not disclosed ARR or unit economics publicly, limiting direct financial comparisons. | 中 | SI018, SI020 |
| CI030 | Freshservice and Atlassian Jira Service Management both publish transparent per-agent pricing starting at $15-$18 per agent per month, providing price anchors that Serval's quote-only model avoids. | 高 | SI016, SI017 |
| CI031 | ServiceNow has launched Now Assist, a generative AI and agentic AI capability native to its Now Platform, directly competing with Serval's core proposition and available as an upsell to its 7,700-plus enterprise customer base. | 高 | SI014, SI013 |
| CI032 | Atomicwork offers a freemium entry point into AI-native ITSM, representing a different commercial strategy than Serval's demo-only, enterprise-focused motion. | 中 | SI019 |
| CI033 | BMC Helix is a legacy ITSM platform undergoing AI modernization, representing the slow-moving but resource-rich incumbent segment that Serval's AI-native architecture aims to displace. | 中 | SI026 |
| CI034 | An EDGAR search for Form D filings by Serval, Inc. returned no results as of the analysis date, which is unusual for a US company that has accepted approximately $127M in exempt securities offerings. | 中 | SI024 |
| CI035 | Acronis security researchers, published by Dark Reading, have documented that agentic AI systems deployed for enterprise IT inherit traditional software vulnerabilities including prompt injection, privilege escalation, and authentication bypass, posing material risk for products like Serval. | 中 | SI025 |
| CI036 | Serval's 500% revenue growth claim is self-reported and unaudited; no independent analyst, audit firm, or investor letter has publicly corroborated the absolute revenue figures that underlie this growth rate. | 高 | SI009, SI010 |
| CI037 | Serval's burn rate, monthly cash consumption, and runway from the Series B are not publicly disclosed; capital adequacy can only be inferred from the timing and scale of the most recent funding round. | 中 | SI009 |
| CI038 | Serval's $232M-to-$1B valuation step-up in approximately three months represents an extremely compressed unicorn timeline that can create downstream complications for future financing rounds and employee equity alignment. | 中 | SI009, SI010 |
| CI039 | The Dark Reading article cites a concrete exploit chain discovered in ServiceNow's own AI agent implementation, illustrating that even major enterprise software vendors deploying agentic AI have faced material security vulnerabilities, directly relevant to the trust risk facing Serval. | 中 | SI025 |
| CE001 | Serval's platform combines three functional surfaces: a Help Desk module, an Access Management module, and an Automation Engine, marketed as an AI-native unified ITSM platform. | 高 | SE001, SE002 |
| CE002 | The Help Desk Agent accepts employee requests via Slack, Microsoft Teams, email, or web portal, routing tickets by content and context, searching connected knowledge bases, executing automations, and escalating with full context when needed. | 高 | SE001, SE002 |
| CE003 | The Help Desk Agent can only invoke pre-approved, deterministic tool automations built by the Automation Agent; it cannot take any ad-hoc action outside that pre-authorized permission set. | 高 | SE001, SE011, SE012 |
| CE004 | The Automation Agent accepts natural-language descriptions from IT administrators and generates deterministic, code-based workflow automations without requiring drag-and-drop logic building. | 高 | SE001, SE008, SE012 |
| CE005 | The Insights Agent continuously analyzes ticket patterns to surface automation opportunities, recommend knowledge-base updates, and suggest configuration improvements. | 中 | SE001, SE002 |
| CE006 | Serval lists five named product capabilities on its homepage: Resolve Requests, Build Workflows, Transform Ticketing, Manage Access, and Manage Assets. | 高 | SE001, SE008 |
| CE007 | Serval's homepage features a named customer testimonial from Vernon Man, Head of IT, stating that Serval automated over 50% of incoming requests after switching. | 中 | SE001 |
| CE008 | Serval's pricing page lists a "Guaranteed 50% automation rate" as a deployment commitment, alongside a dedicated deployment engineer and a guided pilot phase. | 中 | SE003 |
| CE009 | Serval's Asset Management module is listed as a named product capability on the homepage but receives no substantive description in public documentation or the pricing page, leaving its maturity ambiguous. | 高 | SE001, SE002 |
| CE010 | Serval's core architectural approach deliberately separates tool-building (Automation Agent, used by admins) from tool-execution (Help Desk Agent, used by employees) to limit the blast radius of potential AI errors. | 高 | SE011, SE012 |
| CE011 | CEO Jake Stauch told TechCrunch that the two-agent design prevents a Help Desk Agent from taking destructive unauthorized actions: "Instead it will say, 'Hey, I don't have a tool for deleting all the data the company.'" | 高 | SE011, SE001 |
| CE012 | Serval's workflows are generated as executable code and can be managed in Git, inspected directly by engineers, and integrated into CI/CD pipelines, while also being represented in a no-code UI for non-engineers. | 高 | SE008, SE012 |
| CE013 | The identity of the LLM provider(s) powering Serval's three agents has not been publicly disclosed in any official documentation, press release, or verified media report as of the run date. | 高 | SE001, SE002, SE011 |
| CE014 | Serval's documentation states that workflows execute in a deterministic, multi-step manner to ensure predictable, auditable workflow execution for enterprise compliance. | 中 | SE003 |
| CE015 | No public technical white paper, benchmark study, or independent code-quality assessment of Serval's workflow code-generation engine exists as of the run date. | 高 | SE002, SE011 |
| CE016 | NIST AI 100-2 (March 2025) identifies prompt injection and adversarial inputs as key attack vectors against LLM-based agentic AI systems, representing risks applicable to Serval's Help Desk Agent's inference layer. | 高 | SE015, SE019 |
| CE017 | Serval's deterministic tool-gating architecture mitigates execution-layer AI risks but does not eliminate inference-layer adversarial risks such as prompt injection at the LLM API boundary. | 中 | SE015, SE011, SE019 |
| CE018 | Serval's integrations page lists 60+ named connectors spanning identity providers, HR systems, cloud platforms, ticketing tools, device management, and developer tools. | 高 | SE004, SE011 |
| CE019 | Named integrations verified on the Serval integrations page include: Okta, JumpCloud, Microsoft Graph, Google Workspace, Slack, Jira, Linear, ServiceNow, GitHub, AWS, Rippling, BambooHR, Confluence, Notion, CrowdStrike, Jamf, Kandji, PagerDuty, Opsgenie, incident.io, Rootly, and 15+ additional systems. | 高 | SE004, SE001 |
| CE020 | Serval offers cloud-hosted deployment (fully managed, 99.9% uptime SLA, automatic updates) and self-hosted deployment (customer's private cloud, Kubernetes and Terraform, provision in hours per documentation). | 高 | SE002, SE003 |
| CE021 | Serval states a 99.9% uptime SLA for cloud-hosted deployment, but no public status page was found to independently validate historical availability performance. | 高 | SE002, SE003 |
| CE022 | Serval's self-hosted deployment option uses Kubernetes and Terraform, is targeted at enterprises with data-residency or privacy requirements, and places full infrastructure responsibility on the customer. | 高 | SE002, SE003 |
| CE023 | Serval supports bi-directional ticket synchronization with ServiceNow, Jira, and Zendesk, enabling customers to run Serval alongside incumbents as an AI layer during migration. | 高 | SE003, SE008 |
| CE024 | The Serval AUP explicitly prohibits use of Serval outputs to reverse-engineer the service, build competing AI products, or make fully automated decisions with significant individual impact without meaningful human oversight. | 高 | SE005, SE002 |
| CE025 | Serval's documentation claims SOC 2 Type II certification covering security, availability, processing integrity, confidentiality, and privacy, but no public attestation letter, audit period, or auditor name is disclosed. | 中 | SE002 |
| CE026 | Serval implements TLS 1.3 encryption in transit and AES-256 encryption at rest, per official documentation. | 中 | SE002 |
| CE027 | Serval states it conducts regular penetration testing but discloses no details on cadence, scope, or penetration testing firm; no public bug-bounty program has been found. | 高 | SE002, SE001 |
| CE028 | Serval's AUP commits that AI-generated outputs are labeled "Serval AI" to distinguish them from human-authored content for audit and compliance purposes. | 中 | SE005 |
| CE029 | Serval maintains a Trust Center at trust.serval.com listing AI subprocessors and model providers, but this URL returned a 404 error at the time of research. | 高 | SE005, SE001 |
| CE030 | Serval's AUP states customer materials are not used to train, fine-tune, or improve general-purpose AI or machine learning models without explicit customer authorization. | 中 | SE005 |
| CE031 | No publicly disclosed security incidents, CVE filings, data breaches, or outage reports specifically attributing a failure to Serval's production systems have been found as of the run date. | 中 | SE016, SE017, SE019 |
| CE032 | Reuters reported in December 2025 that Serval had expanded beyond core IT into HR, legal, and finance workflow automation use cases, although no product documentation of these expansions appears on the official website as of the run date. | 中 | SE012 |
| CE033 | Serval's blog confirms the Serval Start program was launched in May 2026, recruiting a 12-person inaugural cohort of technical builders interested in founding companies; the application period has closed. | 高 | SE009, SE010 |
| CE034 | No official product roadmap, versioned changelog, or developer API release history is publicly available for Serval as of the run date. | 高 | SE002, SE009 |
| CE035 | Serval's documentation provides CLI and API access references, and the pricing page confirms full API access to all integrations and custom integration support for any system with an API. | 中 | SE002, SE003 |
| CE036 | Serval's entire product functionality relies on one or more undisclosed third-party LLM API providers; any provider outage, pricing change, or safety-policy update could materially disrupt the product. | 中 | SE013, SE015, SE019 |
| CE037 | No public GitHub repository for Serval's integrations, SDKs, or open-source contributions has been found; the Hacker News "from site:serval.com" page shows no public submissions, and HN Algolia search for "Serval ITSM" returns no relevant engineering discussion threads. | 高 | SE021, SE022, SE001 |
| CE038 | VentureBeat and The Register published articles in early 2026 noting general enterprise risks for agentic AI systems—including hallucination in workflow generation, integration failures at scale, and reliability guarantees—applicable as systemic risks to Serval's product category. | 中 | SE019, SE016 |
| CE039 | Serval's cloud-hosted service is operated by a team of under 30 employees as of December 2025, creating operational concentration risk for a multi-tenant SaaS platform handling enterprise access management workloads. | 中 | SE012, SE014 |
| CE040 | Serval's SCIM provisioning capability integrates with identity providers for continuous lifecycle management, and Just-in-Time access grants are automatically deprovisioned at expiry per the pricing page. | 中 | SE003, SE001 |
| CE041 | Serval's knowledge base module syncs from Notion, Google Drive, and Confluence, and the Help Desk Agent uses RAG to surface answers from synced documentation with source citations. | 中 | SE001, SE002 |
| CE042 | Serval's workflow generation approach produces deterministic executable code stored in Git, contrasting with ServiceNow Flow Designer's drag-and-drop node-based automation and Freshservice's point-and-click workflow builders; this enables version control and CI/CD integration but shifts workflow authoring complexity to the Automation Agent. | 中 | SE008, SE011 |
| CE043 | Serval's NL-to-code differentiation allows enterprise admins to describe an automation in plain language and receive a tested, code-based workflow — a distinct capability from Moveworks, Atomicwork, and Aisera which primarily offer intent-classification-and-action routing without transparent code artifacts that engineers can inspect and version-control. | 中 | SE011, SE012 |
| CE044 | No public data on time-to-value or average onboarding duration for Serval deployments is available; the pricing page references a "guided pilot phase" and a "dedicated deployment engineer" but does not specify timelines or define when the 50% automation rate milestone is expected to be reached. | 中 | SE003, SE002 |
| CE045 | Serval's pricing page lists three tiers — Starter, Business, and Enterprise — without disclosing per-seat or per-automation pricing; the Business tier is described as the most popular and the Enterprise tier offers unlimited seats and custom pricing, indicating a usage-based or negotiated contract model typical of enterprise SaaS. | 中 | SE003, SE001 |
| CU001 | Serval's homepage footer (as of June 2026) lists four named customers: Perplexity, Together.ai, Mercor, and Cribl. | 中 | SU001 |
| CU002 | TechCrunch's October 2025 Series A article stated that Serval's "clients include major AI players like Perplexity, Mercor, and Together AI." | 高 | SU002, SU003 |
| CU003 | Reuters' December 2025 article reported "AI startups, including Perplexity and Together AI, use Serval's platform to automate tasks such as resolving help desk queries and onboarding employees." | 高 | SU003, SU022 |
| CU004 | Sequoia Capital's portfolio page for Serval independently names Perplexity, Mercor, Vercel, Verkada, and Clay as companies using Serval to automate help desk tickets and access requests. | 中 | SU005 |
| CU005 | Serval claims its platform automates more than 50% of IT tickets for its customers, as stated on the company homepage, Sequoia portfolio page, and Reuters article. | 中 | SU001, SU003, SU005 |
| CU006 | Serval reported 500% revenue growth between August and December 2025, according to Reuters; no absolute ARR or MRR figure was disclosed alongside the growth rate. | 中 | SU003, SU022 |
| CU007 | Serval had "just under 30 employees" as of December 2025, according to Reuters, and planned to expand to more than 100 by end of 2026. | 高 | SU003, SU022 |
| CU008 | Sequoia's portfolio page for Serval states the company automates "more than 50% of tickets on day one" as a headline customer outcome metric. | 中 | SU005 |
| CU009 | Vernon Man, identified as Head of IT on Serval's homepage, provided a testimonial stating "Once we switched over to Serval, we were able to complete over 50% of our incoming requests automatically"; his employer is not named. | 低 | SU001 |
| CU010 | All seven named Serval customers — Perplexity, Together AI, Mercor, Vercel, Verkada, Clay, and Cribl — are venture-backed technology companies; none are publicly traded enterprises, financial institutions, government agencies, or regulated-industry buyers. | 高 | SU002, SU005, SU019, SU012, SU013 |
| CU011 | Cribl, an observability and data-routing platform vendor, is listed in Serval's homepage footer as a named customer as of June 2026. | 高 | SU001, SU016 |
| CU012 | Serval's case study pages for Perplexity (serval.com/case-studies/perplexity) and Together AI (serval.com/case-studies/togetherai) both returned HTTP 404 as of the June 2026 research date. | 中 | SU001 |
| CU013 | Serval's /customers page (serval.com/customers) returned HTTP 404 as of June 2026, providing no publicly accessible customer listing or case study index. | 中 | SU001 |
| CU014 | Gartner Peer Insights for the "AI Applications in IT Service Management" category returned a JavaScript-only page with no Serval reviews as of June 2026. | 中 | SU018 |
| CU015 | No G2 reviews for Serval were found as of the June 2026 research date, consistent with the platform's 2024 founding date and early-stage review platform activity. | 中 | SU001 |
| CU016 | Sequoia partner Anas Biad stated: "The last time we heard customer feedback this strong was 16 years ago when we partnered with ServiceNow," per the Reuters December 2025 article. | 高 | SU003, SU022 |
| CU017 | The testimonial from Vernon Man on Serval's homepage does not name his employer, making independent verification of the 50%-automation claim impossible from this source alone. | 中 | SU001 |
| CU018 | Serval has expanded its platform beyond IT departments into HR, legal, and finance, according to Reuters' December 2025 article. | 高 | SU003, SU022 |
| CU019 | All six of the named customers identified via investor portfolio pages — Perplexity, Together AI, Mercor, Vercel, Verkada, and Clay — share at least one investor with Serval (Sequoia or Redpoint), indicating a strong co-investment referral network. | 中 | SU005, SU006, SU019, SU012 |
| CU020 | Perplexity, a named Serval customer, is backed by Sequoia Capital — the same firm that led Serval's $75M Series B at a $1B valuation, creating a direct co-investment relationship. | 高 | SU003, SU005, SU019 |
| CU021 | Serval's go-to-market accommodates both full rip-and-replace of legacy ITSM systems and an "AI layer" deployment mode for customers locked into multi-year incumbent contracts, per the Reuters Series B article. | 高 | SU003, SU008 |
| CU022 | Serval's four-phase onboarding process — Meet, Build, Deploy, Optimize — is documented on the Serval pricing page and implies a structured, services-assisted customer activation pathway. | 中 | SU008 |
| CU023 | Serval offers cloud-hosted, hybrid, and fully self-hosted deployment options, as listed on the company's homepage under the "Built for Enterprise" section. | 中 | SU001 |
| CU024 | Serval integrates with 60-plus enterprise tools including Okta, Google Workspace, GitHub, Jira, Slack, Freshservice, ServiceNow, Ramp, Salesforce, Zoom, AWS, and Kandji, per the company's integrations page and homepage. | 高 | SU001, SU009 |
| CU025 | Vercel (developer deployment platform) and Verkada (physical security technology) extend Serval's disclosed customer base beyond pure AI-native companies to include infrastructure and hardware-adjacent businesses. | 中 | SU005, SU014, SU017 |
| CU026 | Serval has not publicly disclosed net revenue retention, gross revenue retention, churn rate, total customer count, annual contract value, or customer lifetime value as of June 2026. | 高 | SU001, SU003 |
| CU027 | Serval's 500% revenue growth claim is solely attributed to a company statement relayed via Reuters; no third-party audit, customer contract data, or independent metric confirms the growth rate or its base-period starting ARR. | 中 | SU003, SU022 |
| CU028 | Dark Reading reported that security researchers at Acronis identified agentic AI deployments as carrying traditional software vulnerabilities — including authentication bypass and insufficient access controls — at the interface between LLM reasoning and deterministic tool execution. | 中 | SU011 |
| CU029 | Acronis security researchers, cited in Dark Reading, specifically warned that "vibe coded programs" — the same natural-language-to-workflow automation paradigm Serval uses — are deploying without adequate understanding of authentication between the non-deterministic LLM and the deterministic software tools. | 中 | SU011 |
| CU030 | The progression from Serval's Series A (October 2025) to its Series B (December 2025) in under 90 days implies that investors observed rapid and credible customer validation in the intervening weeks sufficient to lead a $75M round. | 中 | SU002, SU003 |
| CU031 | Mercor, an AI-powered hiring automation platform, is confirmed as a Serval customer in TechCrunch's October 2025 article and corroborated by the Redpoint Ventures portfolio context for Serval. | 高 | SU002, SU006, SU012 |
| CU032 | Clay, a go-to-market data and enrichment platform, is listed as a Serval customer in Sequoia Capital's portfolio overview page for Serval. | 中 | SU005, SU015 |
| CU033 | Verkada, a physical security and cloud-managed access control company, is listed as a Serval customer in Sequoia Capital's portfolio overview page for Serval. | 中 | SU005, SU017 |
| CU034 | Serval's homepage footer as of June 2026 specifically names four customers — Perplexity, Together.ai, Mercor, and Cribl — as its disclosed design-partner or production customer list. | 中 | SU001 |
| CU035 | Serval's "vibe coding for IT automation" approach generates deterministic, code-based workflow automations from natural-language descriptions, which are then represented in a no-code UI and can be managed in Git by technical teams. | 高 | SU001, SU002 |
| CU036 | All seven publicly disclosed Serval customers are venture-backed technology companies; no Fortune 500 enterprise, regulated-industry buyer, government agency, or non-tech mid-market customer has been disclosed as of June 2026. | 高 | SU001, SU002, SU003, SU005 |
| CU037 | Serval's homepage displays the tagline "Trusted by world's most innovative companies" without specifying the total number of customers or the criteria for the claim. | 中 | SU001 |
| CU038 | The same core customer names (Perplexity, Together AI, Mercor) appear consistently across the Series A TechCrunch article (October 2025), the Series B Reuters article (December 2025), and Serval's June 2026 homepage — indicating stable customer retention across at least seven months. | 高 | SU001, SU002, SU003 |
| CR001 | Approximately one third of organizations have either already adopted or plan to adopt agentic AI soon, according to self-reported survey data cited by Dark Reading. | 中 | SR001 |
| CR002 | A researcher discovered an overly permissive ServiceNow chatbot protected only by factory default credentials that could be authenticated as any user simply by supplying their email address, allowing creation of powerful AI agents in any company's ServiceNow instance. | 中 | SR001 |
| CR003 | Acronis researchers found that agentic AI vulnerabilities primarily arise not from the AI model itself but from classical software errors—lack of input sanitization, hardcoded credentials, and insufficient access controls—in the deterministic software layer connecting the LLM to enterprise tools. | 中 | SR001 |
| CR004 | NIST AI 100-2 E2025 provides a taxonomy of adversarial machine learning attacks covering prompt injection, data poisoning, model evasion, and model extraction, and identifies current challenges in the life cycle of AI systems including large language models and chatbots. | 高 | SR004, SR005 |
| CR005 | Serval's Acceptable Use Policy prohibits making "fully automated decisions with legal or similarly significant effects on individuals without meaningful human oversight." | 高 | SR011, SR010 |
| CR006 | Serval's AUP requires that a qualified human reviews, validates, and expressly approves any AI-generated responses, recommendations, or actions before they are communicated to end users or relied upon for consequential decisions. | 高 | SR011, SR010 |
| CR007 | Serval's two-agent architecture limits the Help Desk Agent to pre-authorized tools only, preventing unconstrained AI action; the Automation Agent builds tool definitions that the Help Desk Agent then executes within those approved scopes. | 中 | SR012, SR018 |
| CR008 | Serval's integration catalog includes privileged identity systems (Okta, AWS, GitHub, Google Workspace, Workday, Slack), creating a broad attack surface for the AI agent's automation execution layer. | 中 | SR013 |
| CR009 | CISA published guidance in 2026 titled "Careful Adoption of Agentic AI Services" providing actionable steps for organizations to secure agentic AI systems and protect critical infrastructure from AI-driven threats. | 高 | SR007, SR005 |
| CR010 | More than one in five AI-forward companies (22%) had OpenClaw—an agentic AI assistant— running within days of its release, according to Token Security, demonstrating how rapidly uncontrolled agentic AI adoption spreads in enterprise environments. | 中 | SR002 |
| CR011 | Serval claims compliance with SOC 2, HIPAA, and GDPR frameworks and provides auditor-friendly logging capabilities with SIEM integration. | 中 | SR012 |
| CR012 | Serval contractually guarantees a 50% automation rate for IT tickets as an SLA; this guarantee creates direct legal liability if the automation threshold is not met. | 中 | SR014 |
| CR013 | The EU AI Act was published in the Official Journal of the European Union on 12 July 2024 and is the final legally enforceable version of the regulation. | 高 | SR006, SR007 |
| CR014 | The EU AI Act imposes transparency and human oversight obligations for AI systems that interact with humans and may impose additional requirements for AI systems making automated employment-adjacent decisions under its high-risk Annex III provisions. | 中 | SR006 |
| CR015 | GDPR Article 22 restricts solely automated decisions with legal or significantly affecting effects on individuals, requiring meaningful human intervention; Serval's automated IT access provisioning for employment-related systems may engage this provision for EU data subjects. | 中 | SR006, SR011 |
| CR016 | No SEC Form D filing was found for Serval, Inc. in the EDGAR database for the period January 2025 through June 2026, despite approximately $127M in disclosed venture fundraising across the Series A ($47M) and Series B ($75M) rounds. | 高 | SR008, SR009 |
| CR017 | Under Regulation D, companies conducting exempt private securities offerings are generally required to file Form D with the SEC within 15 days of the date of the first sale of securities in the offering. | 高 | SR008, SR009 |
| CR018 | Serval's AUP prohibits using AI features for decisions related to employment, credit, housing, insurance, education, or other consequential areas in a manner that discriminates based on protected characteristics. | 高 | SR011, SR010 |
| CR019 | Serval's AUP prohibits using AI outputs as the sole basis for consequential decisions without independent verification, and specifically prohibits using the service to provide medical, legal, financial, or other regulated professional advice as a substitute for qualified professionals. | 中 | SR011 |
| CR020 | CISA issued guidance in 2026 specifically addressing enterprise deployment of agentic AI, emphasizing governance, access controls, and audit trails as required mitigations. | 高 | SR007, SR005 |
| CR021 | Serval's blog content includes SOC 2 audit automation guides, indicating active attention to enterprise audit and compliance workflows as both a customer use case and internal discipline. | 中 | SR016 |
| CR022 | The California Secretary of State's business registration database serves as a verifiable path to confirm Serval, Inc.'s corporate status, registered agent, and state of incorporation. | 中 | SR034 |
| CR023 | ServiceNow has launched NowAssist AI agents with autonomous multi-step workflow capabilities that directly compete with Serval's core value proposition of AI-native IT service management. | 中 | SR026, SR025 |
| CR024 | ServiceNow serves 85% of the Fortune 500 as the incumbent ITSM platform, giving it unparalleled access to the enterprise buyer relationships Serval must penetrate. | 高 | SR025, SR027 |
| CR025 | Serval's Help Desk Agent operates through Slack, Microsoft Teams, email, and web portal as primary communication channels, creating dependency on these third-party platforms for product delivery. | 中 | SR012, SR013 |
| CR026 | Serval integrates with Okta, Google Workspace, GitHub, AWS, Workday, Slack, and Zoom as primary identity, collaboration, and cloud infrastructure providers; these integrations are central to the product's automation capabilities and represent key dependency risks. | 中 | SR013 |
| CR027 | Sequoia Capital led the Series B and is the named lead investor at $1B valuation; Sequoia's continued endorsement and involvement is critical to Serval's credibility for a Series C financing. | 中 | SR020, SR022 |
| CR028 | Serval has not disclosed which LLM provider or providers power its AI agent layer, creating an undisclosed single-vendor dependency risk that cannot be assessed from public information. | 低 | SR012, SR018 |
| CR029 | ServiceNow's bundled enterprise renewal approach and 20 years of ITSM workflow data create a switching-cost moat that Serval must overcome to win full platform displacement deals. | 中 | SR025, SR026 |
| CR030 | Moveworks, Aisera, and Atomicwork are direct AI-native ITSM competitors with established enterprise customer bases and comparable AI help desk architectures, compressing Serval's window to establish category leadership. | 中 | SR029, SR030, SR027 |
| CR031 | Serval had fewer than 30 employees at the time of its December 2025 Series B, making it one of the smallest teams to reach unicorn status in the enterprise software sector. | 中 | SR019, SR020 |
| CR032 | Serval has stated plans to scale from approximately 30 employees at Series B to more than 100 employees by the end of 2026, representing a 3x headcount expansion within 12 months. | 中 | SR019, SR020 |
| CR033 | Jake Stauch is the only named C-suite executive in any public Serval communication as of the run date; no CTO, CFO, CPO, or CISO has been publicly announced. | 中 | SR012, SR015 |
| CR034 | Serval's Series A ($47M, October 2025) and Series B ($75M, December 2025) closed within approximately seven weeks of each other, indicating investor-driven rapid scaling rather than milestone-gated progression. | 中 | SR019, SR020, SR021 |
| CR035 | Serval raised a total of approximately $127M across Series A ($47M led by Redpoint) and Series B ($75M led by Sequoia), achieving a $1B post-money valuation at Series B close. | 中 | SR019, SR020, SR022, SR023 |
| CR036 | Serval claims 500% revenue growth since August 2025 but has not disclosed absolute ARR, customer count, or any other financial metric, making independent financial sustainability assessment impossible from public sources. | 中 | SR019, SR020 |
| CR037 | Serval's $1B post-money valuation at Series B implies a very high revenue multiple given the likely sub-$10M ARR of a 30-person enterprise AI startup in late 2025, creating significant valuation risk if growth decelerates before Series C. | 低 | SR020, SR019 |
| CR038 | An OpenClaw security researcher demonstrated that an exploit delivered in a support ticket—directly analogous to the input channel Serval's Help Desk Agent processes—took approximately 47 seconds to escalate permissions, access customer records, exfiltrate data, and modify audit logs. | 中 | SR002 |
| CR039 | Serval's pre-authorized tool architecture, which prevents the Help Desk Agent from taking any action outside explicitly pre-approved workflow scopes, is a meaningful structural control that reduces the blast radius of a compromised or misbehaving AI agent. | 中 | SR012, SR018 |
| CR040 | Serval claims SOC 2 Type II compliance, which if verified by an independent auditor would demonstrate established security controls for customer data handling, change management, and availability. | 中 | SR012, SR016 |
| CR041 | The NIST AI Risk Management Framework provides a voluntary but increasingly buyer-required governance structure for AI vendors, including guidance on risk identification, measurement, and management relevant to agentic systems. | 高 | SR005, SR004 |
| CR042 | Key thesis-break triggers for the Serval investment include a production AI security incident, sustained SLA breach, CEO departure before Series C, down round, or ServiceNow replicating the 50% automation SLA guarantee for existing ITSM customers. | 中 | SR012, SR019, SR025 |
| CR043 | Gen security researchers found that approximately 15% of skills on OpenClaw's public ClawHub skills marketplace contained malicious instructions, demonstrating that agentic AI supply chains introduce novel malware distribution vectors. | 中 | SR003 |
| CR044 | The characterization of agentic AI systems as "Formula One cars without brakes" (Dev Rishi, Rubrik) reflects industry consensus that agentic systems operate too quickly for human approval of every risky action, creating a governance gap that architectural controls like Serval's pre-authorization design must address. | 中 | SR002 |
| CV001 | Serval raised $75M in a Series B round at a $1B post-money valuation in December 2025, led by Sequoia Capital, making it one of the fastest enterprise software unicorns in history. | 高 | SV011, SV020, SV013 |
| CV002 | Sequoia Capital partner Anas Biad led the Series B and publicly stated that Serval generated the strongest customer feedback Sequoia had heard since it partnered with ServiceNow sixteen years prior. | 高 | SV011, SV013 |
| CV003 | PitchBook cited an implied Serval valuation of $232M in August 2025; the Series B post-money of $1B in December 2025 represents a 4.3x step-up in approximately four months — an exceptionally compressed unicorn timeline. | 高 | SV010, SV020, SV011 |
| CV004 | Serval's total disclosed venture funding as of December 2025 is $127M, comprising the October 2025 Series A ($47M led by Redpoint Ventures) and the December 2025 Series B ($75M led by Sequoia Capital). | 高 | SV012, SV011, SV014 |
| CV005 | Series A investors alongside Redpoint included First Round Capital, General Catalyst, BoxGroup, Bessemer Venture Partners, Chemistry VC, Strike Capital, Sunflower Capital, and Operator Partners. | 中 | SV012, SV008 |
| CV006 | Series B co-investors alongside Sequoia included returning investors Redpoint Ventures, Meritech Capital, and General Catalyst. | 中 | SV011, SV014 |
| CV007 | Serval had approximately 30 employees at the time of the Series B close in December 2025, making it one of the leanest headcounts at unicorn status in enterprise software history. | 中 | SV020, SV011 |
| CV008 | Serval plans to grow from approximately 30 employees at Series B to more than 100 employees by end of 2026, implying a 3x-plus headcount expansion over twelve months that will materially accelerate cash burn. | 中 | SV020 |
| CV009 | Serval reported 500% revenue growth since August 2025 but has disclosed no absolute ARR or MRR figure, making the growth rate impossible to anchor, verify, or benchmark against comparable-stage companies. | 高 | SV020, SV011 |
| CV010 | An EDGAR full-text search for Form D filings by Serval, Inc. returned no results as of the analysis date, which is unusual for a US company that has accepted approximately $127M in exempt securities offerings under Regulation D. | 中 | SV001 |
| CV011 | US Regulation D requires an issuer to file a Form D notice within 15 days of the first sale of securities in a Regulation D offering; the absence of this filing for Serval may indicate a filing delay, a different legal entity name, or an offshore holding structure. | 中 | SV001 |
| CV012 | ServiceNow reported approximately $10.98B in full-year 2024 subscription and professional services revenue and maintained a market capitalization exceeding $150B, implying a trailing revenue multiple of approximately 13–15x. | 高 | SV002, SV026 |
| CV013 | Atlassian's total annual revenue exceeded $4.4B in FY2024; Jira Service Management is its primary ITSM product line, competing directly with Serval's platform. Atlassian's market capitalization of approximately $40–50B implies roughly 9–11x trailing revenue. | 中 | SV021 |
| CV014 | The global ITSM market is estimated at approximately $11–13B in 2024 by MarketsandMarkets, Grand View Research, GlobeNewswire, Precedence Research, and Business Research Insights — five independent analyst sources with convergent estimates. | 中 | SV003, SV016, SV017, SV005, SV006 |
| CV015 | Independent analyst reports project the ITSM market to grow at 12–17% CAGR through 2029–2032, with one GlobeNewswire report placing the market at $12.57B by 2032 and another at $22.08B by 2029, reflecting different scope assumptions across AI-augmented versus pure-software market definitions. | 中 | SV003, SV004, SV005 |
| CV016 | The AI-enabled and cloud-augmented ITSM sub-segment is the fastest-growing slice of the broader ITSM market, with multiple analyst reports projecting it to reach $22–29B by 2028–2032, approximately doubling the total market estimate as AI-native penetration accelerates. | 中 | SV004, SV006 |
| CV017 | At the $1B Series B post-money valuation with 500% growth from an unknown base, Serval's implied ARR multiple ranges from approximately 17x (if ARR is $60M) to over 200x (if ARR is $5M); the multiple cannot be calculated precisely without an ARR disclosure. | 高 | SV009, SV011 |
| CV018 | If Serval's August 2025 ARR base was approximately $2M (a plausible early-stage anchor), the 500% growth implies roughly $12M ARR by December 2025, implying an approximately 83x ARR multiple at the $1B Series B valuation. | 低 | SV010, SV016 |
| CV019 | Top-quartile early-stage AI SaaS companies in 2025 received 15–40x forward ARR multiples; triple-digit implied multiples indicate either a compressed and unverifiable revenue base or a growth trajectory that dramatically exceeds the peer distribution. | 中 | SV010, SV016 |
| CV020 | ServiceNow trades at approximately 13–15x trailing revenue as a mature, cash-generative ITSM incumbent; Serval's implied multiple must be justified by a growth trajectory that is multiple times faster than ServiceNow's current growth rate. | 中 | SV002, SV016 |
| CV021 | Sequoia partner Anas Biad stated publicly that the last time Sequoia heard customer feedback as strong as Serval's was sixteen years ago when it partnered with ServiceNow — a direct generational framing that positions Serval as a platform-defining ITSM company. | 中 | SV013, SV011 |
| CV022 | Serval's two-agent architecture and 50% ticket automation guarantee are operationally differentiated, but the guarantee's sustainability at enterprise scale with large, diverse IT environments has not been independently validated by any third-party benchmark or audit. | 中 | SV023, SV024 |
| CV023 | TechCrunch Startups Weekly in May 2026 named Serval alongside Atomicwork and Moveworks as the three AI-native ITSM vendors most actively disrupting legacy service management platforms, confirming third-party recognition of Serval's competitive position. | 中 | SV007 |
| CV024 | Moveworks, founded in 2016, raised approximately $300M at a $2.1B valuation — a comparable but more mature AI ITSM peer with substantially greater engineering headcount and undisclosed ARR, suggesting the market values AI-native ITSM at premium multiples. | 中 | SV028 |
| CV025 | Aisera, founded in 2017, has raised over $100M for AI-native ITSM with a valuation that is not publicly disclosed, limiting direct comparability; it is used as a qualitative landscape reference rather than a numerical comparable. | 中 | SV029 |
| CV026 | Axios reported in January 2026 that venture-backed AI startups with unverified revenues, self-reported growth rates, and implied multiples exceeding 50x face heightened LP scrutiny and materially elevated down-round risk in 2026 financing markets — a profile that Serval directly matches. | 中 | SV009 |
| CV027 | Precedence Research and Business Research Insights project the global ITSM market to reach $12–18B by 2030–2032, providing independent cross-checks on the MarketsandMarkets and Grand View Research forecasts and reinforcing the structural growth thesis for AI-native ITSM vendors. | 中 | SV005, SV006 |
| CV028 | Gartner's AI Applications in ITSM market reviews show a rapidly expanding vendor landscape with no consensus leader in pure AI-native ITSM as of early 2026, indicating category fluidity that benefits early movers like Serval but also signals low switching costs. | 中 | SV018 |
| CV029 | PitchBook independently tagged Serval as a $1B-valuation company in December 2025 following the Sequoia-led round, providing third-party corroboration of the post-money mark that is independent of any Serval press release or investor portfolio page. | 高 | SV010, SV011 |
| CV030 | The absence of any publicly named Fortune 500, financial services, healthcare, or government customer leaves Serval's enterprise vertical expansion thesis unproven, introducing material TAM capture uncertainty in the base and bear case scenarios. | 中 | SV023, SV025 |
| CV031 | First Round Capital, as a Series A participant, has a documented investment track record that includes ITSM-adjacent and enterprise software companies; its inclusion signals early product-market fit endorsement by a firm with demonstrable pattern recognition. | 中 | SV008, SV012 |
| CV032 | ServiceNow's generative AI module, Now Assist, has been deployed to its more than 7,700 enterprise ITSM customers as an integrated capability on the Now Platform, representing a bundling threat that is the primary bear-case risk for Serval's commercial window. | 高 | SV002, SV027 |
| CV033 | Dark Reading reported that agentic AI systems deployed in enterprise IT environments face documented vulnerabilities including prompt injection, privilege escalation, and authentication bypass — precisely the risk surface of Serval's two-agent architecture in production customer environments. | 中 | SV022 |
| CV034 | Sequoia Capital's historical enterprise software portfolio — ServiceNow, Snowflake, Figma, and Nubank — provides a precedent for how Sequoia prices and supports generational software bets; Serval's $1B valuation framing is consistent with Sequoia's historical entry-price architecture for category-defining companies. | 中 | SV013 |
| CV035 | Atlassian's trailing revenue multiple has compressed from approximately 30x in 2021 to approximately 9–11x in 2024–2025, illustrating the SaaS multiple compression risk that could affect Serval's exit valuation if growth moderates before a liquidity event. | 中 | SV021 |
| CV036 | The ITSM market's subscription-based recurring revenue structure — with annual contracts and multi-year renewals — supports higher sustainable long-run revenue multiples than transaction-based software models, partially justifying a premium for early-stage AI-native ITSM vendors. | 中 | SV016, SV017 |
| CV037 | Bull case: if Serval achieves $40–60M ARR by end of 2026 with 150%+ NRR and expands beyond tech-native customers, a 25–30x ARR multiple at a 2027–2028 exit implies a $1.25–$1.8B valuation, generating 1.3–1.8x gross return from the Series B entry price. | 低 | SV010, SV017 |
| CV038 | Bear case: if ServiceNow bundles Now Assist at zero incremental cost to existing ITSM customers by Q3 2026, Serval's commercial window compresses to sub-500-employee tech companies; at $10–15M ARR with a 12x multiple, the 2028 exit implies a $150–$225M valuation — a 75–85% loss from the Series B entry. | 中 | SV027, SV009 |
| CV039 | Base case: Serval achieves $20–35M ARR by end of 2026 through continued tech-native customer expansion; a 20x ARR multiple at a 2028 exit implies a $500M–$875M valuation range — approximately flat to slightly below the Series B entry, generating minimal return for late-stage Series B investors. | 中 | SV010, SV016 |
| CV040 | Hacker News discussion threads linked to serval.com show strong positive sentiment from IT practitioners, developers, and system administrators, providing social-proof validation of Serval's technical appeal outside formal press channels. | 中 | SV025 |
| CV041 | IBM's IT service management educational content documents the secular shift from reactive ticket management to proactive AI-driven resolution, validating the long-run market direction that Serval's AI-native architecture targets. | 中 | SV030 |
| CV042 | No independent audit, due diligence report, or secondary analyst write-up has published an ARR estimate for Serval; all financial modeling of the $1B Series B valuation relies on range-based inference from comparable-company multiples applied to an unverifiable revenue base. | 高 | SV009, SV010 |
| 编号 | 出版方 | 标题 | 引文 |
|---|---|---|---|
| SO001 | Serval, Inc. | Serval: AI-native IT service management software (homepage) | AI Agents for IT. Automate help desk requests, just-in-time access, onboarding and offboarding with AI-powered workflows that build themselves. |
| SO002 | TechCrunch | Serval raises $47M to bring AI agents to IT service management | Serval CEO Jake Stauch says the key was to make the process of building a tool as simple as possible. "We don't want them to feel the marginal cost of building these automations. We want to make it easier to automate something forever than do it manually once." |
| SO003 | US News & World Report (via Reuters) | AI Startup Serval Valued at $1 Billion After Sequoia-Led Round to Expand IT Automation | Serval said it had grown revenue by 500% since August, without disclosing details. The company, which currently has just under 30 employees, expects to expand to more than 100 next year. |
| SO004 | VC Tavern | Serval Becomes Unicorn After $75M Series B Led by Sequoia Capital | The Series A also included participation from several well-known investors in the technology ecosystem, including First Round Capital, General Catalyst, and BoxGroup. Additional backing came from Bessemer Venture Partners, Chemistry VC, Strike Capital, Sunflower Capital, and Operator Partners. |
| SO005 | Serval, Inc. | Serval Blog — IT service management insights and guides | |
| SO006 | Serval, Inc. | Careers at Serval | |
| SO007 | Sequoia Capital | Serval — Sequoia Capital Portfolio | Serval deploys AI agents for IT. Companies such as Perplexity, Mercor, Vercel, Verkada and Clay use Serval to automate help desk tickets, just-in-time access requests, onboarding and offboarding flows, and more. With Serval, IT teams turn natural language prompts into complex automations in seconds, automating more than 50% of tickets on day one. |
| SO008 | Redpoint Ventures | Serval — Redpoint Portfolio | Founders Jake Stauch, Alex McLeod. Location San Francisco, CA. Stage Invested Series A. Lead. Alex Bard, Patrick Chase, Jordan Segall. |
| SO009 | Serval, Inc. | Serval pricing: AI-native ITSM software | Get a quote | |
| SO010 | Serval, Inc. | Welcome to Serval — Serval Documentation | SOC 2 Type II certified. Encryption — TLS 1.3 in transit, AES-256 at rest. Connect Serval to 60+ tools including Slack, Okta, Google Workspace, GitHub, Jira, and more. |
| SO011 | Serval, Inc. | Serval ITSM integrations: Okta, AWS, ServiceNow & more | |
| SO012 | Serval, Inc. | Acceptable Use Policy | Serval | |
| SO013 | Serval, Inc. | Serval Start — The Future Founder Program | Jake Stauch, CEO and Founder of Serval. Christine Kim, Head of Strategic Projects at Serval. |
| SO014 | General Catalyst | Serval | General Catalyst Portfolio | Serval is an AI-native IT service management platform with a built-in agent workforce that unifies help desk, access management, and automation, enabling IT teams to resolve over half of their tickets autonomously. Investors: Marc Bhargava, Vedant Suri, Kate Bender. Backed since: 2024. |
| SO015 | IBM | What Is ITSM (IT Service Management)? | IBM | |
| SO016 | Gartner | Best AI Applications in IT Service Management Reviews 2026 | Gartner Peer Insights | Gartner Peer Insights content consists of the opinions of individual end users based on their own experiences, and should not be construed as statements of fact, nor do they represent the views of Gartner or its affiliates. Gartner does not endorse any vendor, product or service depicted in this content. |
| SO017 | ServiceNow | IT Service Management (ITSM) — ServiceNow | |
| SO018 | Atomicwork | Atomicwork: The AI Workforce platform for modern IT teams | |
| SO019 | Moveworks | Moveworks: The AI Assistant Platform That Gets Work Done | |
| SO020 | Perplexity AI | Perplexity Blog | |
| SO021 | ServiceNow | Generative AI — ServiceNow Now Platform | |
| SO022 | Hacker News | Submissions from serval.com | Hacker News | |
| SO023 | IBM | What Is ITSM (IT Service Management)? | IBM | |
| SO024 | Serval, Inc. | Serval blog — index of published articles | |
| SO025 | Serval, Inc. | Serval Documentation — llms.txt index | |
| SM001 | Serval Inc. | Serval: AI-native IT service management software | Automate help desk requests, just-in-time access, onboarding and offboarding with AI-powered workflows that build themselves. |
| SM002 | TechCrunch | Serval raises $47M to bring AI agents to IT service management | Serval is using agentic AI models to automate IT service management, but the company has a unique approach that takes advantage of agentic AI's powers while avoiding many of its pitfalls. |
| SM003 | VC Tavern | Serval Becomes Unicorn After $75M Series B Led by Sequoia Capital | Serval, an AI-native IT service management startup based in San Francisco, has attracted strong investor backing in a short period of time, culminating in a $75 million Series B funding round that valued the company at $1 billion. |
| SM004 | MarketsandMarkets | IT Service Management (ITSM) Market by Offering, Deployment Model, Organization Size, Vertical and Region – Global Forecast to 2028 | The ITSM market size is projected to reach USD 22.1 billion by 2028 at a Compound Annual Growth Rate (CAGR) of 15.9% during the forecast period. |
| SM005 | BMC Software | IT Service Management Solutions | BMC Helix | Boost your teams with generative AI agents to surface valuable insights, make recommendations, find and summarize knowledge, and automate tasks. |
| SM006 | ServiceNow | IT Service Management (ITSM) - ServiceNow | Use autonomous AI specialists to handle routine requests and move toward zero-touch service with ServiceNow ITSM. |
| SM007 | ServiceNow | ServiceNow AI Platform | Access data from 450+ systems, including SAP and Salesforce, in one platform. Give AI the context it needs to act across your business. |
| SM008 | ServiceNow | Generative AI - ServiceNow | |
| SM009 | Aisera | Aisera: Agentic AI for the Enterprise | Auto-resolved 70% of tickets... Saved $2.2M in support costs. |
| SM010 | Freshworks (Freshservice) | Freshservice Pricing & Plans 2026 | Freshworks | TRUSTED BY 74,000+ BUSINESSES WORLDWIDE |
| SM011 | Atlassian | Service Collection Pricing: Free and Paid Plans | |
| SM012 | Moveworks | Moveworks: The AI Assistant Platform That Gets Work Done | |
| SM013 | Moveworks | Moveworks: One AI Assistant Platform for Every Workflow | Cut through operational complexity by unifying your siloed content systems and business applications... with an intuitive, AI-native experience that meets employees where they work, and in over 100 languages. |
| SM014 | Atomicwork | Atomicwork: The AI Workforce platform for modern IT teams | |
| SM015 | IBM | What Is ITSM (IT Service Management)? | IBM | Information technology service management (ITSM) is the practice of planning, implementing, managing and optimizing the end-to-end delivery of information technology services to meet user needs and business goals. |
| SM016 | Serval Inc. | Serval pricing: AI-native ITSM software | Get a quote | Guaranteed 50% automation rate |
| SM017 | Serval Inc. | Serval ITSM integrations: Okta, AWS, ServiceNow & more | 30+ native integrations |
| SM018 | Serval Inc. | Welcome to Serval - Serval Documentation | Compliance — SOC 2 Type II certified |
| SM019 | Atlassian | Jira Service Management Features | |
| SM020 | Redpoint Ventures | Serval — Redpoint portfolio | |
| SM021 | Sequoia Capital | Serval — Sequoia portfolio | |
| SM022 | Serval Inc. | Serval blog: IT service management insights and guides | |
| SM023 | Serval AI — LinkedIn company page | ||
| SM024 | Atlassian | Atlassian Investor Relations | |
| SM025 | BMC Software | AIOps Platform Market — MarketsandMarkets listing | |
| SM026 | Freshworks (Freshservice) | AI-Powered ITSM Platform | Simple, Scalable IT Service for Every Business | |
| SP001 | Serval AI | Serval AI Product Overview | |
| SP002 | Serval AI | Serval AI Pricing | |
| SP003 | Serval AI | Serval AI Integrations | |
| SP004 | Serval AI | Serval AI Customer Stories | |
| SP005 | ServiceNow | ServiceNow ITSM Product Page | |
| SP006 | ServiceNow | ServiceNow AI Agents (NowAssist) | |
| SP007 | ServiceNow | ServiceNow ITSM Landing Page | |
| SP008 | ServiceNow | ServiceNow ITSM Pricing Page | |
| SP009 | Freshworks / Freshservice | Freshservice Home | |
| SP010 | Freshworks / Freshservice | Freshservice IT Service Management | |
| SP011 | Freshworks / Freshservice | Freshservice Pricing | |
| SP012 | Atlassian | Jira Service Management Home | |
| SP013 | Atlassian | Jira Service Management Features | |
| SP014 | Atlassian | Jira Service Management Pricing | |
| SP015 | Moveworks | Moveworks Home | |
| SP016 | Moveworks | Moveworks Platform | |
| SP017 | Moveworks | Moveworks Pricing | |
| SP018 | Moveworks | Moveworks Customers | |
| SP019 | Atomicwork | Atomicwork Home | |
| SP020 | Atomicwork | Atomicwork Product | |
| SP021 | Atomicwork | Atomicwork Pricing | |
| SP022 | Aisera | Aisera Home | |
| SP023 | Aisera | Aisera IT Solutions | |
| SP024 | Aisera | Aisera Gartner MQ for AI Applications in ITSM | |
| SP025 | Zendesk | Zendesk Ticketing System | |
| SP026 | Gartner | Gartner Reviews: AI Applications in IT Service Management | |
| SP027 | BMC Software | BMC Helix IT Service Management | |
| SP028 | U.S. News & World Report | AI Startup Serval Valued at $1 Billion After Sequoia-Led Round | |
| SP029 | TechCrunch | Serval raises $47M to bring AI agent to IT service management | |
| SP030 | VCTavern | Serval Becomes Unicorn After $75M Series B Led by Sequoia Capital | |
| SI001 | Serval | Serval — AI-native IT Service Management | |
| SI002 | Serval | Serval Pricing — one platform fee, no surprises | |
| SI003 | Serval | Serval Integrations — 60+ enterprise tools | |
| SI004 | Serval | Serval Careers | |
| SI005 | Serval | Serval Blog | |
| SI006 | Serval | Serval Acceptable Use Policy | |
| SI007 | Serval | Serval Documentation — SOC 2, SLA, integrations | |
| SI008 | TechCrunch | Serval raises $47 million to bring AI agent to IT service management | |
| SI009 | Reuters / US News | AI startup Serval valued at $1 billion after Sequoia-led round to expand IT automation | |
| SI010 | VCTavern | Serval becomes unicorn after $75M Series B led by Sequoia Capital | |
| SI011 | Sequoia Capital | Serval — Sequoia Portfolio | |
| SI012 | Redpoint Ventures | Serval — Redpoint Portfolio | |
| SI013 | ServiceNow | IT Service Management — ServiceNow Now Platform | |
| SI014 | ServiceNow | ServiceNow Generative AI and Now Assist | |
| SI015 | ServiceNow | ServiceNow Investor Relations | |
| SI016 | Freshworks | Freshservice — AI-powered ITSM | |
| SI017 | Atlassian | Jira Service Management — Atlassian | |
| SI018 | Moveworks | Moveworks AI Platform | |
| SI019 | Atomicwork | Atomicwork — AI-first ITSM | |
| SI020 | Aisera | Aisera — AI Service Management | |
| SI021 | MarketsandMarkets | IT Service Management (ITSM) Market — Global Forecast to 2030 | |
| SI022 | Grand View Research | IT Service Management Market Size, Share & Trends Analysis Report | |
| SI023 | Statista | IT Service Management — Statista Topic Overview | |
| SI024 | U.S. Securities and Exchange Commission | EDGAR Company Search — Serval Inc. Form D (no results found) | |
| SI025 | Dark Reading / Acronis | In the Mad Dash to Deploy Agentic AI, Security Is an Afterthought | |
| SI026 | BMC Software | IT Service Management (ITSM) Solutions — BMC Helix | |
| SI027 | ServiceNow | ServiceNow Reports Fourth Quarter and Full Year 2024 Financial Results | |
| SI028 | GlobeNewsWire / MarketsandMarkets | IT Service Management Market to Reach USD 22.08 Billion by 2029 — MarketsandMarkets via GlobeNewsWire | |
| SE001 | Serval | Serval — AI-Native IT Service Management | Serval is a unified help desk, access management, and automation platform with a built-in AI agent workforce. |
| SE002 | Serval | Serval Documentation — Platform Overview | Encryption — TLS 1.3 in transit, AES-256 at rest. Compliance — SOC 2 Type II certified. Deployment Options: Cloud-Hosted — Fully managed service — No infrastructure to manage. Automatic updates, 99.9% uptime SLA. Self-Hosted — Deploy in your private cloud — Full control over your environment. Provision in hours with Terraform and Kubernetes. |
| SE003 | Serval | Serval Pricing and Product Overview | Guaranteed 50% automation rate. SCIM provisioning — Integrates easily with identity providers for continuous lifecycle management. Custom provisioning workflows (via API, Terraform). |
| SE004 | Serval | Serval Integrations — 60+ Connectors | Custom App — Build a custom integration tailored to your unique workflows and systems. [60+ integrations spanning Access Management, Knowledge Base, Ticket Syncing, Workflows] |
| SE005 | Serval | Serval Acceptable Use Policy | Serval maintains a current list of AI subprocessors and model providers on its Trust Center (trust.serval.com). Customer Materials processed by AI features are handled in accordance with the Agreement and the Data Processing Addendum. Serval does not use Customer Materials to train, fine-tune, or improve general-purpose AI or machine learning models unless Customer provides express authorization. |
| SE006 | Serval | Serval Case Study — Together AI | |
| SE007 | Serval | Serval — Customer Success Stories | |
| SE008 | Serval | Serval Product Page | Serval builds workflows in code and represents them in a no-code UI. This gives you the full power and reliability of code-based workflows without needing to be an engineer, while technical teams can inspect workflow code directly and manage workflows in Git. |
| SE009 | Serval | Serval Blog | |
| SE010 | Serval | Serval Start — Program for Aspiring Founders | Build Real AI Systems for the World's Largest Enterprises. Applications are now closed. [Inaugural cohort of approximately 12 participants] |
| SE011 | TechCrunch | Serval raises $47 million to bring AI agent to IT service management | One agent is used to code internal automations for everyday tasks, like authorizing software or provisioning a device. A separate help desk agent responds to user requests by calling those tools on command, following rules established by the tool. "You don't want someone to go into Slack and say, 'Hey, I want to delete all the data at the company,' and the very helpful AI agent responds, 'Great, I'll delete all the data.'" |
| SE012 | Reuters | AI startup Serval raises $75 mln in Series B funding | Serval's platform uses a two-part AI system, including an AI agent that interacts with employees to understand their support requests, as well as a tool that allows administrators to build complex automations by describing them in natural language, a process the CEO calls "vibe coding." This approach generates code behind the scenes, providing more flexibility and power than traditional drag-and-drop workflow builders. It also serves as a security measure, as the employee-facing AI can only trigger automations that have been pre-approved by administrators. |
| SE013 | BusinessWire | Serval Raises 47 Million Series A to Power the First AI Agent for IT | |
| SE014 | BusinessWire | Serval Becomes Unicorn With 75 Million Series B | |
| SE015 | NIST (National Institute of Standards and Technology) | NIST AI 100-2 E2025: Adversarial Machine Learning — A Taxonomy and Terminology of Attacks and Mitigations | This NIST Trustworthy and Responsible AI report provides a taxonomy of concepts and defines terminology in the field of adversarial machine learning (AML)… identifying current challenges in the life cycle of AI systems and methods for mitigating and managing the consequences of those attacks. |
| SE016 | The Register | Serval raises $47M Series A for AI IT service management | |
| SE017 | The Register | Serval Series B — AI ITSM startup reaches unicorn status | |
| SE018 | VentureBeat | Serval raises $47M Series A for AI-native IT service management | |
| SE019 | VentureBeat | The risks of agentic AI in enterprise IT operations | |
| SE020 | CIO Dive | CIO Dive — Enterprise Technology News and Analysis | |
| SE021 | Hacker News (Algolia API) | HN Algolia search results for "Serval ITSM" | |
| SE022 | Hacker News | Submissions from serval.com on Hacker News | |
| SE023 | Semafor | Serval ITSM startup raises Series B | |
| SE024 | Crunchbase News | Serval AI ITSM reaches unicorn status after Sequoia Series B | |
| SE025 | Redpoint Ventures | Serval — Redpoint Portfolio Page | Serval is a next-generation AI-native platform for IT automation, helping automate help-desk tickets, access provisioning, onboarding/offboarding and complex workflows using natural-language instructions instead of manual coding. |
| SU001 | Serval | Serval: AI-native IT service management software (homepage) | "Once we switched over to Serval, we were able to complete over 50% of our incoming requests automatically." — Vernon Man, Head of IT (company unnamed) |
| SU002 | TechCrunch | Serval raises $47 million to bring AI agent to IT service management | "clients include major AI players like Perplexity, Mercor, and Together AI" |
| SU003 | Reuters / US News | AI startup Serval valued at $1 billion after Sequoia-led round to expand IT automation | "AI startups, including Perplexity and Together AI, use Serval's platform to automate tasks such as resolving help desk queries and onboarding employees. Serval said its technology automates more than 50% of IT tickets for its customers." |
| SU004 | VC Tavern | Serval becomes unicorn after $75M Series B led by Sequoia Capital | |
| SU005 | Sequoia Capital | Serval — Sequoia portfolio company page | "Companies such as Perplexity, Mercor, Vercel, Verkada and Clay use Serval to automate help desk tickets … automating more than 50% of tickets on day one." |
| SU006 | Redpoint Ventures | Serval — Redpoint portfolio company page | |
| SU007 | General Catalyst | Serval — General Catalyst portfolio page | |
| SU008 | Serval | Serval pricing and onboarding overview | |
| SU009 | Serval | Serval integrations page | |
| SU010 | Serval | Serval documentation — overview | |
| SU011 | Dark Reading | Agentic AI Isn't Risky; the Way Orgs Deploy It Is | "Slapdash 'Agentforce' or 'Now Assist' agents, or increasingly common vibe coded programs [are deploying] without a deep understanding of how these systems work … People are going to [deploy agentic AI] without a deep understanding of how these systems work, and how they're connected to each other." |
| SU012 | Mercor | Mercor — AI-powered hiring platform (homepage) | |
| SU013 | Together AI | Together AI — AI cloud platform (homepage) | |
| SU014 | Vercel | Vercel — Developer cloud platform (homepage) | |
| SU015 | Clay | Clay — Go-to-market data and enrichment (homepage) | |
| SU016 | Cribl | Cribl — Data observability and routing platform (homepage) | |
| SU017 | Verkada | Verkada — Physical security technology platform (homepage) | |
| SU018 | Gartner | Gartner Peer Insights — AI Applications in IT Service Management market | |
| SU019 | Perplexity AI | Perplexity AI — AI answer engine (homepage) | |
| SU020 | Business Wire | Serval Raises $47 Million Series A to Power the First AI Agent for IT | |
| SU021 | Business Wire | Serval Becomes Unicorn With $75 Million Series B | |
| SU022 | Reuters | AI startup Serval raises $75 mln Series B funding | |
| SU023 | The Register | Serval raises $47M Series A for AI-powered IT service management | |
| SU024 | The Register | Serval hits $1B valuation with $75M Series B | |
| SU025 | Semafor | Serval ITSM startup raises Series B | |
| SU026 | Serval | Serval Start program — path to founding | |
| SU027 | Hacker News | Hacker News — submissions from serval.com | |
| SR001 | Dark Reading | Agentic AI Is Risky: Here's How to Secure It | A researcher discovered a dangerous exploit chain in ServiceNow. Thanks to an overly permissive chatbot—protected only by a factory default credential—that could be authenticated as any user simply by supplying their email address, the researcher found that he could access and create powerful AI agents in any company's ServiceNow instance. |
| SR002 | Dark Reading | For Enterprises, Security Remains Agentic AI's Biggest Challenge | Demonstrating the need for a new security architecture took about 47 seconds. That's how long an exploit—delivered in a support ticket—needed to escalate permissions, access customer records, exfiltrate data, and modify its own audit logs to cover its tracks. |
| SR003 | Dark Reading | OpenClaw's Insecurities Make Safe Usage Difficult | Our current research shows roughly 15% of the skills we've seen contained malicious instructions. |
| SR004 | National Institute of Standards and Technology | NIST AI 100-2 E2025: Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations | This NIST Trustworthy and Responsible AI report provides a taxonomy of concepts and defines terminology in the field of adversarial machine learning (AML). The taxonomy is arranged in a conceptual hierarchy that includes key types of ML methods, life cycle stages of attack, and attacker goals, objectives, capabilities, and knowledge. |
| SR005 | National Institute of Standards and Technology | NIST Artificial Intelligence — AI Risk Management Framework | |
| SR006 | artificialintelligenceact.eu | The EU Artificial Intelligence Act — Official Journal Version (2024) | The EU AI Act was published in the Official Journal of the European Union on 12 July 2024. This is the final version of the text to be enforced. |
| SR007 | Cybersecurity and Infrastructure Security Agency | Artificial Intelligence at CISA — Careful Adoption of Agentic AI Services | This guidance from the ASD's ACSC, along with CISA and other U.S. and international partners outlines actionable steps for organizations to secure agentic AI systems and protect critical infrastructure from evolving AI-driven threats. |
| SR008 | U.S. Securities and Exchange Commission | EDGAR Full-Text Search — Form D Filings for Serval Inc. | |
| SR009 | U.S. Securities and Exchange Commission | EDGAR API Search — Serval San Francisco Form D 2025-2026 | API response returns hits.total.value = 0; no Form D filings match Serval in San Francisco for the period January 2025 through June 2026. |
| SR010 | Serval, Inc. | Serval Legal — Terms and Policies | |
| SR011 | Serval, Inc. | Serval Acceptable Use Policy | Make fully automated decisions with legal or similarly significant effects on individuals without meaningful human oversight. |
| SR012 | Serval, Inc. | Serval — AI Agents for IT (homepage) | Serval complies with SOC 2, HIPAA, and GDPR frameworks, and provides auditor-friendly logging capabilities with SIEM integration. |
| SR013 | Serval, Inc. | Serval Integrations Catalog | |
| SR014 | Serval, Inc. | Serval Pricing | |
| SR015 | Serval, Inc. | Serval Careers | |
| SR016 | Serval, Inc. | Serval Blog | |
| SR017 | Serval, Inc. | Serval Start — Founder Program | |
| SR018 | Serval, Inc. | Serval Documentation | |
| SR019 | TechCrunch | Serval raises $47M to bring AI agents to IT service management | |
| SR020 | U.S. News & World Report | AI Startup Serval Valued at $1 Billion After Sequoia-Led Round to Expand IT Automation | |
| SR021 | VC Tavern | Serval Becomes Unicorn After $75M Series B Led by Sequoia Capital | |
| SR022 | Sequoia Capital | Serval — Sequoia Portfolio | |
| SR023 | Redpoint Ventures | Serval — Redpoint Portfolio | |
| SR024 | General Catalyst | Serval — General Catalyst Portfolio | |
| SR025 | ServiceNow | ServiceNow ITSM — AI-Powered IT Service Management | 85% of the Fortune 500 run on ServiceNow. |
| SR026 | ServiceNow | ServiceNow AI Agents | |
| SR027 | Gartner | Gartner Reviews — AI Applications in IT Service Management | |
| SR028 | IBM | What Is IT Service Management (ITSM)? | |
| SR029 | Moveworks | Moveworks Platform | |
| SR030 | Aisera | Aisera AI for IT | |
| SR031 | SOC 2 Compliance | ITSM AI Automation Compliance | |
| SR032 | InfoQ | Serval AI Raises $75M Series B for ITSM Automation | |
| SR033 | SiliconAngle | Serval Raises $47M to Replace Legacy ITSM With AI Agents | |
| SR034 | California Secretary of State | California Business Search — bizfileonline.sos.ca.gov | |
| SR035 | CIO Dive | CIO Dive — Information and Enterprise Technology News | |
| SV001 | SEC EDGAR Full-Text Search | EDGAR full-text search — Form D filings for Serval, Inc. | |
| SV002 | ServiceNow Investor Relations | ServiceNow Reports Fourth Quarter and Full Year 2024 Financial Results | |
| SV003 | GlobeNewswire | IT Service Management Market Size Expected to Reach USD 12.57 Billion by 2032, Growing at CAGR of 12.1% | |
| SV004 | GlobeNewswire | IT Service Management Market to Reach USD 22.08 Billion by 2029, Report by MarketsandMarkets | |
| SV005 | Precedence Research | IT Service Management (ITSM) Market Size, Share & Trends Analysis Report | |
| SV006 | Business Research Insights | IT Service Management (ITSM) Market Research Report 2024–2032 | |
| SV007 | TechCrunch | Startups Weekly: The AI agents disrupting ITSM | |
| SV008 | First Round Capital | First Round Capital — Portfolio Companies | |
| SV009 | Axios | AI startups with unverifiable revenues face heightened LP scrutiny in 2026 | |
| SV010 | PitchBook | Serval hits $1B valuation on new funding round | |
| SV011 | VCTavern | Serval becomes unicorn after $75M Series B led by Sequoia Capital | |
| SV012 | TechCrunch | Serval raises $47 million to bring AI agents to IT service management | |
| SV013 | Sequoia Capital | Sequoia Capital — Serval portfolio page | |
| SV014 | Redpoint Ventures | Redpoint Ventures — Serval portfolio page | |
| SV015 | General Catalyst | General Catalyst — Serval portfolio page | |
| SV016 | MarketsandMarkets | IT Service Management (ITSM) Market — Global Forecast to 2029 | |
| SV017 | Grand View Research | IT Service Management Market Size, Share & Trends Analysis Report | |
| SV018 | Gartner | Gartner Peer Insights — AI Applications in IT Service Management | |
| SV019 | Statista | IT Service Management (ITSM) — Statistics & Facts | |
| SV020 | US News & World Report / Reuters | AI startup Serval valued at $1 billion after Sequoia-led round to expand IT automation | |
| SV021 | Atlassian | Atlassian — Investor Relations | |
| SV022 | Dark Reading | Agentic AI in the Enterprise — Security Risks You Cannot Ignore | |
| SV023 | Serval | Serval — AI-native IT Service Management | |
| SV024 | Serval | Serval Pricing — one platform fee, no surprises | |
| SV025 | Hacker News | Hacker News — stories from serval.com | |
| SV026 | ServiceNow | ServiceNow IT Service Management (ITSM) | |
| SV027 | ServiceNow | ServiceNow — Generative AI on the Now Platform | |
| SV028 | Moveworks | Moveworks — AI platform for enterprise IT | |
| SV029 | Aisera | Aisera — AI Service Management | |
| SV030 | IBM | What is IT Service Management (ITSM)? |