You.com
$1.5B 估值节点上的企业 AI 搜索基础设施
You.com 的产品宽度、客户验证和融资能力都可信,但公开证据更能支持「仍有相关性」,还不足以支撑耐久经济性;$1.5B Series C 估值显得偏紧。
封面要素
公司概况
You.com 是一家私营 AI 搜索与研究基础设施公司,由 Richard Socher 和 Bryan McCann 于 2020 年创立。公司起步时是 AI 原生搜索挑战者,如今定位成企业 AI 基础设施,围绕有网页依据的 API、私有数据检索、高级研究工作流和定制企业部署搭建产品。公开证据支持 Palo Alto 总部锚点,以及更广的多地运营足迹;2025 年公司完成真实的 $100 million Series C,估值 $1.5 billion;在机构、企业和 API 平台场景中也有实质客户证明。但关键经营指标仍然稀疏。
- 官网
- you.com
- 成立时间
- 2020-01-01
- 创始人
- Richard Socher, Bryan McCann
- 创立地点
- Palo Alto, CA
- 总部
- Palo Alto, CA
- 产品
- You.com 销售分层产品栈,包括网页搜索和内容 API、高级 Research 和 Finance Research API、私有数据检索,以及定制企业代理或工作流部署。
- 客户
- 需要新鲜、有引用依据的网页 grounding、研究工作流,以及跨私有和公开数据的权限感知检索的企业团队和开发者。
- 商业模式
- 混合变现模式,结合自助式按用量计费 API、高级研究和金融 API,以及经济条款定制程度更高、服务更重的企业项目。
- 阶段
- Series C / late-stage private
- 融资情况
- 截至 2025 年 $100M Series C(估值 $1.5B),公开披露融资约 ~$195M;其他地方引用的更高合计金额,尚未被已撰写章节充分验证。
执行摘要
主要优势
- 创始人市场契合度异常强,Richard Socher 与 Bryan McCann 的 NLP 和企业软件履历提供核心支撑。
- 产品宽度真实存在:You.com 现在覆盖基础搜索、高端研究、金融工作流和私有数据企业部署。
- 公开客户验证强于普通 AI 创业公司叙事:NIH、Maryville、Mimecast 给出可量化工作流结果,API 客户也有具名案例。
- 融资能力可信:$100M Series C、$1.5B 估值有较好交叉验证,可见披露融资额约 $195M。
- 在带引用、模型无关、基于网页 grounding 的工作流里,公司对企业和 agentic 用例有一个说得通的切入口。
主要风险
- 收入、毛利率、留存、集中度和 runway 均未披露,业务无法被干净承销。
- 直接 retrieval 与搜索 API 定价在同业中已被压缩,基础层毛利率耐久性承压。
- 多个头部 traction 信号,包括月 API 查询量 >1B,主要仍是公司自称,缺少独立承销。
- 产品栈实质依赖外部模型、平台生态,以及企业信任与支持交付能力。
- 创始人集中度高,公开治理界面滞后,增加关键人和沟通风险。
未决问题
- ARR 与分产品线收入结构仍未披露。
- 毛利率、服务成本和分 SKU 实际价格仍未公开可见。
- 客户集中度、GRR、NRR 和合同耐久性未披露。
- 当前现金余额、烧钱和 runway 仍未知。
- Series C 优先权条款、二级交易和其他 cap table 细节仍为私有信息。
目录
01公司概览
1.1 身份、法律足迹和当前定位
现在再把 You.com 描述成消费级答案引擎,已经不准确。最新官方页面把它定位为企业 AI 基础设施,帮助组织更快获得准确、新鲜、可行动的情报;法律文件则把运营实体写作 SuSea, Inc. d/b/a you.com。公开足迹也不只是单办公室创业公司的故事:法律通知地址和多条公司新闻 dateline 指向 Palo Alto,招聘网站则强调 San Francisco 和 New York 办公室。这个组合符合公司轨迹:2020 年以 AI 原生搜索挑战者起步,随后把前门转向 API、私有数据检索和企业工作流支持。商业模式现在明显是混合型。You.com 一边提供自助定价和公开开发者入口,一边营销定制企业搭建和专家实施支持。后续章节最稳妥的概览框架是:You.com 是一家私营 AI 搜索与研究基础设施公司,在美国有多地足迹,企业化姿态比最初消费搜索发布时更强。[CO001, CO002, CO003, CO004, CO005, CO023]
| 指标 | 数值 / 状态 | 日期 | 置信度 | 缺口 |
|---|---|---|---|---|
| 成立 | 2020 | 2020 | 高 | 本章来源未找到确切 incorporation 文件 |
| 法律实体 | SuSea, Inc. d/b/a you.com(运营实体) | 当前 | 高 | 公开章节来源未覆盖州级 incorporation 案卷 |
| 总部锚点 | Palo Alto 法律通知地址 | 当前 | 高 | 运营足迹还包括 San Francisco 和 New York 招聘办公室 |
| 当前定位 | 企业 AI 基础设施 / 企业 AI 解决方案 | 2025-2026 | 高 | 公开叙事比最初的消费者搜索故事更成熟 |
| 最新估值 | $1.5B Series C | 2025-09 | 高 | 估值佐证充分;cap table 细节仍未公开 |
| 公开追踪融资 | 已披露轮次约 ~$195M | 2021-2025 | 中 | 反复出现的 ~$222M 数字尚未完全调和 |
| 使用规模 | 声称每月 1B+ API 查询 | 2025-09 | 中 | 月查询量仍主要来自公司自述 |
| 当前员工数 | 未公开披露 | 2026-06 | 低 | 需要管理层、工资单或招聘系统证据核验 |
混合了独立佐证的融资事实与公司自述的规模和足迹披露;员工数仍未解决。
[CO001, CO002, CO003, CO004, CO016, CO017]创始人履历、产品范围、客户和融资,共同强化了公司向企业 AI 基础设施转型的叙事。
[CO004, CO005, CO011, CO018, CO022, CO023]概览级 KPI 显示融资和使用信号强劲,但承销数据仍缺失。
[CO016, CO017, CO019, CO026, CO034, CO039]1.2 创始人履历、领导层延续性和治理信号
创始人与市场的匹配,是 You.com 最清晰的优势之一。Richard Socher 仍是公司最主要的公开声音,来自 Salesforce 和学术 NLP 的履历带来少见可信度;Bryan McCann 则通过公司用于解释技术根源的 Natural Language Decathlon 工作,连到同一条研究脉络。这种履历让公司从搜索 UI 转向检索基础设施的转变更可信:创始团队过去做过语言建模、企业产品化和评估。不过,领导层延续性比旧版公司页面暗示的更复杂。公开 About 页面仍把 McCann 标为 CTO,但 2026 年 3 月公司公告已把运营 CTO 角色交给最早期员工之一 Saahil Jain,McCann 则以顾问和董事会身份继续参与。除此之外,治理披露仍然很薄。主要硬信息是 David Yang 随 Series C 加入董事会,说明董事会演进终于开始对外可见,但还远未充分透明。[CO006, CO007, CO008, CO009, CO010, CO011]
| 人物 | 角色 / 状态 | 背景 | 重要性 | 依赖 / 缺口 |
|---|---|---|---|---|
| Richard Socher | CEO 兼联合创始人 | 前 Salesforce 首席科学家、公开 NLP 研究者 | 锚定技术可信度和企业叙事 | 公开定位中关键人物集中度高 |
| Bryan McCann | 联合创始人;据 2026 年 3 月更新,不再日常担任 CTO | 前 Salesforce NLP 负责人、Socher 合作者 | 把公司连接到 prompt-engineering 和检索研究根基 | 公开 About 页面仍显示他为 CTO |
| Saahil Jain | 2026 年 3 月起任 CTO | 早期员工;创始工程师;前 AI 负责人 | 表明技术继任来自内部,而不是外部替换 | 更广工程团队仍未公开映射 |
| David Yang | 通过 Socium Ventures 任董事 | 与 Cox Enterprises 相关的成长投资人 | 增加与 Series C 融资绑定的正式董事证据 | 观察员权利和完整董事会构成仍未披露 |
| 公开领导层表面 | 创始人主导官方叙事 | About 和访谈页面围绕 Socher 和 McCann | 暗示故事和战略由创始人紧密把控 | 尽调室需要更完整组织架构和高管名单 |
公开材料对创始人履历很强,但对完整董事会、汇报线和继任细节仍偏弱。
[CO006, CO007, CO008, CO009, CO010, CO011]1.3 融资历史、估值演进和投资人组合
融资故事现在足够清楚,可以支撑报告主线,尽管一些反复出现的合计数字仍然模糊。TechCrunch 记录了 2021 年 $20 million 种子轮,随后是 Radical Ventures 领投的 2022 年 $25 million 轮;公司和媒体来源均指向 2024 年 $50 million Series B,以及 2025 年 Cox Enterprises 领投、估值 $1.5 billion 的 $100 million Series C。这四轮披露融资合计约 $195 million 可见资本,明显低于二级叙事中有时重复的约 $222 million 数字;除非缺失 tranche 浮出水面,后续章节应把后者视为未验证。投资人基础既有财务意义,也有战略意义:DuckDuckGo 同时是投资人和客户,Cox 通过 Socium 注入与董事会相关的成长资金,更早支持者包括 Benioff、Radical、Salesforce Ventures 和 Norwest。仍然缺失的是对控制权和承销判断更关键的 cap-table 细节,包括老股转让、观察员权利,以及任何债务或结构化资本。[CO013, CO014, CO015, CO016, CO017, CO018]
| 利益相关方 | 角色 | 公开关联轮次 / 关系 | 重要性 | 尽调要求 |
|---|---|---|---|---|
| Time Ventures | 种子轮领投方 | 2021 年 $20M 种子轮 | 来自 Marc Benioff 生态和创始人网络的早期信号 | 确认持股及任何持续治理权利 |
| Radical Ventures | 2022 年领投方 | 2022 年 $25M 轮 | 在后续企业转向完全显现前就支持业务 | 核验 pro rata 和董事会经济性 |
| Georgian | Series B 领投方 | 2024 年 $50M Series B | 验证从消费者搜索转向企业生产力 | 确认 Georgian 如何影响定价和企业 GTM |
| Cox Enterprises / Socium Ventures | Series C 生态领投赞助方 | 2025 年 $100M Series C;David Yang 加入董事会 | 在独角兽估值下增加战略分量和董事会相关资本 | 要求董事席位、观察员和信息权细节 |
| DuckDuckGo | 投资人和客户 | 参与 Series B;在 Series C 文章中被列为 API 客户 | 创造超越资本本身的战略分发和验证价值 | 明确收入集中度、排他性和数据访问条款 |
| Salesforce Ventures / Norwest | 既有后期投资人 | 在 Series C 参与名单中被点名 | 提供来自企业软件投资人的持续验证 | 需要确切持股、跟投权和任何二级交易 |
投资人地图混合了独立报道和公司发布的融资文章;不能替代已签署的 cap table。
[CO013, CO014, CO015, CO016, CO017, CO018]| 日期 | 事件 | 类型 | 金额 / 估值 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2020-12 | 作为 AI-native 搜索引擎公开发布 | 成立 | TechCrunch 报道发布 | Richard Socher、Bryan McCann | 确认最初的消费者搜索 thesis 和创始人组合 |
| 2021-11 | 种子融资完成 | 融资 | $20M 种子轮 | Time Ventures 和既有支持方 | 显示 Benioff 网络的早期信念 |
| 2022-07 | 增长轮与产品扩展 | 融资 | $25M;总额至 $45M | Radical Ventures 加既有投资人 | 为 YouCode 和 YouWrite 时期的产品扩展提供资金 |
| 2023-06 | 关于搜索挑战的独立报道 | 反向 | 只在复杂查询上与 Google 竞争 | TechCrunch 和 Socher | 标记最初消费者搜索战役的边界 |
| 2024-09 | Series B 与生产力引擎重新定位 | 融资 | $50M;总额至 $99M | Georgian、DuckDuckGo、Salesforce Ventures 等 | 让企业转向变得明确 |
| 2024-10 | 宣布 dpa 合作 | 合作 | 新闻发现合作已上线 | dpa 和 You.com | 显示媒体行业企业 traction |
| 2025-09 | Series C 独角兽轮 | 融资 | $100M,估值 $1.5B | Cox Enterprises、Socium、既有投资人 | 确立独角兽地位并扩充董事会 |
| 2026-03 | Saahil Jain 晋升 CTO | 治理 | Bryan McCann 转为顾问角色 | Saahil Jain、Bryan McCann、Richard Socher(管理层) | 表明内部继任,同时保留创始人影响力 |
这是本章唯一的记录时间线;后续章节应复用这些日期,而不是凭记忆重建。
[CO001, CO007, CO013, CO014, CO015, CO016]公开里程碑显示,公司已从消费者搜索清晰转向企业 AI 基础设施,同时治理也在成熟。
[CO001, CO007, CO013, CO014, CO015, CO016]1.4 规模标记、产品转向和概览层面的风险背景
最强的公开规模标记不是收入,而是用量。Series C 材料称 API 现在每月处理超过 10 亿次查询;此前 Series B 文章已经声称累计 10 亿次查询,并称自 2024 年 1 月以来 ARR 增长 500%。这些信号有意义,但大多仍来自公司自述,需要谨慎处理。独立报道确实确认了战略转向:TechCrunch 2024 年报道称,在简单查询上对抗 Google 已经徒劳,因此 You.com 转向更高端、更难的研究和生产力任务,在这些场景里引用和编排更重要。客户故事也支持这一叙事,因为它们集中在 NIH 研究工作流、Maryville 学生成果、Mimecast GTM 生产力、dpa 新闻编辑室发现,以及与 Harvey 和 DuckDuckGo 的具名关系。同一组证据也暴露了主要概览风险。分发仍不如产品质量有证明力,一些关键指标仍为自报,员工数未披露,公开治理页面也尚未完全跟上最新领导层变化。[CO019, CO020, CO021, CO022, CO025, CO026]
1.5 图表与附录
02市场分析
2.1 市场边界、纳入支出和替代方案
分析 You.com 时,不应假设它在争夺所有搜索、所有生成式 AI 或所有知识工作软件支出。证据更具体。AI 搜索由自然语言理解、跨多来源检索以及带引用的综合答案定义;企业搜索还叠加内部授权、索引和连接器要求,这些在消费级答案引擎里并不存在。这意味着,You.com 最相关的市场边界,位于外部网页 grounding、企业检索和面向代理的研究基础设施交汇处。纳入支出包括搜索调用、答案生成 API、抽取、RAG grounding,以及让开发者或企业团队基于新鲜数据搭建工作流的高价值研究端点。排除支出包括原始模型训练预算、广义消费搜索广告预算,以及并不真正提供搜索或 grounding 基础设施的软件类别。现实替代方案是传统关键词或联邦搜索、围绕基础模型 API 自建的 DIY RAG 栈,以及承诺为代理提供类似 grounding 的 API-first 搜索竞品。[CM001, CM002, CM003, CM004, CM005, CM006]
| 细分 / 类别 | 纳入支出 | 排除支出 | 买方 / 付款方 | 与 You.com 的相关性 |
|---|---|---|---|---|
| 外部网页 grounding API | 搜索调用、摘要、页面提取、引用、面向实时网页工作流的答案生成 | 消费者搜索广告、通用浏览流量变现 | 使用 usage-based API 预算的开发者平台团队或产品负责人 | 最贴合 You.com 公开 API 定位和基准材料 |
| 企业检索 / workplace search | 索引、连接器、权限感知检索、员工答案体验 | 没有搜索或 grounding 层的通用生产力套件 | CIO、知识平台负责人、AI enablement 负责人 | 与企业扩张相关,但不是 You.com 今天最清晰的公开楔子 |
| 垂直研究工作流 | 更高价值的领域端点、研究代理、合规或分析师工作流 | 缺乏搜索深度的宽泛横向生产力软件 | 用工作流软件预算付费的职能负责人或平台团队 | 相关,因为 You.com 销售高端研究型端点 |
| Agent 平台 grounding 层 | 位于聊天机器人、copilot 和自主代理下方的 API 和检索服务 | 没有检索或引用层的纯模型支出 | AI 工程或平台预算负责人 | 重要,因为 agents 需要带引用且新鲜的检索,而不只是模型 tokens |
| 消费者答案引擎 UX | 面向终端用户的自然语言搜索界面和答案呈现 | 企业连接器和内部权限栈 | 消费者产品预算或流量变现负责人 | 历史上与 You.com 有关,但对当前市场 sizing 来说太宽 |
| DIY 或替代栈 | 传统企业搜索、vector DB 加 LLM 栈,或竞争搜索 API | N/A | 与上述相同的买家 | 这些替代方案争夺同一笔 grounding 和知识工作预算 |
各行把更宽的 AI 和搜索背景,与更窄且看起来最贴近 You.com 的检索、grounding 和工作流支出分开。
[CM005, CM006, CM007, CM008, CM018, CM021]证据受限的金字塔,从广义企业 AI 支出收窄到 You.com 似乎参与竞争的搜索基础设施层。
这个金字塔叠放的是不可相加的市场视角:上层是广义 AI 预算或价值信号,底层是在更窄 AI 搜索基础设施赛道里的公司规模锚。
[CM009, CM010, CM011, CM016, CM027, CM031]2.2 受证据约束的规模测算和变现视角
公开数据不足以为 You.com 支撑一句干净的 TAM 表述,因此更稳妥的做法是使用多个受约束的视角。最宽一层,Andreessen Horowitz 报告称 2023 年企业生成式 AI 平均支出为 $7 million,并预期 2024 年该支出约增长 2 到 5 倍;同时,少于四分之一受访企业预期资金会继续来自创新预算。这一点重要,因为它意味着品类正在转向经常性软件归属,而不只是实验性试点。BVP 提供第二个视角,展示价值已经如何流经 AI 栈:基础模型公司在 2023 年融资 $23 billion,而垂直 AI 初创公司的年度合同价值已经达到传统垂直 SaaS 的约 80%,且增长约 400%。这些信号是需求背景,不是 You.com 的 TAM。变现方面,公开厂商定价比公开市场规模更清楚。基础搜索访问目前集中在一个窄区间——Brave 每 1,000 次答案请求约 $4,You.com 每 1,000 次网页搜索调用 $5,Exa 每 1,000 次搜索请求 $7——而更深的答案和研究工作流售价更高。这个价格带说明,You.com 的上行空间较少来自商品化检索,更多取决于能否说服买家为工作流级结果付费。[CM009, CM010, CM011, CM012, CM015, CM016]
| 发布方 / 视角 | 年份 | 地理范围 | 数值 | CAGR / 增长信号 | 方法 | 置信度 | 主要限制 |
|---|---|---|---|---|---|---|---|
| a16z 企业 genAI 支出视角 | 2024 年发布 / 2023 年基准 | 大型企业 | 2023 年平均支出 $7M | 2024 年计划增加 2x 到 5x | 围绕模型、托管和微调预算对企业领导者做的调查 | 中 | 面向广义生成式 AI 的预算视角,不是搜索特定收入 |
| a16z 软件预算迁移视角 | 2024 年发布 | 大型企业 | 来自创新预算的支出 <25% | 转向经常性软件科目 | 关于预算来源的调查证据 | 中 | 显示所有权和耐久性,不是品类规模 |
| BVP foundation-layer 资本视角 | 2024 年发布 / 2023 年数据 | 全球 | 融资 $23B;合计市值 $124B+ | 资本仍集中在基础层 | 投资人对模型公司融资和估值的综合 | 中 | 资本市场视角,不是直接搜索支出 |
| BVP application-layer 付费意愿视角 | 2024 年发布 | 全球垂直软件市场 | 传统 SaaS ACV 的 ~80%;增长 ~400% | 应用层快速扩张 | 投资人对 vertical-AI 公司 ACV 的分析 | 中 | 不特指搜索基础设施 |
| You.com 运营规模视角 | 2025 年发布 | 全球 API 使用 | 每月 >1B API 查询 | 互联网级使用量主张 | 公司关于当前月查询量的声明 | 中 | 使用量不是收入,细分组合未披露 |
| 开发者搜索 API 价格视角 | 2026 年观察价格 | 全球开发者市场 | $4 到 $7 / 1,000 次基础搜索或答案请求 | 各供应商基础价格聚得很紧 | Brave、You.com 和 Exa 的当前标价 | 中 | 标价不是实际净价或企业合同价值 |
这张表有意混合预算、资本、使用量和价格视角,因为公开证据没有暴露 You.com 独立市场规模的清晰口径。
[CM009, CM010, CM011, CM012, CM015, CM016]对 agentic-search 构建者最相关的 API-first 供应商中,基础网络搜索访问的当前标价区间。
这些区间比较的是公开发布的基础搜索或答案访问标价,不代表议价后的企业折扣或完整工作流成本。
[CM024, CM025, CM026, CM038]2.3 买家、用户和预算归属
这个品类的买家地图分成企业平台所有者和开发者主导团队两类。企业买家想要跨 SharePoint、Salesforce、Jira 和内部文档等系统做安全、权限感知的检索,因此 Glean、Azure AI Search 和 Google Agent Search 都强调连接器、治理和企业数据控制。在这些案例里,买家可能是 CIO、知识平台负责人、支持平台所有者或 AI enablement 团队;用户是员工、分析师、支持客服或下游内部 copilots;付款方通常是经常性软件或云预算。开发者主导买家不同。OpenAI、Exa、Tavily、Brave 和 You.com 都把搜索作为可编程组件,卖给聊天机器人、编码助手、研究代理和有网页依据的应用;这意味着第一笔采购可以从工程或产品预算开始,只有工作流证明价值后才扩张。对 You.com 来说,关键含义是它横跨两种动作。它的 API 定价和 benchmark 材料瞄准自下而上的 builders,但其规模声称、垂直索引和企业定位,又暗示后续会向更高价值工作流上移。[CM018, CM019, CM020, CM021, CM022, CM023]
| 细分 | 买方 | 用户 | 付款方 | 工作流 | 预算负责人 | 采用触发点 |
|---|---|---|---|---|---|---|
| 企业知识平台 | CIO、IT 平台或知识负责人 | 员工、分析师、支持人员 | 经常性软件或云预算 | 跨内部文档和业务系统搜索 | 知识或 AI enablement 负责人 | 需要跨多来源的安全、权限感知答案 |
| 客户支持 / 服务运营 | 支持平台或 CX 负责人 | 客服坐席和自助服务客户 | 运营软件预算 | 面向工单和知识库的有依据答案检索 | 支持运营负责人 | 分流工单,提高解决速度 |
| 开发者工具 / Agent 构建者 | 工程经理或产品负责人 | 开发者和自主 Agent | 产品或平台工程预算 | 把实时网页搜索或研究嵌入应用 | 工程或产品负责人 | 需要新鲜引用、低延迟和简单 API |
| 受监管或专家研究团队 | 职能运营或研究负责人 | 分析师、律师、财务或合规人员 | 部门工作流软件预算 | 带引用的垂直多源研究 | 业务单元负责人 | 需要可追溯、最新的外部情报 |
| 云原生企业构建者 | AI 平台团队 | 内部 Copilot、聊天机器人和应用 | 云与平台预算 | 在 RAG 应用中接入内部或网站数据 | 平台工程负责人 | 希望让模型输出扎根公司数据,又不必自建整套栈 |
| 独立开发者 / 初创公司 | 创始人或独立开发者 | 应用终端用户 | 按用量计费的 API 支出 | 快速原型并发布带搜索能力的产品 | 创始人或产品构建者 | 定价门槛低、赠送额度可用、集成快 |
关键分野在于企业平台归属权和自下而上的开发者采用;You.com 已显示出同时服务两种销售打法的证据。
[CM018, CM019, CM020, CM021, CM022, CM023]这张矩阵展示谁付款、谁使用,以及 You.com 在企业和开发者主导细分中最契合的位置。
[CM018, CM019, CM020, CM021, CM022, CM023]2.4 增长驱动、约束,以及 You.com 所处位置
最强需求驱动是结构性的。代理工作流越来越需要新鲜引用、页面级访问和语境丰富的输出,而不是蓝色链接;基础 RAG 研究和厂商文档都把 grounded search 视为解决幻觉和知识陈旧的答案,进一步强化了检索的技术逻辑。预算所有者也在寻找生产力和准确性收益,而不只是 AI 新鲜感,这解释了为什么企业支出正在流入经常性软件科目。约束同样重要。连接器、权限和治理仍是部署瓶颈;DIY RAG 仍需要实质系统和 ML 人才;买家也越来越想要多模型可选性,避免锁定单一供应商。与此同时,基础检索的价格带已经很紧,品类里挤满了 hyperscalers、企业工作 AI 平台和 API-first 厂商。这让 You.com 处在一个合理但并非牢不可破的位置:公开证据指向它在外部网页 grounding 和更高层研究工作流上的差异化切口,但公开证据仍无法隔离出清晰的 SAM、利润率画像,或足以在缺少客户级尽调时承销的持久护城河。[CM013, CM014, CM017, CM027, CM028, CM029]
| 驱动因素 / 约束 | 方向 | 时间 | 影响 | 尽调追问 |
|---|---|---|---|---|
| 企业 AI 预算转入经常性软件科目 | 驱动因素 | 当前 | 提高搜索和数据扎根支出的持续性 | 追问今天哪些买方用经常性软件或云预算采购 You.com |
| Agent 需要新鲜、有引用、到页面级的检索 | 驱动因素 | 当前 | 支撑超出模型 token 之外的搜索基础设施需求 | 追问用量中 Agent 工作流与简单搜索各占多少 |
| 多模型可选性 | 驱动因素 | 当前 / 正在显现 | 有利于模型无关或可组合检索供应商 | 追问 You.com 集成在不同模型栈之间的可迁移性 |
| 从搜索向答案 / 研究端点做工作流级别加售 | 驱动因素 | 正在显现 | 如果买方越过商品化调用,可打开更高 ARPU 路径 | 追问从基础搜索 API 转向高级端点的 cohort 转化 |
| 连接器和摄取复杂度 | 约束 | 当前 | 拖慢企业部署,也让预置集成供应商更占优 | 追问当前连接器路线图,以及不同客群的部署负担 |
| 权限、治理和隐私要求 | 约束 | 当前 | 可能挡住受监管行业或企业级推广 | 追问合规姿态、权限模型和默认数据留存设置 |
| 自建 RAG 实施复杂、人才稀缺 | 约束 | 当前 | 有助于打包型供应商胜出,但也拉长销售周期 | 追问 You.com 仍把哪些实施工作留给客户 |
| 基础搜索 API 价格压缩 | 约束 | 当前 | 把毛利和差异化压力推向更高层工作流 | 追问不同客户 cohort 的实际 ASP 和折扣行为 |
| 品类拥挤,超大云厂商和 API 优先对手并存 | 约束 | 当前 | 护城河取决于性能、分发或专有数据,而不只是品类新鲜感 | 追问相对 Google、Azure、Exa、Brave 和 Glean 的赢单 / 输单数据 |
这个品类的需求面看起来有吸引力,但承保仍取决于 You.com 能否在商品化检索之上、完整企业工作平台之下做出差异化。
[CM011, CM027, CM028, CM032, CM033, CM034]示意性采用漏斗,展示广义 AI 指令如何收窄为耐久的搜索基础设施部署。
指数值是示意性阶段权重,总结公开证据显示的品类瓶颈;它们不是公司披露的转化率。
[CM027, CM028, CM033, CM034, CM035, CM037]2.5 图表与附录
03竞争对手
3.1 竞争格局,以及最重要的竞品类别
不应只拿 You.com 对标一个答案引擎或一个企业搜索 incumbents。公开产品界面显示出三类不同竞争者。第一类是直接的 API-first 同行——Exa、Tavily、Brave、Perplexity 和 OpenAI 的网页搜索栈——它们向代理和应用销售可编程检索、答案生成或研究工作流。这些厂商攻击的买家问题与 You.com 的自助式搜索和研究 API 相同:如何在不从零搭建检索层的情况下,给 LLM 提供当前、带引用、有网页依据的上下文。第二类是企业 incumbents 和近似 incumbents——Glean、Azure AI Search、AWS Kendra、Google Agent Search 和 Elastic——它们较少在简单每次调用价格上竞争,更多争夺私有数据、权限、索引、治理和采购关系的控制权。第三类是看起来完全不像搜索厂商、却能捕获同一预算的替代品,包括专业服务工作流里的 Harvey、拥有消费分发规模的 DuckDuckGo,以及由 OpenAI 的 gpt-oss 加托管搜索工具促成的内部自建栈。这个框架重要,因为 You.com 已不再把自己呈现为泛消费搜索挑战者;TechCrunch 2024 年报道和 You.com 自己的企业界面都显示,它正转向更深的研究和生产力代理。因此,承销问题不是 You.com 有没有对手——显然有——而是在某个具体交易中,哪一类压力最大:开发者采用、企业扩张,还是内部自建决策。[CP004, CP005, CP010, CP012, CP016, CP017]
| 竞争对手 / 类别 | 品类 | 规模或采用信号 | 目标客群 | 差异化 | 局限 |
|---|---|---|---|---|---|
| You.com | 网页扎根与研究平台 | 公司称月 API 查询量超过 1B;自助服务和企业入口均已上线 | 构建 Agent 的开发者,以及需要带引用研究工作流的企业 | 组合网页搜索、深度研究、金融研究、模型无关编排和私有数据连接器 | 核心规模和基准测试主张仍主要来自公司自身,实际企业定价未披露 |
| Perplexity | 答案引擎与搜索 API 同业 | 官方文档披露原始搜索、Sonar 和 Agent 工具定价 | 需要带引用答案或答案引擎能力的开发者和团队 | 在带引用答案 UX 和 API 变现上与 You.com 高度重叠 | 公开文档对定价的呈现,比对企业私有数据工作流深度的呈现更清晰 |
| Exa | API 优先的搜索与研究同业 | 公开端点覆盖搜索、内容、答案、研究和 Agent 工作流 | 把实时网页检索或研究嵌入应用的开发者 | 开发者定位清晰,端点价格和研究工作流明确 | 公开证据最强的是 API 界面和定价,而不是企业分发 |
| Tavily | Agent 搜索工具包同业 | 搜索、提取、爬取、映射和研究 API 按 credits 销售 | 构建懂网页的 Agent 和研究工作流的团队 | 网页工具工作流覆盖面广,credit 模型透明 | credit 定价清楚,但企业安装基数和差异化分发不公开 |
| Brave Search API | 独立索引搜索同业 | 声称索引 30B+ 页面,提供答案 API、AWS Marketplace 上架和企业隐私姿态 | 希望直接访问独立网页索引的开发者和企业 | 自有索引、零数据留存叙事和强性价比故事 | 产品界面在搜索上更强,在完整企业私有数据编排上较弱 |
| Glean | 企业工作 AI 既有玩家 | 100+ 连接器、嵌入式工作流界面和 Enterprise Flex 席位打包 | 在内部系统中标准化 AI 的知识工作者和企业 | 靠企业数据、权限和嵌入式应用体验分发 | 公开定价偏打包口径,实际合同经济性不透明 |
| Google Agent Search | 超大云厂商企业检索既有玩家 | Search Enterprise 定价、网站 / 非结构化 / 结构化搜索,以及基于企业数据的扎根 | 已在 Google Cloud 上构建,或需要网站加内部数据搜索的企业 | Google 质量检索、引用、行业版本和云分发 | 企业数据控制更强,低摩擦自助式网页搜索采用更弱 |
| Azure AI Search | 超大云厂商企业检索既有玩家 | 混合搜索、向量存储行为、通过 Microsoft 工具获得的 80+ 连接器路径,以及分层定价模式 | 在 Microsoft 云、数据和身份栈上标准化的企业 | 深度接入 Microsoft 生态,并主打 Agentic retrieval 定位 | 公开页面更强调分层和企业部署,而不是简单的按调用价格对比 |
| AWS Kendra | 托管式企业搜索既有玩家 | GenAI 和企业索引版本,分别收取查询、存储和连接器费用 | 已在 AWS 内部的内部知识搜索和 RAG 构建者 | 成熟的基础设施式打包,连接器经济性明确 | 相比 You.com 或 Brave,不太明显为低摩擦外部网页扎根用例优化 |
| 用 gpt-oss 加工具自建 | 自建替代方案 | 开放权重模型可在本地或自有基础设施上运行,并可接入网页搜索工具 | 由工程牵头、优化控制权、定制化或成本结构的团队 | 避免供应商锁定,并可按具体工作流调优 | 需要集成工作、评估纪律和持续运营;产品供应商把这些抽象掉了 |
| Harvey | 垂直工作流替代方案 | 公司网站声称获得顶级律所和企业内部法务团队信任,覆盖 20+ 专业领域、142,000+ 用户和 1,500+ 家律所 | 法律和专业服务团队 | 掌握高价值专家工作流,而不只是检索层 | 领域范围比横向搜索或研究平台更窄 |
这张表有意偏向买方可以替代 You.com 的决策选项,而不是列出所有邻近 AI 搜索的公司。
[CP001, CP002, CP005, CP006, CP007, CP008]按与 You.com 当前论点的直接重合度,以及企业分发或数据控制力,对竞争对手类型做序数映射。
坐标轴为 1–10 的序数评分,来自对产品重叠和分发力证据的审阅,不是实测市场份额或收入。
[CP005, CP012, CP016, CP018, CP020, CP024]3.2 直接同行在外部 grounding 上竞争,incumbents 则争夺数据控制
在最直接的同行中,重叠最清楚的是向开发者开放搜索、答案或研究端点的厂商。Exa 提供 search、contents、answer 和 research 端点;Tavily 提供 search、extract、crawl、map 和 research;Brave 基于自有网页索引提供 raw search 和 answer API;Perplexity 官方文档把 raw search 与 Sonar、agent tooling 组合起来。OpenAI 的 web search 也应放入这组替代组件,尽管它被包装成 Responses API 里的内置工具,而不是专门的搜索产品界面。这些直接同行会立刻对 You.com 的外部网页产品形成价格和功能压力。incumbent 一侧不同。Glean 的姿态围绕权限感知企业检索、连接器广度、嵌入式工作流界面和按用户计费的 Enterprise Flex 包装展开。Azure AI Search、AWS Kendra 和 Google Agent Search 同样强调混合检索、知识库管理和企业数据访问,定价模型围绕容量、订阅或基础设施式计量,而不是简单的自助式网页搜索调用。Elastic 又扩展了 incumbent 集合,提供企业搜索和向量搜索 building blocks,而不是 turnkey 答案引擎。结果是一场双线战。You.com 既要在底层 web-grounding 同行面前赢得开发者偏好,也要证明自己能足够强地承载企业上下文、治理和工作流深度,避免被 Glean 或 hyperscaler-native 栈替代。[CP005, CP006, CP007, CP008, CP009, CP010]
| 公司 / 类别 | 外部网页扎根 | 私有数据连接器 | 研究或 Agent 工作流层 | 定价透明度 | 信任 / 治理界面 | 分发优势 | 关键未知 |
|---|---|---|---|---|---|---|---|
| You.com | 强 | 中高 | 强 | 标价透明度强 | 强 | 中 | 赢单 / 输单和实际企业合同细节 |
| Perplexity | 强 | 低到中 | 中高 | 强 | 中 | 中 | 企业私有数据和治理部署深度 |
| Exa | 强 | 低 | 强 | 强 | 中 | 中 | 企业分发和工作流所有权 |
| Tavily | 强 | 低 | 中高 | 强 | Unknown | 低到中 | 安装基数和治理深度 |
| Brave Search API | 强 | 低 | 中 | 强 | 强 | 中 | 搜索和答案层之外的覆盖宽度 |
| Glean | 低到中 | 强 | 强 | 中 | 强 | 强 | 网页扎根广度在其赢单 / 输单组合中有多重要 |
| Google / Azure / AWS 既有栈 | 中 | 强 | 中高 | 中 | 强 | 强 | 开发者牵头的买方是否会为了云原生接受更高复杂度,而不是选择产品简单性 |
| 自建 | 不定 | 不定 | 不定 | 不定 | 不定 | 初期低 | 总运营成本和价值兑现时间 |
单元格总结已审阅来源直接支持的内容;不定或未知意味着公开证据高度取决于实施选择,或没有给出可清晰横向对比的证据点。
[CP003, CP006, CP007, CP008, CP012, CP013]比较企业或开发者可能用来替代 You.com 的主要竞争类别的能力。
单元格把异质来源证据压缩成买方相关标签;「可变」表示能力取决于实施选择,而不是某个标准厂商界面。
[CP003, CP006, CP007, CP008, CP012, CP016]3.3 定价、包装和分发权力解释了为什么市场显得拥挤
最清楚的公开定价信号是,基础外部 grounding 已经聚集到一个窄区间。You.com 标价为每 1,000 次网页搜索调用 $5,Brave 的 answered search 为每 1,000 次请求 $4,Exa 的 search 为每 1,000 次请求 $7,Perplexity 文档列出 raw search 每 1,000 次请求 $5。Tavily 使用 credits 而不是 requests,但它的定价仍可换算成透明的 pay-as-you-go 费率。这个价格聚集很重要,因为它说明原始检索正在变成 utility,而不是持久护城河。因此,厂商都在向上游栈推进。You.com 以明显更高价格销售 research endpoints,Exa 把 deep-search 和 agent workflows 定价在基础搜索之上,Perplexity 增加 Sonar 和 agent/tool 定价层。Glean 把包装迁移推得更远,把企业搜索、chat、agent runs 和 embedded experiences 捆进按席位的 Enterprise Flex 权益与超额 credits 中。Hyperscalers 追求另一种模型:Google 和 AWS 给出明确的基础设施式查询和存储定价,而 Azure 当前定价页围绕 tier 选择、search units 和 agentic-retrieval activity 组织,而不是像消费搜索账单那样计费。分发会放大这些差异。Glean 嵌入 Slack、Teams、GitHub、Zoom 和服务平台;Google、Microsoft 和 AWS 紧贴更广的云和身份资产;DuckDuckGo 拥有终端用户搜索分发;Brave 通过 AWS Marketplace 获得可见渠道界面。相比之下,You.com 的公开分发优势更偏产品驱动,而非生态锁定。只要性能和工作流深度能持续吸引买家,这条路可以成立;但价格压力和 incumbent 捆绑应被视为基本假设,而不是被轻易忽略。[CP001, CP002, CP006, CP007, CP008, CP009]
| 公司 / 类别 | 公开价格或单位 | 合同模式 | 包含能力 | 折扣 / 未知项 | 影响 |
|---|---|---|---|---|---|
| You.com | 每 1k 次 web-search 调用 $5;每 1k 次 research 请求 $12;每 1k 次 finance-research 请求 $110 | 按用量计费的自助服务加企业销售动议 | 外部网页搜索、研究工作流、金融研究、引用和企业选项 | 实际企业折扣和席位经济性不公开 | 自助入口强,但毛利防守可能取决于更高价值工作流 |
| Exa | 每 1k 次 search $7;每 1k 次 deep search $12;每 1k 次 deep-reasoning search $15;Agent 请求 $0.025 起 | 按用量计费的 API 定价 | 搜索、内容、答案、研究和异步 Agent | 企业折扣和自定义数据集定制定价 | 纯 API 经济性上,可与 You.com 清晰做开发者侧对比 |
| Tavily | 每月 1,000 个免费 credits;按量付费每 credit $0.008;高级搜索消耗 2 credits | 基于 credit 的用量模型,提供月度套餐和企业定制定价 | 搜索、提取、映射、爬取和研究工作流 | credits 到请求经济性的换算取决于端点和深度 | 适合实验和 Agent 工具打包,虽不如按调用定价直观 |
| Brave Search API | 每 1k 次答案请求 $4,另收 token 费用;搜索套餐含每月免费 credits | 按用量计费 API,另有定制企业套餐 | 原始搜索、答案、引用和独立索引访问 | 存储权利和大规模条款需要升级套餐 | 在这组来源中,可见直接答案价格最低,加大基础检索的商品化价格压力 |
| Perplexity | 每 1k 次搜索请求 $5;每次 web_search 工具调用 $0.005;Sonar 分层另收 token 和请求费用 | 跨搜索、Agent 和模型层的 API 用量模型 | 原始搜索、答案引擎模型、深度研究和 Agent 工具 | 企业席位定价存在,但本章证据集中在文档定价 | 在外部答案和研究变现上与 You.com 高度重叠 |
| OpenAI | 每 1k 次 web-search 调用 $10,另加模型 token 定价 | 模型加工具定价,而不是独立搜索产品定价 | Responses API 内的托管网页搜索 | 总成本取决于模型、工具频率和 token 量 | 作为构建者栈组件有吸引力,但本身不是开箱即用的企业搜索平台 |
| Glean | 按用户计费的 Enterprise Flex 席位加共享 FlexCredits;标准 thinking 包含每用户每周最多 100 次查询 | 席位许可,高级 AI 功能按用量超额收费 | 企业搜索、聊天、Agent、嵌入式应用体验和 API | 这组来源中的官方文档没有发布简单公开席位费率 | 打包方式更偏企业推广和嵌入式使用,而不是自下而上的 API 采用 |
| Google Agent Search | 每 1k 次标准查询 $1.50;每 1k 次企业查询 $4.00;存储和附加项另计费 | 查询加存储定价,并提供可配置订阅选项 | 覆盖网站、结构化和非结构化数据的搜索应用,并带生成式答案 | 高级附加项和可配置模型带来比简单 API 同业更多的定价分支 | 买方已接受云式计量且需要企业扎根时具备竞争力 |
| AWS Kendra | 基础 GenAI index 每小时 $0.32 起,另收查询、存储和连接器费用 | 预置基础设施定价 | 内部搜索索引、retrieve/query API 和托管连接器 | 即便尚未重度使用,成本也会随预置容量扩张 | 更适合作为内部知识 RAG 成本基准,而不是直接外部搜索价格对比 |
| Azure AI Search | 专用和 serverless 分层,agentic retrieval 在 Azure AI Search 下计费;页面展示分层结构,而不是简单标价 | 基于容量的搜索服务定价 | 混合检索、索引和 agentic-retrieval 操作 | 实际按地区价格需要计算器或报价上下文 | 不利于开发者快速比较,但企业云买方熟悉这种模式 |
该领域的公开定价高度分散;有的供应商按调用收费,有的按 credit 收费,有的按席位收费,也有的按云容量收费,因此定价背后的影响比标价本身更重要。
[CP001, CP002, CP006, CP007, CP008, CP009]3.4 护城河耐久性取决于拥有工作流,而不只是拥有一次搜索调用
You.com 最强的公开论据,不是它没有竞争,而是它试图逃离最商品化的一层。企业页面、ARI 发布和 Finance Research API 都显示,公司正从网页搜索推向更高价值、带引用、多步骤的研究工作流,并把公开和私有数据结合起来。这个策略合理,因为纯搜索层看起来脆弱。公司自述的 benchmark 可以解释 You.com 希望外界如何看它,但单靠 benchmark 不能证明相对直接同行有持久优势。与此同时,incumbent 压力是结构性的:Glean 和 hyperscalers 更靠近身份、数据和采购层,而这些层决定企业搜索能否粘住。替代威胁同样是结构性的。Harvey 说明,垂直工作流工具可以吸收本来可能流向横向研究平台的预算;gpt-oss 和工具 API 也降低了企业内部搭建定制代理的成本。因此,公开证据支持一个更细的护城河判断。买家想要一个平台,把当前网页 grounding、引用、模型选择和可配置企业工作流混在一起时,You.com 看起来最强;如果买家已经有可信内部数据基础设施、垂直应用完全占住用户工作流,或工程团队愿意自己组栈,You.com 就弱一些。主要尽调障碍不是不知道有竞争者,而是缺少公开的 win-loss、实际定价和渠道经济数据,无法判断 You.com 的切口是否真正转化为持久企业权力。[CP003, CP024, CP025, CP027, CP028, CP029]
| 护城河主张 | 威胁 | 严重性 | 有证据支撑的理由 | 缓释方式 / 尽调索取项 |
|---|---|---|---|---|
| 引用式网页 grounding 质量 | 直接 API 同行之间价格压缩 | 高 | 多个同行的基础外部搜索定价已大致集中在每 1,000 次请求 $4-$7。 | 索取按端点拆分的留存和毛利率证据,尤其比较研究产品与普通搜索 |
| 企业工作流深度 | Glean 与 hyperscaler 控制内部数据、身份和采购 | 高 | 与 You.com 的公开呈现相比,Glean 和云端既有厂商更贴近连接器、权限和企业落地界面。 | 按用例和买方画像,提供与 Glean、Microsoft、Google、AWS 竞争的赢单 / 输单数据 |
| 模型无关灵活性 | 用开放模型和托管搜索工具自建内部方案 | 中高 | 工程团队想掌握控制权时,gpt-oss 加网页搜索工具降低了拼装定制技术栈的门槛。 | 与内部自建替代方案相比的价值实现时间、维护负担和总成本基准 |
| 研究型 agent 差异化 | Harvey 等垂直工作流产品 | 中 | 专用工具可以占住高价值工作流,降低对横向研究平台的需求。 | 证明 You.com 已扩展到特定领域、留存稳住,且客户愿意在试验之外把 You.com 标准化使用 |
| 信任与治理姿态 | 竞争对手的信任叙事趋同 | 中 | You.com、Glean 和 Brave 都强调安全或零数据留存式主张,削弱了信息层面的差异化。 | 第三方安全尽调、红队结果,以及因合规赢单的客户证明 |
| 公司自编基准领先 | 未被独立复现的基准叙事 | 中 | You.com 最强的相对性能主张来自自家基准和发布材料,而不是中立第三方测试。 | 独立 bake-off 结果、具名客户证明和可复现评估方法 |
严重性反映截至 2026-06-14 公开证据集中的承销相关性,并非实测市场份额。
[CP024, CP027, CP031, CP032, CP033, CP034]用简表展示公开证据显示的 You.com 当前最强和最弱竞争位置。
[CP002, CP024, CP028, CP031, CP033, CP035]3.5 图表与附录
04财务
4.1 收入模式明显是混合型,不是单 SKU API 故事
当前公开足迹支持的是混合变现模式,而不是纯搜索 API utility。最清晰的可见收入流是自助开发者用量:You.com 定价页列出按次调用的网页搜索定价、按页计费的 contents 定价、成本较低的 research tier,以及价格高得多的 finance research tier。但公司卖的并不只是一个计量端点。Series B 和 Series C 材料、enterprise-solutions 页面以及 2025-2026 产品更新显示,至少有四条重叠变现路径:基础检索 API、高级研究 API、Team 或协作式产品,以及绑定私有数据集成的企业定制代理部署。这一点重要,因为它提高了收入可选性,但也让收入确认分析更复杂。用量计费、按席位导向的协作计划和定制企业项目,很可能按不同节奏确认收入,并带来不同毛利率。公开来源让我们看见菜单,却看不见结构。因此,最稳妥的财务框架是:You.com 已经大幅拓宽可销售界面,但投资人仍无法判断收入中有多少来自商品化搜索,又有多少来自更高价值的工作流产品或服务。[CI001, CI002, CI003, CI004, CI005, CI009]
| 收入流 | 公开机制 | 当前标价 / 单位信号 | 收入质量判断 | 证据状态 | 尽调索取项 |
|---|---|---|---|---|---|
| Web Search API | 面向开发者和 agents 销售、按使用量计费的自助式 API | $5 per 1,000 calls | 量大,但基础层可能承受价格压缩 | 官方定价页可直接观察 | 索取按查询区间拆分的实际净价、客户结构和毛利率 |
| Contents API | 用于页面内容的按量抽取或 livecrawl 附加产品 | $1 per 1,000 pages | 有用的附加产品,但客单价可能低于搜索或研究 | 官方定价页可直接观察 | 索取从搜索到 contents 的附加率及相关基础设施成本 |
| Research API | 面向多步骤开放网页研究的高价引用答案工作流 | 公开页面文字为每 1,000 次查询 $12 | 若采用真实,变现好于基础搜索 | 公开标价可见;实际用量未知 | 索取按研究层级和客户 cohort 拆分的量级结构 |
| Finance Research API | 使用授权数据和公开数据的金融专用 agentic research | $110 / 1,000 次查询 | 可见标价最高,但服务成本可能也最高 | 公开发布和定价已披露 | 索取扣除授权数据费、计算和支持后的毛利率 |
| 团队 / 协作产品 | 2024 年推出的席位式或工作区导向工作流产品 | 公开曾释放信号,但当前官方席位定价未在 API 定价页清晰呈现 | 若规模可观,可提升收入耐久性,但当前可见度弱 | 证据通过公司和媒体来源间接获得 | 索取当前席位定价、活跃席位数和按套餐拆分的留存 |
| 企业定制 agents 与私有数据部署 | 定制构建、私有 RAG、集成和定制工作流交付 | 定制 / 基于合同 | 支撑更大的 ACV,但会引入服务和部署复杂度 | 官方叙事很强;无公开合同经济性 | 索取服务收入占比、实施工作量,以及按账户拆分的投产时间 |
将可直接观察的 API 标价,与更高接触度或间接可观察的变现路径分开;实际净价和收入结构仍未披露。
[CI001, CI002, CI003, CI004, CI005, CI009]| SKU / 视角 | 当前公开价格 | 单位 | 标价与实际经济性 | 竞争背景 | 来源依据 |
|---|---|---|---|---|---|
| You.com Web Search API(网页搜索) | 5 | USD / 1,000 次调用 | 仅为标价;明确提供量级折扣和年度节省 | 位于可见同行价格带中部附近 | 官方定价页 |
| You.com Contents API(内容抓取) | 1 | USD / 1,000 页 | 仅为标价;附加行为和利润率未知 | 可与 Exa 页面抽取定价比较 | 官方定价页 |
| You.com Research API(研究) | 12 | USD / 1,000 次查询 | 公开页面文字显示其高于基础搜索,但不是最终合同价 | 相比简单搜索有溢价,低于金融专用研究 | 官方定价页 |
| You.com Finance Research API(金融研究) | 110 | USD / 1,000 次查询 | 高标价可能反映授权数据和深度研究成本 | 明显高于基础搜索,因为任务不同 | 官方定价页和发布文章 |
| 报道中的高端聊天 / 助手套餐 | 15 | USD / 月 | 通过 2025 年 6 月独立报道观察到,当前以 API 为中心的官方定价页并未清晰呈现 | 低于 Verdict 引用的 $20 月费消费者 AI 基准 | Verdict 文章 |
| 基础搜索同行价格带 | 4 / 5 / 7 / 10 | USD / 1,000 次请求或调用 | 同行标价适合压力测试价格压缩,不适合推断利润率 | Brave 为 4,You.com 为 5,Exa 为 7,OpenAI 网页搜索工具调用为 10 | 官方竞品定价页 |
将按使用量计费的 API 标价,与报道中的月费助手定价、以及仅作为市场压力 proxy 的竞品标价分开。
[CI002, CI003, CI004, CI005, CI006, CI031]展示 You.com 的混合变现模型里,产品触达如何转化为使用量、高级工作流和企业收入路径。
这是根据产品和企业材料绘制的结构性收入地图,不是公司披露的会计瀑布或客户转化漏斗。
[CI001, CI007, CI008, CI009, CI010, CI011]4.2 标价和公开 traction 是真实信号,但不是实际经济性
标价栈现在已经足够可观察,可以对收入质量说些有用的话。基础网页搜索定价为每 1,000 次调用 $5,内容抽取为每 1,000 页 $1,更高价值的研究产品价格大幅上台阶;公开页面文本显示 research tier 为 $12,finance-research SKU 为 $110。这个价差方向上是好事,因为它说明公司试图变现推理深度和工作流价值,而不只是廉价检索。限制同样重要:同一价格页宣传批量折扣和年度节省,所以标价是天花板,不是实际净费率披露。公开 traction 指标也大多来自公司自述。Series C 文章称每月 API 查询超过 10 亿次,前一年构建了 100,000+ 个代理;Series B 文章称自 2024 年 1 月以来 ARR 增长 500%。这些是有意义的需求信号,但没有显示绝对 ARR、cohort 质量或续约行为。NIH、Maryville 和 Mimecast 的客户故事展示了强工作流价值,却没有披露合同规模、部署范围或扩张行为。从财务角度,证据支持产品市场相关性和变现宽度,但还不足以证明持久经常性收入质量。[CI005, CI006, CI019, CI020, CI021, CI022]
在有边界的地方展示有来源支撑的公开区间,同时把未披露的基本面明确排除在外。
Research SKU 区间混合了不同产品,而非等价层级;估值区间把 2025 年 6 月融资讨论和之后完成的 Series C 合在一起。SPV 区间代表两份独立申报金额,不是单一融资工具。
[CI005, CI029, CI031, CI032, CI033, CI036]4.3 GTM 和单位经济代理信号指向双速动作
公开 go-to-market 证据显示一个双速商业模型。一边,You.com 已经搭起摩擦较低的开发者漏斗:$100 免费 credits、无需注册的免费 MCP 搜索 profile、SDKs、playground,以及接入 OpenAI、Claude、Zapier、Vercel 和 Databricks 生态。这些功能应能降低 API 买家的自助获客成本,并缩短首次价值实现时间。另一边,企业材料和创始人文章清楚暗示,私有数据集成、定制代理和高级研究工作流需要更重服务。这很可能意味着更长销售周期、解决方案工程,以及更昂贵的客户成功,尤其当公司承诺零数据留存、垂直索引、金融专用证据裁决和组织专属部署时。竞争性定价强化了这种结构迁移需求:You.com、Brave 和 Exa 的基础网页搜索定价已经紧密聚集,而 OpenAI 的相邻搜索工具位于该价格带之上,但包装方式不同。实际看,You.com 可能需要用 premium research、finance 和 enterprise workflow 产品,来补贴正在变得相当压缩的基础搜索层。公开证据没有说明的是,这些 premium 产品能否足够高效地转化,以抵消支持、推理和授权数据成本。[CI007, CI008, CI012, CI013, CI014, CI015]
| 指标 | 公开数值 / 状态 | 置信度 | 重要性 | 替代指标或限制 | 尽调索取项 |
|---|---|---|---|---|---|
| ARR 增长 | 自 2024 年 1 月以来 500%(公司声称) | 中 | 若属实,说明增长方向在加速 | 未披露绝对 ARR 基数,也无独立佐证 | 索取月度 ARR 桥表和董事会 KPI 历史 |
| 绝对 ARR 或收入 | 未公开披露 | 低 | 估值和 runway 承销的核心输入 | 没有变现结构,无法把使用量主张换算为收入 | 索取过去 12 个月收入、ARR,以及按产品拆分的月度 run rate |
| 月度 API 调用量 | 每月 >1 billion 次查询(公司声称) | 中 | 展示漏斗顶端规模和基础设施负载 | 可能包含低价或合作伙伴流量;单独看不是收入 proxy | 索取付费查询结构、免费查询结构,以及按查询区间拆分的毛利 |
| 按产品拆分的毛利率 | 未公开披露 | 低 | 决定高价产品能否抵消基础搜索价格压缩 | 授权数据和深度研究可能让不同 SKU 的成本栈差异很大 | 索取 search、contents、research、finance 和 services 的毛利率 |
| CAC / 回本周期 | 未公开披露 | 低 | 需要区分高效 PLG 和昂贵企业销售 | 免费 credits、文档、SDK 和企业集成只能提供方向性线索 | 索取获客漏斗、按细分市场拆分的回本周期和销售周期长度 |
| 净收入留存 | 未公开披露 | 低 | 扩张与流失,是经常性软件收入质量的核心信号 | 客户故事无法显示 cohort 扩张 | 索取按产品线拆分的 NRR、GRR 和 cohort 扩张 |
| 客户集中度 | 未公开披露 | 低 | 少数大型 API 合作伙伴可能主导使用量或收入 | 具名 logo 未披露支出占比或合同价值 | 索取前 10 大客户收入占比和最大合作伙伴经济性 |
| 每次深度研究或金融查询成本 | 未公开披露 | 低 | 只有多步骤检索和授权数据之后贡献毛利仍站得住,高端标价才有效 | 产品称会引用数百个来源和外部数据集 | 索取按层级拆分的每次查询计算、授权和支持成本 |
用已披露 traction 和定价作为 proxy,但所有核心效率指标仍需要管理层直接披露或提供 data-room 摘录。
[CI005, CI018, CI019, CI020, CI021, CI040]展示从获客到利润率的可能经济链条,并标出公开数据消失的位置。
这条经济链没有任何公开数值;图中只映射方向性逻辑。
[CI007, CI008, CI014, CI015, CI018, CI038]4.4 融资能力可见;现金是否充足不可见
资本形成看起来明显好于经营披露。既往轮次形成的已披露融资时间线是连贯的:2021 年 $20 million、2022 年 $25 million、2024 年 $50 million,以及 2025 年 $100 million Series C,合计约 $195 million 可见公司层面资本。独立新闻和官方材料均指向估值 $1.5 billion 的 $100 million Series C;2025 年 6 月报道则显示,公司在后续 close 前仍在寻求新融资。更有意思的增量证据来自 SEC filings:两个名为 GSBackers You.com Fund I 和 II 的 2025 年 8 月 Form D 载体,报告已全部售出的发行金额合计约 $15.1 million。它们看起来像与融资相关的 pooled investment vehicles,但并不能直接证明运营公司资产负债表上的现金。这个区别重要,因为公开材料仍未披露账面现金、月 burn、runway、债务或净留存。因此,财务结论是有细微差别的。作为一家软件基础设施公司,You.com 看起来不像缺资本,但授权金融数据、多步骤研究计算、企业集成和客户成功较重的部署,仍可能快速消耗现金。在管理层开放利润率和 runway 数据室之前,资本充足性只能判断为合理,而不是已证明。[CI026, CI027, CI028, CI029, CI030, CI031]
| 项目 | 公开信号 | 来源支撑的细节 | 承销含义 | 置信度 | 尽调索取项 |
|---|---|---|---|---|---|
| 可见已披露公司轮次 | ~195 | 20 / 25 / 50 / 100 已披露轮次合计,数百万美元 | 足以支撑严肃的基础设施建设,但不是现金余额披露 | 高 | 提供当前 cap table、现金余额,以及累计 primary 与 secondary 融资款 |
| 最新估值 | 1.5 | 2025 年 9 月 Series C 中的数十亿美元 | 支撑高端市场定位,但本身不能证明经营质量 | 高 | 提供展示估值理由、增长假设和投资人权利的董事会材料 |
| 2025 年中资本需求信号 | 在后续交割前,曾寻求按 1.4B 估值融资 | 2025 年 6 月独立报道称资金将加强 AI 助手产品 | 暗示 Series C 交割前仍依赖资本 | 中 | 提供融资流程时间线、投资人 pipeline 和资金用途计划 |
| 以 You.com 命名的 2025 Form D 载体 | 15.085 | Fund I 和 Fund II pooled venture-capital filings 合计的 USD millions | 显示融资生态深度,但不能直接证明运营公司现金 | 高 | 澄清这些载体是否进入 Series C,以及其经济条款是否不同于领投轮 |
| 手头现金 | 未披露 | 已审阅材料中未出现公开现金余额 | 无法承销 runway | 低 | 提供最近月末现金、受限现金和短期投资 |
| 月度 burn / runway | 未披露 | 未出现公开经营亏损或现金消耗披露 | 资本充足性仍只是看似可行,尚未被证明 | 低 | 提供 24 个月现金流、基础 / 下行情景 runway 和人员计划 |
| 下一轮触发条件 | 未公开 | 未披露明确触发条件;高端工作流转化和利润率证明似乎最可能成为关口 | 未来融资需求可能取决于高端 SKU 变现是否快过基础设施成本增长 | 中 | 提供融资计划、covenant 触发条件和董事会新增资本门槛 |
| 债务或项目融资义务 | 公开不可见 | 已审阅来源中未出现债务融资或项目融资义务 | 若属实是利好,但没有公开证据不等于零杠杆 | 低 | 提供债务明细、授信额度、信用证和表外承诺 |
将可见股权证据与未经证实的现金头寸假设分开;SPV filings 有信息价值,但不能与运营公司现金互换。
[CI030, CI031, CI032, CI033, CI034, CI043]| 缺失指标 | 重要性 | 当前公开状态 | 精确尽调路径 |
|---|---|---|---|
| 按 SKU 拆分的收入与确认政策 | 需要判断收入结构质量、季节性,以及服务是否掩盖软件增长 | 公开页面展示产品,但没有收入结构或会计处理 | 索取按产品家族拆分的月度收入、收入确认备忘录、递延收入,以及服务与软件的拆分 |
| 按 SKU 拆分的毛利率与服务成本 | 需要知道高端产品扣除计算和授权后,是否仍有吸引人的贡献毛利 | 无公开利润率披露 | 索取产品级毛利率、云支出、供应商授权和按 SKU 拆分的支持负担 |
| 现金、burn 和 runway | 决定 2025 年融资是否足够,还是只是一个中途站 | 无公开现金或 burn 披露 | 索取月度现金流、当前现金、招聘计划,以及基础 / 下行情景 runway |
| 留存、流失和扩张 | 区分一次性使用高峰与耐久的经常性收入 | 无公开 NRR、GRR 或 cohort 指标 | 索取过去 8 个季度按产品拆分的 cohort 留存、扩张率和流失原因 |
| 客户集中度和合作伙伴经济性 | 大型合作伙伴可能带来流量,但不一定带来健康利润率 | 具名 logo 公开,支出占比不公开 | 索取头部客户和头部合作伙伴收入占比、合同条款、收入分成或折扣安排 |
| 实际成交价和折扣政策 | 仅靠标价无法支撑利润率或估值判断 | 官方页面提到折扣和年度节省,但未披露实际净价 | 索取近期订单、价格例外、续约费率,以及按细分市场拆分的折扣 waterfall |
每个缺口都对应具体尽调材料包,因为公开记录不够细,无法支撑标准软件承销。
[CI041, CI042, CI043, CI045, CI047, CI051]展示已募集资本如何投入产品深度和企业交付,但仍无法揭示资金续航期。
这张图区分了资本获取能力和现金充足性;公开记录对前者的支撑强于后者。
[CI030, CI032, CI033, CI034, CI043, CI044]4.5 图表与附录
05产品与技术
5.1 产品界面多层展开,横跨快速 grounding、深度研究和企业代理交付
2026 年 You.com 最重要的产品技术事实,是它已经不再只卖一件东西。公开界面横跨快速网页 grounding、页面抽取、长篇研究、金融专用研究,以及服务更重的企业代理部署。自助侧,文档和定价页展示了离散的 Search、Contents、Research 和 Finance Research 界面;release notes 又增加 Express,用于单次调用的 grounded answering,以及用于已验证 snippet retrieval 的 Deep Search。企业侧,You.com 销售更宽的情报栈,把私有数据、实时网页结果、模型路由和应用层交付结合起来。这一点重要,因为公司正试图用一个架构捕获多种 jobs:给代理用的低延迟搜索、RAG pipelines 的检索材料、供分析师和高管使用的长周期综合,以及面向企业团队的私有数据助手。工作流证据也与这套栈一致。NIH 使用引用密集型研究,Maryville 在学术工作流中使用多模型定制代理,Mimecast 使用共享代理提升 GTM 生产力。实际承销读法是:产品宽度真实存在,但它抬高了执行门槛,因为 You.com 必须同时维护多个可靠性和成本边界,而不是一个范围狭窄的搜索 API。[CE001, CE002, CE003, CE004, CE005, CE006]
| 模块 / 资产 | 主要买方 / 用户 | 状态 / 成熟度 | 差异化 | 尽调缺口 |
|---|---|---|---|---|
| Web Search API | 开发者、agent 构建者、搜索驱动应用 | 已上线核心产品 | 在一个界面中提供低延迟网页 grounding,包含 snippets、metadata 和 news | 需要独立 uptime 和大客户留存数据 |
| Contents API | RAG 构建者、监控团队、agent pipelines | 已上线附加产品 | 无需自建爬取 / 解析栈,就能返回干净的 markdown 或 HTML 及 metadata | 无公开抽取失败率、覆盖统计或滥用控制 |
| Express API | 需要一次调用即可获得有来源支撑答案的应用 | 2025 年商业化推出 | 把搜索、答案和引用打包进一次请求,减少编排工作 | 公开披露的答案路由、模型选择和错误处理信息较薄 |
| Deep Search | 需要精确段落的 agents 和 copilots | 已可见推出的早期访问能力 | 用页面抓取、高亮和 fuzzy-match 验证来检索已验证摘录 | 未发现独立基准或大规模生产部署数据 |
| Research API / ARI | 分析师、咨询顾问、企业知识工作者 | 有意义的商业化界面 | 跨数百个来源做带引用的多步骤研究,并输出精修结果 | 多数性能比较由公司自编,且绑定特定工作负载 |
| Finance Research API | 金融分析师、尽调和市场研究工作流 | 新产品,但战略价值高 | 在研究编排之上叠加授权结构化数据和证据仲裁 | 依赖第三方授权数据且 latency 长,可能压缩利润率 |
| 企业情报栈 | IT、运营和部署私有数据 agents 的职能团队 | 高接触度企业产品 | 私有 RAG、定制 agents、模型无关路由、连接器和 onboarding 支持 | 需要更清晰的安全架构、SLO 和实施工作量公开细节 |
行把成熟的核心 grounding 界面与更新的专业层、高接触层分开;成熟度判断只依赖 2024-2026 年公开证据。
[CE001, CE002, CE003, CE004, CE005, CE006]| 用户任务 | 当前工作流 | You.com 方案 | 可衡量收益 | 局限 |
|---|---|---|---|---|
| 为快速 agent 回答提供 grounding | 应用需要新鲜、事实性的网页上下文 | Search API 或 Express API | 带引用的最新摘要,或一次调用得到答案 | 商品化定价压力很高,质量主张也多由公司自己撰写 |
| 把网页转成模型可用证据 | RAG 或监控管线需要干净的源文本 | Contents API | 不必自建爬虫、解析器和清洗栈 | 公开文档没有披露抽取召回率、覆盖范围或 SLA 指标 |
| 找回经过验证的支持段落 | Copilot 或 agent 需要精准摘录,而不是整页内容 | Deep Search | 经验证的片段降低 token 负担,也降低幻觉风险 | 仍被定位为早期访问,独立验证有限 |
| 产出带引用的长篇研究 | 知识工作者需要跨大量来源做多步综合 | Research API / ARI | 覆盖数百个来源,引用可追溯,并输出报告式结果 | 相比快速查询界面,延迟和成本明显上升 |
| 运行金融专项尽调 | 分析师需要匹配期间的数字,并做来源对账 | Finance Research API | 把授权金融数据、带引用研究和证据仲裁合在一起 | 公开披露的深度运行延迟可能超过两分钟 |
| 部署组织专属 agent | 企业团队需要在工作流内把私有数据和网页 grounding 结合起来 | Enterprise AI Solutions / Private RAG 企业方案 | 模型无关部署、连接器、上线导入和支持 | 安全控制只做高层描述,没有详细公开架构 |
| 提高校园或 GTM 生产力 | 部门希望共享定制 agent,并按工作流选择模型 | You.com 平台上的定制 agent | Maryville 和 Mimecast 有公开客户证据 | 案例研究没有披露硬留存或实际合同经济性 |
从客户任务出发定义产品,而不是把端点孤立罗列。
[CE003, CE004, CE005, CE006, CE010, CE016]You.com 产品技术栈的公开可见层级。
[CE001, CE007, CE012, CE016, CE017, CE018]一个典型的 You.com 驱动工作流如何从查询走向有依据的输出。
[CE002, CE009, CE010, CE017, CE018, CE019]5.2 架构强调检索编排、引用保真,以及按用例路由的专门化
公开技术材料持续把 You.com 描述成覆盖多种检索模式的编排层,而不是单体模型公司。架构从对实时网页、垂直索引和企业私有数据的采集与索引开始,然后根据任务把请求路由到不同检索路径。Fast search 返回上下文 snippets;Contents 抓取干净的页面文本;Express 把搜索和答案打包进一次调用;Research 和 ARI 跑更深的多步骤综合;Finance Research 则把开放网页替换成针对金融优化、并由授权和机构数据增强的索引。公开材料中技术差异化最强的组件是 Deep Search。公司称它会并行抓取页面,运行基于 LLM 的 highlighter,再通过 fuzzy matching 验证返回段落是否真实出现在页面上。这个设计选择值得注意,因为它明确瞄准 agentic retrieval 系统最大的失败模式:有说服力但 grounding 薄弱的摘录。栈的其余部分遵循类似逻辑。RAG 被描述为先检索再生成,引用贯穿研究界面,产品更新也反复把快速答案路径与更慢但更深的研究路径分开。架构上行空间在于跨延迟和深度的灵活性;下行是运营复杂度:更多检索界面、更多排序逻辑、更多模型路由决策,也有更多端点之间成本或质量漂移的机会。[CE011, CE012, CE013, CE014, CE015, CE016]
| 层 / 组件 | 作用 | 依赖 | 风险 |
|---|---|---|---|
| 采集与索引层 | 持续收集公开网页数据、垂直索引和企业私有数据 | 需要爬取新鲜度、连接器健康度和来源权限管理 | 覆盖缺口或摄取滞后会拖累所有上层 |
| 搜索与排序层 | 返回摘要、元数据和结果集,为快速 grounding 服务 | 依赖排序质量、新鲜度和查询路由 | 市场定价被压缩后,质量差异很难在商业上持续变现 |
| 内容抽取层 | 为下游模型抓取完整页面文本或 markdown | 依赖网页可访问性、解析、清洗和反机器人韧性 | 抽取失败或内容噪声会污染 RAG 管线 |
| 已验证摘录层(Deep Search) | 找到回答段落,并检查引用文本确实存在于源页面 | 依赖并行抓取编排、高亮质量和验证逻辑 | 如果验证薄弱,所谓幻觉优势会很快收窄 |
| 研究编排层 | 运行多步搜索、来源阅读、综合和引用映射 | 依赖模型质量、成本纪律和来源对账 | 长流程可能变贵,而且跨端点调试困难 |
| 企业数据与交付层 | 连接私有数据、SDK、MCP 界面和终端应用 | 依赖权限、合作伙伴 API、SDK 兼容性和支持运营 | 连接器蔓延和第三方平台变化可能打断客户工作流 |
把公开技术栈建模为检索与编排系统,而不是单一端点。
[CE007, CE011, CE012, CE013, CE014, CE017]5.3 信任姿态在商业上纸面很强,但公开技术披露仍落后于平台野心
You.com 显然知道,企业 AI 买家关心的不只是原始搜索质量,也同样关心信任和可部署性。定价和企业页面强调零数据留存、SOC 2 认证、DPA ready、自定义 QPS limits 和私有数据集成。隐私政策补充了有用的技术背景:服务使用多个 LLM providers,明确点名 OpenAI、Anthropic 和 Google,内容托管在 AWS,并称从已连接第三方系统拉取的数据不会用于训练第三方 AI 模型。条款也说明了更冷静的现实:部分功能依赖第三方 API 按合理条件持续可用,You.com 也保留修改或停止功能的权利。这让信任故事足以打开企业大门,但从尽调角度看仍未完全透明。公开材料没有给出架构层细节,例如模型路由政策、fallback paths、事故历史、签署的 uptime commitments,或安全声明背后的详细控制映射。不过,集成广度是真优势。公司现在接入 MCP-enabled IDEs、Databricks、Zapier、Vercel AI SDK、OpenAI SDK、Claude Agent SDK 和常见企业数据存储。分发扩大了使用,也增加了连接器脆弱性、权限错误和外部平台变化风险。[CE007, CE008, CE009, CE010, CE020, CE021]
| 控制 / 信号 | 状态 | 范围 | 缺口 |
|---|---|---|---|
| 研究界面内联引用 | 公开可见 | Search、Research、ARI、Finance Research,以及 Express 式带 grounding 的答案 | 有引用不等于来源选择质量或答案正确性得到证明 |
| Deep Search 的已验证片段设计 | 公开描述 | 从实时网页抽取段落,并用模糊匹配验证 | 未发现独立审计或详细公开失败分析 |
| Zero Data Retention、SOC 2、DPA-ready 主张 | 公开营销 | 面向企业买家的定价和企业界面 | 公开页面没有展示详细控制映射、审计范围或合同例外条款 |
| 不用接入的第三方企业数据训练 | 公开描述 | 隐私政策如何处理接入数据源和第三方 AI 提供商 | 需要为企业 SKU 和数据流图做合同确认 |
| 模型无关部署与多个 LLM 提供商 | 公开描述 | 跨 GPT、Claude、Gemini 和底层服务提供商的企业路由 | 模型路由、降级和故障行为没有公开文档 |
| Evals 基础设施和开放 benchmark 仓库 | 公开可见 | DeepConsult 仓库以及公司 benchmark 叙述 | 多数 benchmark 设置仍由公司撰写或筛选,而不是独立实验室 |
信任信号具备商业意义,但仍弱于完整公开的安全与可靠性档案。
[CE008, CE017, CE018, CE020, CE026, CE039]决定 You.com 能否守住产品质量和企业信任的关键依赖。
[CE020, CE022, CE026, CE029, CE039, CE040]5.4 成熟度在核心网页 grounding 上最强;在企业信任或专业数据深度依赖外部供应商的地方最弱
You.com 栈的成熟度不均匀,但脉络清楚。Search、Contents 和通用 grounding 层看起来是最成熟的界面,因为它们有公开定价、文档、多条集成路径、客户引用和反复发布节奏。Research 和 ARI 也不像实验品,尤其是公司已经开源评估工具,并围绕引用密集型工作流搭建客户故事。Finance Research 在战略上有意思,因为它从开放网页检索升级到授权金融数据加证据裁决;但它也最清楚地提醒我们,部分护城河位于 You.com 自身之外,依赖付费数据访问和整理。外部市场文件强化了竞争框架:OpenAI、Google、Azure、Brave、Exa、Tavily、Elastic 和 Kendra 都包装了搜索、抽取、grounding 和企业检索的某种组合。这意味着产品问题不是 You.com 是否独占品类,而是它在模型无关路由、已验证 snippets、私有数据混合和开发者分发上的组合,是否仍比相邻栈可衡量地更好。因此,最大的技术风险落在依赖集中、benchmark 可信度和隐藏运营复杂度上。公开证据支持一个强切口,但还不能证明平台故事已经完全 de-risked。[CE018, CE019, CE022, CE023, CE024, CE025]
| 日期 / 阶段 | 功能 / 里程碑 | 状态 | 含义 | 来源 |
|---|---|---|---|---|
| 2024-06 | YOU API 的网页规模搜索定位,以及未来查询重写 / agentic 路线图 | 已完成的历史发布标记 | 显示平台从简单搜索 UI 转向可编程 grounding 基础设施 | You.com 资源 |
| 2025-02 | ARI 发布,面向专业级深度研究 | 已完成发布 | 从检索扩展到长篇研究输出和企业知识工作 | You.com 资源 |
| 2025-09 | Express API 和 Contents API 发布 | 已完成发布 | 把一次调用 grounding 和模型可用页面抽取正式做成独立产品 | 2025 年 9 月汇总 |
| 2025-11 | 增加 Deep Search、更广的 MCP 工具、Python SDK、Zapier、Databricks 界面 | 已完成发布波次 | 让平台在 IDE 和工作流生态中更适合 agent | 2025 年 11 月汇总 |
| 2025-12 | 增加 Evals-as-a-Service、垂直索引、API playground、TypeScript SDK,以及 OpenAI/Claude/Vercel 集成 | 已完成发布波次 | 拓宽开发者分发,并强化按领域定制检索 | 2025 年 12 月汇总 |
| 2025-09 | OpenAI GPT-OSS 把 You.com 集成为核心搜索提供商 | 已完成合作伙伴里程碑 | 外部验证 You 可作为开放权重模型的 grounding 基础设施 | You.com/OpenAI 组合 |
| 2026-05 | Finance Research API 基于授权金融数据和研究编排发布 | 新的专业界面 | 推动公司上行进入更高价值的尽调和金融工作流 | You.com 资源和文档 |
时间线强调产品与平台演进,而不是融资事件。
[CE015, CE016, CE017, CE018, CE020, CE021]You.com 主要产品技术界面的定性成熟度视图。
[CE001, CE018, CE020, CE021, CE022, CE023]5.5 图表与附录
06客户
6.1 客户原型很宽,但公开证明分裂为深度企业故事和渠道主导的 API 分发
公开记录支持一条双轨客户故事。一条是典型企业部署:You.com 向重视引用、私有数据和工作流专用代理的机构销售服务更重的情报栈。NIH、Maryville、Mimecast、dpa、HanseMerkur 和 Wort & Bild 都落在这个桶里,尽管各账户的证明深度差异很大。第二条是平台分发:You.com 也销售 API 和搜索基础设施,经由 MCP、GPT OSS、Vercel、Zapier、MindStudio 和云市场界面流动。这把买家宇宙从中央 IT 团队和知识工作负责人,拓宽到开发者、产品团队,以及可能永远不会购买独立席位产品的合作伙伴生态。关键承销细节是,买家、用户和付款方经常不同。Maryville 领导层和教师治理 rollout,学生消费产品;NIH IT 赞助使用,研究人员和承包商实际使用;Mimecast 为业务团队集中选择产品;GPT OSS 或 Remy 这样的伙伴渠道,则可能把 You.com 嵌进其他产品。这让客户基础比简单的企业搜索标签更宽,但也意味着单靠公开数据很难读出商业结构。[CU001, CU002, CU003, CU004, CU013, CU019]
| 分层 | 买方 / 付款方 | 主要用户 | 代表性证据 | 战略价值 | 主要公开缺口 |
|---|---|---|---|---|---|
| 研究密集型机构 | IT 领导 / 部门预算 | 研究人员、分析师、承包商 | NIH 案例研究 | 验证重引用的深度研究工作流 | 没有合同期限或续约数据 |
| 高等教育机构 | 大学领导 / 学术部门 | 教师和学生 | Maryville 案例研究 | 展示全校级落地和学生重复使用 | 未披露席位数或支付价格 |
| 企业业务团队 | CIO 或职能运营领导 | GTM、内容和业务团队 | Mimecast 案例研究 | 展示跨团队 agent 共享和可衡量内容提升 | 未披露 GTM 工作流之外的范围 |
| 媒体与新闻机构 | 编辑或产品领导 | 编辑和记者 | dpa 合作 | 扩展到私有数据和通讯社工作流 | 没有 ROI 或续约证据 |
| API / 平台客户 | 产品和工程负责人 | 客户产品内的开发者和最终用户 | Series C 文章点名 DuckDuckGo、Windsurf、Harvey | 可带来大规模查询量,并把 You.com 深度嵌入 | 未披露支出、合同规模或留存 |
| 合作伙伴带来的构建者和自动化用户 | 开发者或业务运营者 | 应用构建者、无代码团队、agent 框架 | GPT OSS、Zapier、Vercel、MindStudio、MCP 等集成 | 降低获客摩擦,并扩大 TAM | 未披露渠道转化和变现 |
行把直接企业部署与嵌入式 API、合作伙伴渠道采用分开,因为买方、用户和付款方往往不是同一角色。
[CU001, CU002, CU003, CU013, CU019, CU021]| 指标 | 数值 | 日期 / 来源背景 | 置信度 | 含义 | 缺失分母 |
|---|---|---|---|---|---|
| NIH 试点用户 | 30-35 | NIH 案例研究 | 中 | 显示在更广访问前先做受控初始推出 | 未披露全机构合资格用户基数 |
| NIH 扩展目标 | 120 个许可 | NIH 案例研究 | 中 | 支持部门扩展不只是试用好奇 | 没有付费席位数或续约时间 |
| NIH 影响的承包商基数 | 50-60 名承包商 | NIH 案例研究 | 中 | 暗示使用延伸到单一核心团队之外 | 没有直接 MAU 或 WAU 指标 |
| Maryville 教师采用 | 76% | Maryville 案例研究 | 中 | 内部推动强,且重复使用 | 未披露教师绝对人数 |
| Maryville 定制 agent | 600+ | Maryville 案例研究 | 中 | 表明用户主动创建,而不只是被动聊天 | 未披露创建者到 agent 的分布 |
| Maryville 学生使用频率 | 50%+ 每周多次 | Maryville 案例研究 | 中 | 在高要求项目中,重复使用代理清晰 | 没有全校活跃用户数 |
| 公开 API 查询量 | 1B+ 月查询量 | Series C 公告 | 中 | 显示 API 侧已有可观规模 | 未按客户或产品披露查询量 |
本表把公开原始指标和业务解读分开,因为多数指标缺少真正做 cohort 分析所需的客户数分母。
[CU006, CU008, CU011, CU012, CU027, CU029]公开证据显示,客户可能从工作流痛点走向试点、团队铺开,最终进入嵌入式分发或更广账户扩张。
这是基于公开证据的旅程,不是公司披露的内部漏斗。
[CU001, CU002, CU006, CU012, CU017, CU019]6.2 最强的具名证明来自 NIH 和 Maryville;之后的证据明显变薄
最有说服力的客户证据,不是 logo 出现,而是工作流细节、用户扩张和量化结果。NIH 是最清楚的研究密集型案例:客户故事解释了旧工具为何失败、30-35 人 pilot 如何扩展到接近 120 个 license、研究人员和承包商实际如何使用产品,以及买家为什么关心引用可追溯性。Maryville 是最清楚的教育案例:它把 You.com 与护理课程退课减少、更多学生保持进度、教师高频采用、数百个定制代理和学生重复使用联系起来。Mimecast 提供第三个证明点,但披露结果更窄:sales-enablement content 提升 20%,GTM 流程更快。dpa 增加了新闻编辑室工作流证明和 Team Plan access,但没有续约或 ROI 数据。企业页面还加入 HanseMerkur 和 Wort & Bild,但这些引用明显比旗舰案例研究更轻。正确读法是,You.com 显然有真实企业部署,但公开证明质量集中在少数精心挑选的故事上,而不是广泛、独立交叉印证的 reference base。[CU004, CU005, CU006, CU007, CU008, CU009]
| 客户 / 标识 | 分层 | 部署或用例 | 生产 / 试点 | 公开结果 | 局限 |
|---|---|---|---|---|---|
| NIH | 公共卫生研究 | 带引用的深度研究、审阅、起草、承包商支持 | 试点正扩展到更广生产使用 | 30-35 名试点用户正扩展到 120 个许可;估计年度生产力价值 >$1M | 所有证据都来自公司发布的客户故事 |
| Maryville University | 高等教育 | 护理课程和更广教师工作流中的定制 AI agent 与全天候学生支持 | 生产课堂部署 | 退课率下降 78%;教师采用率 76%;600+ 个 agent;收入影响 >$3M | 未披露合同期限或付费席位数 |
| Mimecast | 企业网络安全软件 | 面向 GTM 和内容工作流的共享定制 AI agent | 生产内部工作流部署 | 销售赋能内容增加 20% | 结果范围较窄,且没有公开续约数据 |
| dpa | 媒体 / 新闻 | Agentic 新闻发现、策展、Team Plan、定制 agent | 商业合作 / 部署 | 工作流细节和客户访问路径公开 | 没有量化采用或 ROI 指标 |
| HanseMerkur | 保险 | 跨团队安全 AI 集成 | 仅具名引用 | 显示与受监管行业相关 | 没有量化结果或推出细节 |
| Wort & Bild Verlag | 健康出版 / 媒体 | 面向记者的定制研究 agent | 仅具名引用 | 显示适配编辑工作流 | 没有规模或留存证据 |
| DuckDuckGo / Windsurf / Harvey | 搜索 / 开发者 / AI-native 软件 | 融资公告点名的 API 客户关系 | 仅具名关系 | 释放平台级客户信号,且整体月查询量 >1B | 没有客户专属范围、定价或期限 |
公开具名客户证据只是截至运行日期公司重点展示的部分样本,不是完整客户名单。
[CU004, CU006, CU008, CU010, CU011, CU012]具名客户标识之间的差异更多来自证明深度,而不是行业广度;NIH 和 Maryville 的证据明显更强。
评级是截至 2026-06-14 对公开证据质量的定性判断,不是客户健康度评分。
[CU004, CU017, CU018, CU027, CU028, CU039]6.3 Marketplaces 和框架集成降低部署摩擦,但公开采购证明落后于营销叙事
You.com 的 2026 年客户动作越来越有渠道意识。合作伙伴和集成文档把公司定位成不仅是按席位销售的 AI 应用,更是一个可通过 MCP、云 marketplaces、SDKs、工作流自动化和模型生态获得的基础设施层。商业上,这很重要,因为这些渠道缩短开发者 time-to-value,也让团队无需 re-platforming 就能采用 You.com。GPT OSS 把 You.com 变成默认浏览器工具后端,Zapier 把 API 带入 no-code 工作流,Vercel 把搜索和研究工具打包进既有开发者栈,MindStudio/Remy 则把网页情报变成可编译进生成应用的依赖。企业页面也强化同一信息,承诺低负担 go-live 和 white-glove onboarding。限制在于,采购证明仍然混杂。You.com 合作伙伴页面声称 AWS、Azure 和 Databricks marketplace 可用,但截取的 AWS Marketplace 搜索结果本身并没有显示清晰可读的 listing 或 price card。因此,更广的部署故事可信且战略上有吸引力,但并非每个声称渠道都有同等强度的公开证据支撑。[CU019, CU020, CU021, CU022, CU023, CU024]
| 渠道 / 合作伙伴 | 暴露内容 | 主要买方或用户 | 证据强度 | 商业含义 | 关键限制 |
|---|---|---|---|---|---|
| MCP 与 IDE 客户端 | 免费起步的网页访问,加 MCP server | 开发者和技术团队 | 直接文档 + 合作关系 | 降低获客漏斗顶部的采用成本 | 免费使用未必能清晰映射到付费转化 |
| GPT OSS | search/open/find 的默认浏览器后端 | 交付开放权重 agent 的开发者 | 官方和代码仓库证据强 | 把 You.com 嵌入正在增长的 agent 技术栈 | 单个终端部署的客户经济性未披露 |
| Zapier | 搜索、内容抽取和新闻动作 | 业务运营人员 / 无代码用户 | 文档证据强 | 把 You.com 拉进非技术工作流 | 成功可能更多取决于连接器质量,而非核心搜索质量 |
| Vercel AI SDK | 可直接接入的搜索、内容和研究工具 | 应用开发者 | 文档证据强 | 缩短工程上线时间 | 可能停留在功能使用,而非持久账户扩张 |
| MindStudio / Remy | 为生成应用内置的实时 web 层 | AI 应用构建者 | 合作伙伴文章 + 合作伙伴网站证据强 | 形成嵌入式分发 | 合作伙伴终端用户可能不会自认是 You.com 客户 |
| AWS / Azure / Databricks marketplace 渠道 | 采购和部署入口 | 企业平台团队 | 公司声称,部分得到佐证 | 可释放已承诺云支出和联合销售 | 可读公开抓取弱于公司的主张集合 |
分发表聚焦采用动作,而不是技术功能细节,因为本章判断门槛关注的是客户触达和商业耐久性。
[CU019, CU020, CU021, CU022, CU024, CU025]漏斗从大量公开入口收窄到少数有重复使用或续约证据披露的账户。
末端节点刻意较薄,因为公开记录没有披露留存 KPI。
[CU004, CU019, CU024, CU025, CU026, CU029]6.4 重复使用代理信号存在,但留存耐久性仍是假设,而不是已发布 KPI
公开证据足以说明,部分客户不只是试用一周就停;但还不足以支撑传统 SaaS 式耐久性的判断。NIH 的试点扩展和 Maryville 参与度上升,都是合格的重复使用信号。Mimecast 的共享代理工作流、dpa 的 Team Plan 访问,也说明 You.com 卖的不是纯个人工具,而是带协作属性的产品。评论数据给出另一层视角:G2 整体正面,Trustpilot 则明显负面,支持、计费和可靠性投诉很多。这个分化很关键,因为它暗示了两套体验:一边是经过筛选的企业案例,另一边是更粗糙的大众客户体验。企业信任叙事方向上有帮助——零数据留存、不训练连接方数据、部署支持,都能降低企业顾虑——但条款也承认公司依赖第三方 API,并保留变更或停用功能的权利。合起来看,You.com 看起来有能力赢下并扩张客户,但公开记录仍没有披露 GRR、NRR、续约率、平均合同期限,或按 cohort 展开的扩张情况。[CU029, CU030, CU031, CU032, CU033, CU034]
| 指标 | 数值 / null | 分层 | 置信度 | 说明 | 尽调追问 |
|---|---|---|---|---|---|
| Maryville 教师采用 | 76% | 高等教育 | 中 | 教师群体中很强的重复使用代理 | 要求按学期提供 cohort 留存和付费席位转化 |
| Maryville 学生每周多次使用 | 50%+ | 高等教育 | 中 | 显示后续课程中学生已形成习惯使用 | 要求按项目和课程提供 MAU/WAU |
| NIH 推出扩展 | 30-35 名试点用户 -> 120 个许可 | 研究机构 | 中 | 企业耐久性最好的公开扩张代理指标 | 索取付费续约日期、席位使用率和拥护者地图 |
| G2 评分 | 19 条评论给出 4.4/5 | 一般用户群 | 中 | 第三方评论信号偏正面,但样本仍小 | 索取按套餐类型和企业规模拆分的评论构成 |
| Trustpilot 评分 | 36 条评论给出 2.4 | 一般用户群 | 中 | 支持和计费信号偏负面,可能影响续约 | 索取投诉解决和退款指标 |
| NRR / GRR / logo 流失 | 全部细分 | 低 | 核心留存指标仍未披露 | 索取过去 12 个月 NRR、GRR、总 logo 留存和净 logo 留存 |
Null 表示未找到公开留存 KPI;代理指标不应被当作 NRR 或 GRR 的替代品。
[CU029, CU030, CU031, CU032, CU033, CU040]6.5 最大的客户尽调风险不是没有 logo,而是这些 logo 背后的集中度和续约可见度缺失
本章最主要的反向解读是:客户证明真实存在,但相对于故事的雄心仍然偏薄。Series C 文章称每月 API 查询超过 10 亿次,并点名 DuckDuckGo、Windsurf 和 Harvey,战略意义很强;但这些关系都没有给出合同规模、工作负载组合或留存细节。多数伙伴驱动和市场渠道也是如此:它们显然扩大了触达,却也可能把技术和商业依赖集中到外部生态上。公开采购证据没有完全闭环,第三方评论还暴露了支持和激活问题;在团队场景里,这些问题可能被放大。最重要的是,公司没有披露客户数、头部客户集中度、分业务 ARR 或续约统计。投资人因此必须把已被证明的事实和只是合理推测的部分分开。已证明的是:You.com 有可部署产品、真实机构背书和有意义的分发能力。未证明的是:收入基础有多分散、客户第一年之后有多粘、渠道流量能否转成耐久经济性。[CU027, CU028, CU031, CU033, CU037, CU038]
| 驱动因素或风险 | 当前公开信号 | 潜在影响 | 置信度 | 尽调路径 |
|---|---|---|---|---|
| 跨团队 agent 共享 | Maryville、Mimecast 和 dpa 都暗示存在协作 / 团队工作流 | 如果使用在一个账户内扩散,就支撑先落地再扩张 | 中 | 索取席位增长、部门数量和交叉销售附加率 |
| API 合作伙伴量集中 | 月查询量 1B+,且有具名 API 客户,但未披露单账户构成 | 少数大客户可能主导收入或基础设施负载 | 中 | 索取前 10 大客户收入和查询集中度 |
| Marketplace 与渠道依赖 | AWS / Azure / Databricks / GPT OSS / Zapier / Vercel / MindStudio 入口均被重点强调 | 分发摩擦低,但更容易受外部平台变化影响 | 中 | 索取渠道来源管线、转化率和支持成本数据 |
| 采购可见性缺口 | AWS 可读抓取文本中未出现具体 listing | 可能拖慢对企业采购就绪度的尽调 | 中 | 索取直接市场平台上架 URL、SKU 和计费机制 |
| 续约可见性缺口 | 未公开 NRR、GRR、合同期限或流失指标 | 无法为收入耐久性做承保 | 低 | 索取 cohort 留存、续约日历和流失原因 |
本表把上行杠杆和未解决风险拆开,避免把缺失的留存数据误读成耐久性证据。
[CU037, CU038, CU040, CU042, CU043, CU044]6.6 展示材料
07风险
7.1 即使没有头部诉讼,法律、隐私和企业信任风险依然真实
法律和监管栈是尽调最先该感到不舒服的地方。You.com 有足够公开文件去卖给企业,但这些文件也明确写出公司不控制什么。隐私政策点名多个 LLM 供应商,提醒用户不要上传受限数据,披露继续转移和服务提供商共享,并保留法律留存例外。条款更进一步,把第三方输出和 API 的风险转回用户,同时保留修改、暂停或停用功能的广泛权利。这些条款对快速迭代的 AI 产品很常见,但在这里更重要,因为 You.com 把信任、引用和适配受监管工作流作为价值主张的一部分在销售。2026 年 AI Act 时间线和 HanseMerkur 的公开评论显示,受监管买家越来越需要可审计性、文档和可预测控制,而不只是好答案。缓释因素是,You.com 确实发布了法律中心、DPA-ready 语言、治理模板和企业信任主张。未解的问题在于深度:公开网页更清楚展示的是文件壳和营销说法,而不是谈判后的赔偿、控制映射、事件义务或客户级 SLA 语言。[CR001, CR002, CR003, CR004, CR006, CR007]
| 规则 / 许可证 / 案件 | 司法辖区 | 状态 | 可能性 | 严重性 | 缓释措施 | 剩余暴露 | 尽调路径 |
|---|---|---|---|---|---|---|---|
| EU AI Act / GPAI 与透明度义务 | 欧盟 | 实施时间表已启动;GPAI 规则已生效,透明度规则预计 2026 年 8 月到期 | 中 | 高 | 治理模板、引用和企业控制信息有帮助,但不能替代合规运营 | 中高 | 索取 AI Act 角色映射、部署方 / 提供方划分,以及面向受监管部署的逐产品合规备忘录。 |
| Prompt 数据与后续传输隐私风险 | 美国 / 欧盟 / 英国 / 全球 | 持续运营风险 | 高 | 高 | 受限数据警告、联网应用不训练声明和 DPF 措辞能降低暴露,但无法消除 | 高 | 审阅数据流图、默认留存设置、分处理方、删除 SLA,以及任何客户特定的数据驻留承诺。 |
| 第三方输出、API 与链接内容责任分配 | 全球合同法 | 目前已写入公开条款 | 高 | 高 | 条款清晰分配风险并保留服务变更权;模型无关路由在边际上降低单一供应商锁定 | 高 | 获取超出公开条款的谈判赔偿立场、客户救济和备用承诺。 |
| 企业信任包充分性 | 客户签约 / 采购 | 公开材料可见但不完整 | 中 | 中高 | Legal hub、安全附录、DPA 模板和治理套件为安全审查提供起步包 | 中高 | 获取现场企业交易中使用的完整 DPA、安全概览、SLA 和控制映射。 |
| 受监管或事实型工作流中的内容与引用责任 | 全球 | 未发现公开案件,但产品界面带来剩余暴露 | 中 | 高 | 以引用优先的产品设计和来源验证信息,相比纯聊天产品可降低幻觉风险 | 中高 | 索取内部事故日志、争议处理,以及外部律师对检索、版权和错误归因暴露的意见。 |
各行按剩余承保严重性给法律与合规暴露排序,而不是按是否存在公开诉讼排序;最大缺口在于经谈判的合同条款和实施证据。
[CR001, CR003, CR004, CR006, CR007, CR009]You.com 最高的剩余风险集中在法律合规负担、模型依赖和商业化不透明;公开缓释措施已经可见,但仍有一部分由营销叙事驱动。
这个矩阵使用有来源支撑的序位标签,而不是编造概率。它用于给剩余暴露排序,不是统计预测。
[CR007, CR010, CR012, CR013, CR018, CR020]7.2 模型供应商依赖和可靠性不透明,最可能把风险传导成客户痛点
核心经营风险不是 You.com 今天没有模型或功能,而是太多体验仍压在交易对手和难以观察的运行质量上。隐私政策点名 OpenAI、Anthropic 和 Google 作为 LLM 供应商,条款又明确说第三方 API 可能以不合理条件消失。这不是理论上的样板条款。Trustpilot 投诉已经指向供应商施加的请求限制,GPT OSS 文档也让 You.com 成为另一平台浏览器工具的后端。与此同时,公司自己的幻觉和延迟指南承认了投资人担心的同一组运营事实:幻觉没有被消除,监控仍然重要,下游依赖可能主导延迟和故障表现。公司确实宣传零留存、SOC 2 和基准测试胜出,但这些信号大多仍由公司自己撰写。关于 uptime 承诺、事件历史、复盘纪律或连接器权限治理,独立公开证据很少。换句话说,信任故事在商业上有用,但剩余敞口仍然像是模型路由、配额管理和运营支持质量都可能在公司证明机构级可靠性之前先被客户看见。[CR005, CR007, CR014, CR015, CR016, CR017]
| 失败模式 | 可能性 | 严重性 | 缓释成熟度 | 剩余暴露 | 未解决缺口 |
|---|---|---|---|---|---|
| 上游模型限制、额度变化或重新定价会直接传导给付费用户 | 高 | 高 | 低中 | 高 | 需要逐供应商备用政策、预留容量和价格传导规则。 |
| 高风险工作流中的幻觉或引用失败 | 中 | 高 | 中 | 中高 | 需要独立错误率证据、客户 QA 流程和部署后监控阈值。 |
| 下游依赖和排队导致延迟飙升或体验降级 | 中高 | 中高 | 中 | 中高 | 需要真实 P95/P99 数据、事故历史和合同约束的服务等级。 |
| 付费套餐的支持和激活摩擦 | 中 | 中 | 低中 | 中 | 需要支持 SLA 指标、升级规则,以及按工单主题绑定流失的分析。 |
| 集成扩展后,连接器和私有数据权限膨胀 | 中 | 高 | 中 | 中高 | 需要权限模型、连接器审计日志,以及覆盖受支持数据源的管理员控制证据。 |
本登记表强调最可能先被客户感知、再成为头条事件的失败模式:限制、质量漂移、延迟和支持摩擦。
[CR005, CR014, CR016, CR018, CR019, CR020]| 依赖 | 交易对手 | 角色 | 集中度 | 失败场景 | 严重性 | 缓释措施 | 剩余暴露 |
|---|---|---|---|---|---|---|---|
| LLM 推理供应商 | OpenAI / Anthropic / Google | 模型访问、答案质量,以及部分核心能力路由 | 高 | 供应商重新定价、限流或弃用,迫使路由突然切换或压缩利润率 | 极高 | 模型无关定位和多供应商降低单一供应商锁定 | 高 |
| 浏览器工具分发 | OpenAI GPT OSS 生态 | 浏览器工具 web action 的默认后端 | 中高 | OpenAI 更改默认集成,或开发者心智转向自托管 / 开放替代方案 | 高 | API 界面和其他集成在一定程度上分散渠道 | 中高 |
| 超大云厂商采购渠道 | AWS Marketplace / AWS EDP 动作 | 企业采购捷径和发票整合 | 中 | 渠道经济性或 AWS 合同机制塑造谁控制采购和折扣 | 中高 | AWS Marketplace 之外仍有直销动作 | 中 |
| 企业搜索竞争 | Google Cloud / Brave / OpenAI | 带隐私、引用或打包优势的竞争性有依据搜索基础设施 | 高 | 资本更厚的平台压低价格,或把搜索打包进更大的 AI 合同 | 高 | You.com 靠模型无关路由、引用和企业服务竞争 | 高 |
| 具名 API 客户和参考账户 | DuckDuckGo / Windsurf / Harvey / 旗舰企业 logo | 采用证据,也可能是查询量锚点 | unknown | 少数大客户在缺乏公开可见性的情况下主导工作负载或收入 | 高 | 参考账户广度覆盖多个垂直和渠道 | 中高 |
依赖风险不是单一交易对手,而是横跨模型、分发、超大云厂商采购,以及一组仍不透明的大型使用锚点。
[CR001, CR007, CR016, CR017, CR031, CR032]大多数下行情形都经过一条短因果链:信任和平台风险会传导到客户信心、定价权、支持负担,最终影响估值纪律。
这张图是定性、方向性的。它展示一种风险如何传导到另一个业务变量,而不是衡量影响有多大。
[CR018, CR020, CR025, CR039, CR042, CR043]You.com 的外部依赖网横跨模型、采购和竞争平台;其中多个节点可以同时影响产品质量和议价杠杆。
这张依赖图标出外部交易对手在哪些位置会同时影响产品体验和商业杠杆,并不意味着各节点权重相同。
[CR001, CR016, CR031, CR032, CR033, CR034]7.3 商业化风险来自价格压力、证明集中,以及仍在成形的收入模型
承销问题不再是 You.com 有没有人想用的产品,而是公司能否在更大平台压缩品类之前,把需求转成耐久且分散的经济性。TechCrunch 2023 年和 2024 年报道像是在公开记录中呈现一场战略转向:先是一家探索私人广告的消费者搜索公司,后来变成一家生产力和企业公司,同时承认聊天里的广告仍未解决。这个转向可能合理,但也抬高了执行门槛,因为公司现在试图同时服务自助 API、市场、嵌入式平台渠道和企业项目。定价有两面性。公开价格表足够有吸引力,可以赢下开发者;但它也在多个产品面上设定了激进预期,而这些产品可能依赖昂贵的上游模型和检索基础设施。竞争并不抽象。OpenAI 在更大的 API 平台里把网页搜索作为工具出售,Google 用合规功能推广 grounded enterprise search,Brave 则销售带引用的搜索和企业支持。与此同时,客户证明仍集中在少数打磨过的故事里——NIH、Maryville、Mimecast、DPA、HanseMerkur、Wort & Bild——再加上公司撰写的 Series C 文章,点名大型 API 客户却不披露支出、续约或集中度。这足以证明相关性,但不足以让人靠信念承销耐久性。[CR027, CR028, CR029, CR030, CR031, CR032]
| 角色 / 职能 | 依赖或缺口 | 可能性 | 严重性 | 缓释措施 | 尽调路径 |
|---|---|---|---|---|---|
| 创始人 / CEO 叙事和企业战略 | Richard Socher 仍是产品方向、变现叙事和投资者故事的核心公开战略制定者 | 中 | 高 | 更广的高管班底正在成形,客户证据也支撑市场需求 | 索取授权分工、继任材料和董事会层面的运营节奏。 |
| 技术领导层交接 | Bryan McCann 转任顾问;Saahil Jain 于 2026 年 3 月出任 CTO | 中 | 中高 | 这次晋升来自内部长期建设者,而非外部招聘 | 审阅工程组织深度、架构负责人,以及可靠性、安全和平台团队招聘计划。 |
| GTM 跨度 | You.com 同时服务自助式 API、免费 MCP 用户、市场平台和企业部署 | 高 | 中高 | 市场平台、法律和客户引导脚手架显示,公司在有意搭建可重复动作 | 索取按细分拆分的管线、转化率、支持负担和各动作利润率。 |
| 信任与合规运营 | 公开信任材料偏营销,事故、SLA 和控制权属披露不足 | 高 | 高 | Legal hub、治理套件和企业文档显示公司意识到该问题 | 索取具名控制负责人、事故响应证据、支持人员配置和客户审计结果。 |
执行风险与其说来自野心不足,不如说来自公开运营证据尚未跟上之前,就要同时扩张多条渠道、信任义务和产品界面。
[CR027, CR028, CR029, CR039, CR040, CR041]7.4 可见缓释因素有帮助,但投资案例仍需要明确否决标准和私下尽调
正确姿态应是有条件判断,而不是非黑即白。You.com 在公开视野里确实有不少缓释因素:模型无关定位、来自受监管和重视引用买家的客户故事、DPA 和安全文件壳、治理模板、免费到付费的开发者漏斗,以及 AWS Marketplace 这类采购路径。这些都是真实的降风险信号。但它们不能替代买家级尽调。对本章来说,最高优先级的否决标准也都可以跟踪。如果隐私、信任或监管事件在公司证明成熟控制之前落地,企业叙事应迅速重估。如果模型供应商涨价、弃用或限流,而公司没有可信后手,利润率和可靠性故事会同时变弱。如果管理层无法披露客户集中度、续约行为和产品级毛利率,公开证明就仍然过于精修,无法支撑有信心的承销。如果付费计划的可靠性或支持质量反复浮出水面,而不是孤立噪音,那就是商业化问题,不是表面瑕疵。实际尽调要求很直接:先拿到信任包、供应商地图、集中度清单、支持和事件历史,以及真实合同经济性,再把企业故事视为已充分降风险。[CR010, CR011, CR018, CR020, CR034, CR041]
| 风险 | 可监控触发因素 | 阈值 / 事件 | 行动含义 |
|---|---|---|---|
| 信任或监管事故 | 与隐私、留存或 AI 合规相关的公开披露、监管机构接触或客户数据泄露通知 | 任何与零留存、安全或受监管工作流主张相矛盾,且没有快速控制和买方级解释的事故 | 立即重估投资论点;暂停承保,直到事故范围和补救措施得到独立验证。 |
| 上游供应商重新定价或访问丧失 | 影响主要模型的支持通知、产品限制变化或供应商合同重置 | 丧失旗舰模型的同等访问能力,或在没有替代经济性的情况下发生实质成本传导 | 重做利润率假设,并在继续推进前要求证明备用路由。 |
| 客户集中度不透明 | 管理层在尽调中拒绝披露头部账户构成、续约行为或查询集中度 | 无法提供可信的头部客户、NRR/GRR 或工作负载集中度明细 | 将收入耐久性视为未经证明,并把估值立场调得更保守。 |
| 付费套餐可靠性和支持恶化 | 激活、额度或未解决企业阻塞点上反复出现投诉 | 一个季度内反复出现未解决的付费客户失败,而非一次性噪音 | 下调商业化可扩展性信心,并要求更强的支持指标。 |
| 竞争性价格压缩 | OpenAI、Google、Brave 或类似打包产品缩小有依据搜索与引用上的价值差距 | 持续需要匹配竞争对手价格,却没有利润率韧性或更高价值企业附加的证据 | 假设贡献利润率更低,并降低承保溢价倍数的意愿。 |
| 领导层或执行断裂 | 高管离职、信任包交付延误,或无法连贯支撑多条 GTM 动作 | 创始人 / 高管出现重大扰动,且没有清晰继任或运营冗余 | 重新评估执行折价,并要求更强治理权,或等待稳定后再推进。 |
这些触发因素刻意保持可监控。目标不是虚假的精确性,而是定义哪些事件应改变估值纪律或停止尽调流程。
[CR010, CR018, CR020, CR023, CR025, CR031]08估值
8.1 投资正题与反题
You.com 的多头逻辑不难理解。公开来源显示,它有真实产品面,而不只是一个通用 AI wrapper:低摩擦搜索 API、更高价的研究和金融产品、私有数据集成,以及企业部署构建。客户证据也好于普通早期 AI 故事。NIH、Maryville、DPA、HanseMerkur 和 Mimecast 都支持一个判断:在受监管或重视准确性的场景里,带引用的搜索和研究工作流可以有价值;Series C 文章还点名了 DuckDuckGo、Windsurf、Harvey 等 API 客户。在这个背景下,$1.5 billion 的 Series C 表面上并不荒唐。反题在于,投资人真正需要用来估值的几乎每个指标都还藏着。公司声称每月 API 查询超过 10 亿次,也曾称 ARR 加速,但公开来源没有披露实际净定价、收入组合、毛利率、留存或客户集中度。与此同时,基础检索在 You.com、Brave、Exa 和 Perplexity 之间已经被价格压缩,OpenAI 和 hyperscalers 这类更大平台还能把检索打包进更广的栈里。因此,估值问题不是 You.com 有没有意思,而是叙事背后的私有经济性是否足够好,能支撑在当前标记上支付溢价。[CV001, CV005, CV007, CV008, CV010, CV011]
| 维度 | 评估 | 支撑证据 | 决策含义 |
|---|---|---|---|
| 建议 | 继续研究 | 真实牵引力与融资支持让机会有研究价值,但经济性披露仍太薄,不足以支撑确信买入。 | 没有对收入质量和利润率做私下尽调前,不应按当前估值承销。 |
| 置信度 | 中 | 产品、资本获取能力和客户验证都真实存在;收入、留存和利润率数据缺失也同样真实。 | 估值区间要留足宽度,避免给出单点精度。 |
| 风险评级 | 高 | 支持、集中度、供应商依赖和信任缺口可能同时传导到估值倍数压缩。 | 仓位要谨慎,愿意付高价前必须拿到尽调权利。 |
| 估值立场 | $1.5B 估值偏紧 | Series C 估值有其合理性,但公开经济性还不足以完全支撑。 | 更偏好相对上一轮折价,或看到更强的私下证据。 |
| 观点改变条件 | 披露 ARR、毛利率、留存,以及高端工作流背后分散的大客户使用情况。 | 这些指标会证明 You.com 不只是承受价格压力的检索层。 | 只有当估值支撑来自经济性而非叙事本身时,才上调判断。 |
| 最可能的下一检查点 | 新一轮私募融资、战略交易,或更丰富的二级市场定价 | 目前公开层面还看不到 IPO 式披露。 | 把融资和披露质量作为下一次市场检验。 |
这份摘要有意把价格和证据放在核心位置:公司质量本身不足以支撑当前估值。
[CV001, CV013, CV039, CV040, CV041, CV042]| 支柱 | 多头论点 | 反论点 | 观点改变条件 |
|---|---|---|---|
| 产品表面 | You.com 现在销售搜索、研究、金融和企业工作流产品,而不再只是单一的消费者搜索工具。 | 仅有产品宽度,不能证明更高价值 SKU 已成为经济重心。 | 按产品线展示收入结构和贡献利润率。 |
| 客户验证 | NIH、Maryville、DPA、HanseMerkur、Mimecast 和具名 API 客户显示,它切入了受监管且知识密集的工作流。 | 公开客户证据经过筛选,没有揭示账户集中度、合同规模或续约情况。 | 提供头部客户结构、合同规模和 cohort 留存数据。 |
| 融资信号 | $100M Series C、$1.5B 估值说明市场仍愿意为这个故事出资。 | 2025 年 6 月约 $1.4B 的融资信号和不透明轮次条款,都反对盲目沿用该估值。 | 披露 term sheet 逻辑、清算优先权结构和当前二级市场价格。 |
| 竞争姿态 | 模型无关定位和企业控制能力,让 You.com 在报告提到的研究工作流中有真实切口。 | 直接搜索价格正在压缩,更大平台也能把检索打包进更广的套件。 | 证明高端工作流能把实现价格推到商品化检索之上。 |
| 估值方法 | 当收入、利润率和留存仍属私有信息时,情景分析能保留不确定性。 | 单一可比倍数会制造公开记录并不支持的精确感。 | 释放足够财务数据,支撑更紧的估值框架。 |
每个多头观点都配上对应证据;只有这些证据变得可见,论点才配得上更紧的倍数。
[CV003, CV010, CV011, CV012, CV017, CV030]从牵引力和市场背景出发,经由披露与竞争,推导到继续研究建议的链条。
这是逻辑链,不是加权模型;它展示哪些公开事实驱动建议,以及缺失证据在哪里打断了信心。
[CV001, CV010, CV013, CV017, CV033, CV034]8.2 融资背景、估值纪律,以及公开标记真正说明什么
融资记录现在给了投资人一个清晰标题,但没有给出清晰承销文件。You.com 可见融资合计约 $195 million,包括 2021 年 $20 million 融资、2022 年 $25 million 轮、2024 年 $50 million Series B、2025 年 $100 million Series C。这条序列重要,是因为它说明公司在战略转向中仍能持续吸引资本。但估值纪律也要求看成交前一刻的信号。Verdict 2025 年 6 月报道,公司正在寻求以约 $1.4 billion 估值融资,随后才有官方宣布的 $1.5 billion Series C。这个差距不足以称为 distress,但确实说明最终轮次应被视为一次谈判形成的市场事件,而不是突破性经济性的自明证明。两个 2025 年 Form D 载体又增加了一层细节:它们显示轮次周围有融资生态支持,但不必然等同于运营公司直接进账,或干净的普通股条款。因此,投资人应把 $1.5 billion 标记当作真实锚点,但不能把它当成稀释、runway 或优先权结构问题的充分答案。[CV001, CV002, CV003, CV004, CV025, CV027]
8.3 多头、基准和空头情景
You.com 不披露收入、毛利率或留存,因此用情景法比给单点估值更诚实。空头情景假设,大部分经济价值落在日益商品化的检索里,而不是有防御力的工作流软件里。在这个世界里,公司仍有真实技术和客户相关性,但市场会折价看待不透明集中度、支持负担和供应商依赖;这指向约 $0.8 billion 到 $1.2 billion 的区间。基准情景假设 Series C 标记大体守住,因为企业工作流、金融研究和私有数据部署都是真实的,但公开经济性仍太薄,不足以支撑 Glean 式溢价。这导向约 $1.2 billion 到 $1.8 billion 的区间。多头情景需要远多于今天的公开故事:私下尽调必须证明强 ARR、耐久留存、健康毛利率,以及查询量声明背后分散的大客户。只有到那时,约 $1.8 billion 到 $2.6 billion 的区间才说得通。换句话说,当前轮次已经计入了有意义的上行;公开证据主要支持基准情景,而不是多头情景。[CV013, CV017, CV019, CV020, CV032, CV033]
| 情景 | 概率信号 | 估值区间 | 关键假设 | 主要下行触发因素 |
|---|---|---|---|---|
| 空头 | 25% | 0.8B-1.2B | 基础检索更像商品而不是软件;集中度和供应商依赖仍不透明;高端 SKU 未能抵消价格压缩。 | 平轮 / 下调轮或疲弱利润率证据,迫使市场给 2025 年估值打折。 |
| 基准 | 55% | 1.2B-1.8B | Series C 大体站得住,因为企业工作流和研究产品真实存在,但公开经济性太薄,无法支撑更高溢价。 | 留存、集中度或支持证据削弱耐久性。 |
| 多头 | 20% | 1.8B-2.6B | 私下尽调显示 ARR 强劲、毛利率健康、大客户足够分散,高端工作流变现改变了实现价格。 | 管理层无法证明高端工作流显著改善经济性。 |
| 概率加权中心 | 100% | ~1.35B-1.7B | 当前公开记录支持接近上一轮的估值,但不支持明显再上一个台阶。 | 没有新证据却大幅高于上一轮付款,会抹掉安全边际。 |
区间刻意拉宽,因为公司尚未披露收入、利润率或留存输入项,无法建立更紧的估值模型。
[CV032, CV033, CV035, CV036, CV037, CV038]以证据质量为中心,而不是围绕单一可比倍数,给出示意性的悲观、基准和乐观估值区间。
这些是情景区间,不是按模型估值输出。由于公开收入、利润率和留存输入不可得,区间刻意拉宽。
[CV035, CV036, CV037, CV038, CV039, CV040]8.4 可比公司组合与市场锚点
最干净的可比教训不是某一个倍数,而是不同证据层级能拿到什么定价。Glean 是这组来源里最有用的私人参照,因为它把高得多的 $7.2 billion 估值、披露的 $100 million 以上 ARR,以及清晰的企业平台故事放在一起。公开软件公司则框定了经济性可见后市场奖励的外沿。Elastic 是一家与搜索和基础设施相邻的上市公司,2026 年 6 月市值约 $6.3 billion、约 3.6x sales;Datadog 和 Snowflake 的收入倍数和市值则高得多。重点不是 You.com 应该接近这些龙头,而是公开市场把这些溢价留给披露收入、增长和利润率质量的公司。下行情景也适用同一原则。搜索 API 同行发布透明价格,hyperscalers 又把检索打包进更广的采购关系;因此,未披露经济质量的公司不应默认拿到高端软件可比逻辑的全部好处。可比公司因此支持谨慎,而不是机械下调或上调。[CV017, CV019, CV020, CV021, CV022, CV023]
| 可比对象 | 指标或状态 | 价值 / 估值信号 | 与 You.com 的相关性 | 局限 |
|---|---|---|---|---|
| You.com Series C 融资 | 私募融资轮 | 融资 100M,估值 1.5B | 公司自身最直接的硬市场锚。 | 未披露 ARR、利润率、留存或清算优先权条款。 |
| Glean Series F | 私募融资加已披露 ARR 信号 | 融资 150M,估值 7.2B;公司称 ARR 已超过 100M | 展示更强企业披露在私募 AI 软件中能拿到什么定价。 | 产品组合不同,公开 ARR 证据也远强于 You.com。 |
| Elastic | 公开市值和销售倍数 | StockAnalysis 显示约 6.3B 市值;约 3.6x 销售额 | 搜索和基础设施相邻的公开可比公司,有真实财务披露。 | 增长较低的公开市场业务,成熟度画像不同。 |
| Datadog | 公开市值和销售倍数 | StockAnalysis 显示约 81.8B 市值;约 22.3x 销售额 | 说明公开市场会给高增长、披露充分的软件基础设施什么溢价。 | 规模大得多,披露也比 You.com 多得多。 |
| Snowflake | 公开市值和销售倍数 | StockAnalysis 显示约 80.7B 市值;约 16.0x 销售额 | 给出第二个高质量公开软件锚,并带有明确估值比率。 | 数据云经济性和成熟度与搜索及研究基础设施差异很大。 |
这些参考用于设定区间和约束证据纪律,不是机械套用倍数来标价。
[CV001, CV019, CV021, CV022, CV023, CV024]将 You.com 上一轮融资与选定私有和公开参考点对比,给出方向性的估值背景。
数值为四舍五入后的公开或公司宣布估值,单位为十亿美元。这张图只提供背景,并不意味着上市公司倍数可直接套用到 You.com。
[CV001, CV019, CV021, CV022, CV023, CV024]8.5 退出准备度与击穿 thesis 的触发因素
公开证据不支持 IPO-ready 判断。You.com 具备一家可信后期私营公司的要素——真实融资、产品广度、客户 logo,以及连贯的企业 AI 叙事——但还没有公开市场通常要求的披露画像。这意味着下一次估值检查点更可能是另一轮私募融资、战略交易,或明显更活跃的二级市场,而不是近期上市。处理这种不确定性的务实方式,是盯住会快速改变估值框架的否决触发因素。平轮或 down round 会告诉投资人,2025 年标记并不耐久。如果高端研究和金融产品无法证明它们实质改变实际定价或利润率,多头情景会坍回商品化检索解读。同样,重大信任或支持故障,或发现少数客户驱动大部分用量,都会降低查询规模叙事的价值。这些不是抽象风险;它们是会迫使重估的具体事件。[CV014, CV015, CV033, CV036, CV042, CV043]
| 触发因素 | 阈值 / 事件 | 对论点的传导 | 行动含义 |
|---|---|---|---|
| 平轮或下调轮 | 下一轮融资在更苛刻条款下,价格等于或低于 2025 年估值 | 证明投资者深入经济性后,当前叙事溢价没能守住。 | 把估值框架推向空头区间,并审查清算优先权悬挂。 |
| 高端 SKU 变现不及预期 | 管理层无法证明研究、金融或企业工作流显著改善实现价格或利润率 | You.com 正从商品化检索毕业的判断会坍塌。 | 在产品经济性得到证明前,移除多头情景权重。 |
| 集中度意外 | 少数 API 客户贡献了不成比例的使用量或收入 | 使每月 >1B 查询量的说法不如表面上有价值,并抬高续约风险。 | 要求更大折扣,并重新评估耐久性假设。 |
| 信任或支持失灵 | 账单、支持、隐私或可靠性问题在尽调或客户访谈中反复出现,形成模式 | 在高端定价最需要信任的位置,推高流失和声誉风险。 | 暂停承诺,直到运营指标和客户访谈改善。 |
| 供应商或平台冲击 | 主要上游模型或平台变化恶化成本、质量或路由灵活性 | 挤压利润率,并削弱模型无关切口。 | 用更低实现利润率和更高运营风险重算情景。 |
| 披露继续停滞 | 管理层在下一轮尽调中仍不披露 ARR、利润率、留存和 runway | 估值继续锚在叙事,而不是经营证据。 | 维持继续研究立场,不要高于保守区间付款。 |
每个触发因素都可监测,并直接关联基准情景是否仍能成立。
[CV014, CV015, CV016, CV033, CV041, CV043]8.6 建议与最终尽调清单
正确投资姿态是继续研究,置信度中等,并且认为公开 $1.5 billion 标记下估值偏紧。这不是说 You.com 没有前景,而是说公开记录对产品相关性的支持远强于对耐久经济性的支持。如果投资人能以显著低于上一轮的价格买入,或管理层能私下证明强 ARR、健康毛利率、分散的大客户,以及高粘性的高端工作流采用,姿态可以转向 fair,最终再转向 track。没有这些证明,为故事支付溢价就是承销希望,而不是承销证据。最终尽调清单直接且杠杆很高:收入组合、产品毛利率、集中度和留存、cap table 条款、现金 runway。如果这五块都强,公司可能值得高于当前公开证据所允许的估值。如果它们偏弱或仍被 withheld,上一轮就应被视为叙事上限锚点,而不是可靠的当前价值。[CV013, CV034, CV039, CV040, CV041, CV042]
| 主题 | 缺失证据 | 重要性 | 尽调路径 |
|---|---|---|---|
| 收入和 ARR 结构 | 按搜索、内容、研究、金融和企业解决方案拆分的过去十二个月收入或 ARR | 需要区分工作流软件业务和低价值检索层。 | 索取月度收入桥和产品级 bookings / ARR 明细表。 |
| 按产品拆分的毛利率 | 按 SKU 拆分的毛利率、云成本、数据授权成本和支持负担 | 判断高端工作流是否真的创造软件式经济性。 | 审阅产品 P&L 和服务成本分析。 |
| 留存和集中度 | 头部客户占比、GRR、NRR、合同期限,以及按账户拆分的工作负载结构 | 没有耐久性和分散化,每月 >1B 查询量这个标题还不够。 | 索取前 10 大客户明细、cohort 表和续约历史。 |
| 股权结构和清算优先权条款 | Series C 周围的清算优先权、二级交易、按比例认购权和任何结构化条款 | 标题估值不等于普通股吸引力。 | 索取当前 cap table、term sheets 和董事会批准的融资备忘录。 |
| 现金和 runway | 当前现金余额、月度 burn、招聘计划,以及基准 / 下行情景 runway | 区分可选增长资本和未来融资依赖。 | 审阅现金桥和 24 个月运营计划。 |
| 运营验证 | 支持指标、正常运行时间或事故历史,以及聚焦部署稳定性的客户访谈 | 信任和支持质量同时影响留存和支付高端价格的意愿。 | 索取 SLA dashboards、事故日志,并与规模化客户做 reference calls。 |
这些要求按最能改变估值置信度的程度排序,而不是按最容易拿到的程度排序。
[CV004, CV013, CV041, CV042, CV043, CV044]仅基于公开证据,对市场、证明、经济性、风险和估值支撑做 IC 风格评分。
分数是定性且受证据约束的;它们用于概括信心,而不是替代财务模型。
[CV013, CV017, CV026, CV034, CV039, CV040]免责声明
本报告是基于公开证据的尽调快照,不构成投资建议。重要的财务、法律、技术和合同事实仍未公开;任何投资决策前,应直接向管理层和原始文件核验。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | You.com was founded in 2020 by Richard Socher and Bryan McCann. | 高 | SO007, SO009, SO011 |
| CO002 | The operating entity is SuSea, Inc. d/b/a you.com. | 高 | SO003, SO004 |
| CO003 | Public legal notices list 228 Hamilton Ave, Floor 3, Palo Alto, California as the company notice address. | 高 | SO003, SO004 |
| CO004 | Current official messaging positions You.com as enterprise AI infrastructure built to deliver accurate, fresh, and actionable intelligence. | 高 | SO002, SO017, SO027 |
| CO005 | The company now sells both developer-facing web-search infrastructure and customized enterprise AI solutions. | 高 | SO001, SO027, SO028 |
| CO006 | The current About page still lists Bryan McCann as CTO and co-founder. | 中 | SO002 |
| CO007 | A March 2026 company announcement says Saahil Jain became CTO while Bryan McCann moved into an advisory role and stayed on the board. | 中 | SO006 |
| CO008 | Saahil Jain joined in 2020 immediately after incorporation and rose from founding engineer to head of AI before being promoted to CTO. | 中 | SO006 |
| CO009 | Richard Socher previously served as Salesforce chief scientist and EVP and is publicly presented as a leading NLP researcher. | 高 | SO005, SO009 |
| CO010 | Socher and McCann publicly tie their founder relationship to joint NLP research that included the 2018 Natural Language Decathlon paper. | 高 | SO007, SO015 |
| CO011 | Founder pedigree gives You.com unusual credibility in search, retrieval, and LLM workflow design relative to a generic application startup. | 中 | SO005, SO009, SO015 |
| CO012 | Series C materials say David Yang from Socium Ventures joined the board as part of the 2025 financing. | 高 | SO001, SO029, SO030 |
| CO013 | TechCrunch reported a $20 million seed round in November 2021 led by Marc Benioff's Time Ventures. | 中 | SO010 |
| CO014 | TechCrunch reported a $25 million round in July 2022 led by Radical Ventures, bringing total disclosed funding to $45 million. | 高 | SO011, SO014 |
| CO015 | You.com's September 2024 Series B announcement said Georgian led a $50 million round that brought total funding to $99 million. | 高 | SO008, SO013, SO018 |
| CO016 | Independent and official 2025 coverage agrees that You.com raised $100 million of Series C funding at a $1.5 billion valuation led by Cox Enterprises. | 高 | SO001, SO029, SO030 |
| CO017 | Adding the publicly disclosed 2021, 2022, 2024, and 2025 rounds yields about $195 million of visible funding, below the higher ~$222 million figure sometimes repeated elsewhere. | 中 | SO010, SO011, SO008, SO001 |
| CO018 | DuckDuckGo appears in the 2024 investor list and also appears in the 2025 customer list, giving the relationship both financing and distribution dimensions. | 中 | SO008, SO001, SO022 |
| CO019 | Series C materials state that You.com processes more than 1 billion API queries per month. | 高 | SO001, SO029 |
| CO020 | The 2024 Series B post claimed You.com had already served 1 billion queries cumulatively and grown ARR by 500% since January 2024. | 中 | SO008, SO013 |
| CO021 | The company started as a consumer search engine but by 2024 was openly calling itself a productivity engine for knowledge workers. | 中 | SO009, SO012, SO013, SO008 |
| CO022 | You.com says its API business emerged after it connected LLMs to live web data with verifiable citations in late 2022. | 中 | SO001, SO007 |
| CO023 | Current enterprise product pages promise model routing, access to private enterprise data, public web data, and custom agent workflows. | 中 | SO027, SO001 |
| CO024 | The public product surface now spans APIs, private RAG, deep research, and custom enterprise builds rather than only a consumer answer engine. | 中 | SO001, SO027, SO028 |
| CO025 | Named customers and partners in public materials include DuckDuckGo, Harvey, NIH, Maryville University, Mimecast, and dpa. | 中 | SO001, SO018, SO019, SO020, SO021 |
| CO026 | Those public references span search, legal AI, public-sector research, higher education, cybersecurity, and media workflows. | 中 | SO018, SO019, SO020, SO021, SO023 |
| CO027 | An interview article on You.com's site claims OpenAI, Amazon, and Alibaba use You.com APIs for up-to-date answers. | 中 | SO007 |
| CO028 | The Harvey homepage corroborates that one named customer category is legal and professional-services AI rather than consumer search. | 中 | SO001, SO023 |
| CO029 | The public customer mix suggests You.com now serves multiple institution types rather than one narrow AI-developer niche. | 中 | SO018, SO019, SO020, SO021, SO023 |
| CO030 | The careers page frames the company as AI building-block infrastructure and highlights offices in San Francisco and New York City. | 中 | SO017 |
| CO031 | Public materials therefore show a multi-site footprint: Palo Alto in legal notices and company datelines, plus San Francisco and New York in recruiting pages. | 中 | SO003, SO004, SO017, SO018 |
| CO032 | TechCrunch's 2024 coverage says it would be futile for You.com to compete with Google on simple queries, which helps explain the move toward harder enterprise research tasks. | 中 | SO013 |
| CO033 | The same TechCrunch piece calls You.com consistently innovative yet often overlooked, signaling that distribution power remains less proven than product quality. | 中 | SO013 |
| CO034 | Official public surfaces lagged the March 2026 CTO handover, creating a mild governance and communications inconsistency. | 中 | SO002, SO006 |
| CO035 | Current valuation and scale disclosures rely heavily on company-authored materials even when financing amounts are independently corroborated. | 中 | SO001, SO008, SO013, SO029, SO030 |
| CO036 | Richard Socher remains the company's dominant public voice across financing, recruiting, mission, and media, creating ongoing key-person concentration. | 中 | SO001, SO005, SO016, SO017 |
| CO037 | Customer case studies emphasize workflow productivity and research automation rather than generic search traffic acquisition. | 中 | SO019, SO020, SO021 |
| CO038 | You.com publicly offers both self-serve pricing and custom enterprise engagements, implying a mixed go-to-market model. | 中 | SO027, SO028 |
| CO039 | Current employee headcount is not disclosed in the public sources reviewed for this chapter. | 低 | |
| CO040 | The chapter sources do not disclose whether any financing round included secondaries, debt facilities, or investor observer rights. | 低 | |
| CM001 | AI search is defined by natural-language understanding plus synthesized answers with citations rather than a ranked list of blue links. | 高 | SM001, SM011 |
| CM002 | Retrieval-augmented generation combines retrieval with generation so answers can be grounded in external sources instead of only model parameters. | 高 | SM001, SM022, SM023 |
| CM003 | Enterprise search covers structured and unstructured company data and relies on indexing, querying, and authorization-aware result delivery. | 高 | SM007, SM009, SM012 |
| CM004 | Agentic search extends one-shot retrieval by letting models plan searches, open pages, inspect page contents, and keep searching inside a workflow. | 中 | SM011, SM015 |
| CM005 | You.com's relevant market is best framed as AI search infrastructure for developers and enterprises rather than the broad consumer search advertising market. | 中 | SM002, SM003, SM024 |
| CM006 | Included spend in You.com's addressable category includes web search APIs, answer-generation endpoints, extraction, private-data grounding, and vertical research workflows. | 中 | SM002, SM006, SM009, SM010, SM015 |
| CM007 | Excluded spend includes generic model-training budgets, broad consumer search ad budgets, and software categories that do not provide search or grounding infrastructure. | 中 | SM002, SM007, SM019 |
| CM008 | Status-quo substitutes include legacy keyword or federated enterprise search, DIY RAG stacks built on model APIs, and rival API-first search providers. | 中 | SM007, SM009, SM010, SM011, SM015, SM018 |
| CM009 | No single public TAM cleanly maps to You.com because available evidence mixes enterprise AI budgets, enterprise search software, and usage-based search API pricing. | 中 | SM006, SM007, SM019, SM020 |
| CM010 | Andreessen Horowitz reported average enterprise spend of $7 million across foundation-model APIs, self-hosting, and fine-tuning in 2023. | 中 | SM019 |
| CM011 | The same a16z survey said enterprises planned to increase that generative-AI spend by roughly 2x to 5x in 2024. | 中 | SM019 |
| CM012 | a16z found that fewer than a quarter of surveyed enterprises expected generative-AI spend to come from innovation budgets in 2024. | 中 | SM019 |
| CM013 | a16z reported that 46% of surveyed enterprise leaders preferred or strongly preferred open-source models going into 2024 and many were targeting a roughly 50-50 closed versus open mix. | 中 | SM019 |
| CM014 | a16z said most enterprises were customizing models through RAG or fine-tuning rather than training large language models from scratch. | 中 | SM019, SM023 |
| CM015 | BVP said foundation-model companies raised $23 billion at more than $124 billion of aggregate market capitalization in 2023, underscoring the capital intensity of the foundational layer. | 中 | SM020 |
| CM016 | BVP said vertical-AI upstarts were already commanding about 80% of the annual contract value of legacy vertical SaaS systems while growing around 400%. | 中 | SM020 |
| CM017 | Microsoft's 2026 Work Trend Index frames AI agents as taking on execution, which supports a broad demand backdrop for agent-enabled knowledge workflows. | 中 | SM021 |
| CM018 | Enterprise buyers consistently require permission-aware retrieval across many internal systems rather than a web-only answer engine. | 中 | SM007, SM013, SM014 |
| CM019 | Glean presents permissioning, governance, explainability, and connector coverage as first-order purchase criteria for enterprise work-AI deployments. | 中 | SM013, SM014 |
| CM020 | Azure AI Search supports vector, keyword, and hybrid retrieval and can connect to SharePoint, the public web, Azure Blob storage, and other knowledge sources. | 中 | SM009 |
| CM021 | Google Agent Search is sold both as better website search and as grounding for generative-AI applications built on enterprise data. | 中 | SM010 |
| CM022 | OpenAI positions web search and deep research as tools that developers embed into Responses API workflows for up-to-date answers and longer-running reports. | 中 | SM011 |
| CM023 | Exa, Tavily, and Brave all position search products as building blocks for agents, chatbots, coding assistants, and automated research tasks. | 中 | SM015, SM017, SM018 |
| CM024 | You.com's price card starts with $5 per 1,000 Web Search API calls and extends to $12 per 1,000 answer calls and $110 per 1,000 finance-research calls. | 中 | SM006 |
| CM025 | Exa lists base search at $7 per 1,000 requests, deep search at $12, and deep-reasoning search at $15, alongside agent modes priced from $0.025 per request. | 中 | SM016 |
| CM026 | Brave lists answer generation at $4 per 1,000 requests plus token fees and advertises 50 queries per second on its search API plan. | 中 | SM018 |
| CM027 | Across vendor materials, modern search APIs are expected to return structured, context-rich outputs fast enough for conversational and autonomous systems. | 中 | SM003, SM011, SM018 |
| CM028 | Fresh citations and page-level access are increasingly core product requirements because agent workflows need verification and follow-up reasoning, not only initial retrieval. | 中 | SM001, SM011, SM015 |
| CM029 | The underlying value proposition for AI search is to reduce hallucination, stale knowledge, and unverifiable reasoning in knowledge-intensive tasks. | 高 | SM022, SM023, SM011 |
| CM030 | You.com's benchmark materials claim 77.84% SimpleQA answer accuracy, a FreshQA F1 of 0.44, and 445 millisecond p50 latency, indicating a speed-plus-freshness positioning. | 中 | SM003, SM004 |
| CM031 | You.com says its API processes more than 1 billion queries per month with 99.9% uptime. | 高 | SM001, SM024, SM025 |
| CM032 | The category is crowded across hyperscalers, enterprise-work AI platforms, and API-first search vendors rather than being owned by one product shape. | 中 | SM009, SM010, SM012, SM013, SM015, SM017, SM018 |
| CM033 | Enterprise adoption is constrained by governance, privacy, and permissioning requirements that are explicitly highlighted across Glean, Google, and Brave materials. | 中 | SM013, SM018, SM010 |
| CM034 | Connectors and ingestion remain a material deployment bottleneck because vendors compete on how much ETL, indexing, and source setup they can hide from customers. | 中 | SM007, SM009, SM010, SM014 |
| CM035 | DIY RAG remains operationally complex because it requires retrieval design, indexing, chunking, embeddings, and specialized implementation talent. | 中 | SM010, SM019, SM023 |
| CM036 | Model optionality matters in vendor selection because enterprises increasingly want multi-model deployment and do not want search infrastructure tied to one model provider. | 中 | SM001, SM019 |
| CM037 | Budget owners are evaluating AI search largely through productivity, accuracy, and cost-to-serve outcomes rather than model novelty alone. | 中 | SM019, SM013, SM021 |
| CM038 | Base web-search pricing clusters in the single-digit-dollar range per 1,000 requests, which implies that value capture increasingly shifts toward higher-level answer, research, and workflow products. | 中 | SM006, SM016, SM018 |
| CM039 | Public evidence suggests You.com's strongest current wedge is external web grounding for agents rather than the broadest internal-enterprise knowledge-platform category. | 中 | SM001, SM003, SM013, SM014, SM024 |
| CM040 | You.com appears to sit between consumer-origin answer-engine UX, developer API distribution, and enterprise-oriented vertical research or search infrastructure. | 中 | SM002, SM005, SM006, SM024 |
| CM041 | Because You.com sells both low-priced search calls and higher-priced answer or research endpoints, its revenue upside likely depends on moving customers up the workflow stack. | 中 | SM006, SM015, SM016 |
| CM042 | Public sources still do not reveal a clean You.com SAM, customer concentration, seat expansion, or gross-margin profile, so market underwriting remains proxy-based. | 低 | |
| CP001 | You.com publicly prices its base web-search API at $5 per 1,000 calls. | 中 | SP001 |
| CP002 | You.com's higher-order products are priced materially above base search, with research at $12 per 1,000 requests and finance research at $110 per 1,000 requests. | 中 | SP001, SP006 |
| CP003 | You.com positions itself as a model-agnostic enterprise platform that mixes private and public data while emphasizing trust and security controls. | 中 | SP002, SP004 |
| CP004 | Independent reporting in 2024 said You.com had shifted from trying to beat Google on simple queries toward deeper productivity agents. | 中 | SP003 |
| CP005 | The most direct current peer group for You.com consists of API-first search and answer vendors such as Exa, Tavily, Brave, Perplexity, and OpenAI's web-search tool stack. | 中 | SP013, SP015, SP017, SP018, SP020 |
| CP006 | Exa exposes search, contents, answer, research, and agent workflows and publicly lists search pricing from $7 per 1,000 requests up to $15 per 1,000 deep-reasoning requests. | 高 | SP013, SP014 |
| CP007 | Tavily sells a broader web-tool workflow than plain search by exposing search, extract, map, crawl, and research APIs inside a credit-based pricing model. | 中 | SP015, SP016 |
| CP008 | Brave Search API combines its own web index, answer-generation endpoints, enterprise privacy positioning, and visible $4 per 1,000 answer pricing. | 中 | SP017 |
| CP009 | Perplexity's official docs price raw search at $5 per 1,000 requests while layering additional tool, Sonar, and deep-research pricing on top. | 中 | SP020 |
| CP010 | OpenAI packages web search as a built-in tool inside the Responses API rather than as a standalone enterprise-search product surface. | 中 | SP018 |
| CP011 | OpenAI's official pricing page lists web search at $10 per 1,000 calls on top of model-token pricing. | 高 | SP018, SP019 |
| CP012 | Glean competes from the enterprise retrieval side with permission-aware search, large connector breadth, embedded workflow surfaces, and per-user Enterprise Flex packaging. | 高 | SP008, SP011, SP012 |
| CP013 | Glean's public surfaces claim 100-plus connectors and enterprise AI experiences that span search, agents, and embedded work surfaces. | 高 | SP009, SP011, SP012 |
| CP014 | Glean's official Enterprise Flex documentation bundles unlimited fast-mode queries and up to 100 standard thinking queries per user per week before extra credit consumption. | 中 | SP012 |
| CP015 | Glean's strongest public advantage is distribution through enterprise systems and user seats rather than visible superiority on open-web API economics. | 中 | SP008, SP011, SP012 |
| CP016 | Azure AI Search competes as enterprise retrieval infrastructure with hybrid search, vector-store behavior, knowledge-base connections, and tiered pricing modes. | 高 | SP021, SP022 |
| CP017 | AWS Kendra competes as a managed internal-search service whose public pricing is built from hourly base index, storage, query, and connector charges. | 高 | SP023, SP024 |
| CP018 | Google Agent Search sells enterprise retrieval over website, structured, and unstructured data with query-plus-storage pricing and built-in generative-answer tiers. | 高 | SP025, SP026 |
| CP019 | Elastic's public positioning is as enterprise-search and vector-search infrastructure rather than as a turnkey external answer engine. | 高 | SP027, SP028 |
| CP020 | Harvey is best viewed as a vertical workflow substitute because it sells secure AI for legal and professional services instead of a horizontal search layer. | 中 | SP029 |
| CP021 | Harvey's existence shows that horizontal search and research vendors can be displaced by domain-specific tools that own a valuable professional workflow. | 中 | SP029 |
| CP022 | DuckDuckGo's public site claims 3 billion monthly searches and 9 million monthly downloads, demonstrating materially stronger consumer distribution than You.com publicly discloses. | 中 | SP030 |
| CP023 | DuckDuckGo's consumer-distribution scale matters more as evidence of search-channel power than as proof that it is building the same enterprise agent platform as You.com. | 中 | SP030 |
| CP024 | Internal build is a credible substitute because gpt-oss can run locally or on owned infrastructure and can attach to search and tool surfaces instead of requiring a single vendor platform. | 高 | SP018, SP019, SP031, SP032 |
| CP025 | OpenAI positions gpt-oss as customizable open-weight models with native tool use and lightweight deployment options, lowering the barrier to building tailored in-house agents. | 高 | SP031, SP032 |
| CP026 | AWS Marketplace featuring Brave Search API as a top listing is a visible channel advantage for Brave and an example of ecosystem distribution that You.com does not publicly match. | 高 | SP017, SP033 |
| CP027 | You.com's 2025 benchmark report is useful as company positioning but cannot serve as independent proof of durable superiority over peers. | 中 | SP034 |
| CP028 | ARI and the Finance Research API show that You.com is intentionally moving up-stack from search plumbing into packaged research workflows. | 高 | SP005, SP006 |
| CP029 | ARI claims to analyze up to 400 sources and combine public with private data for enterprise research outputs. | 中 | SP005 |
| CP030 | You.com's Finance Research API release claims higher benchmark accuracy than Perplexity, Exa, and Tavily on historical financial lookups, but the comparison is still company-authored. | 中 | SP006, SP007 |
| CP031 | The most direct visible base-search price cluster in this chapter is roughly $4 to $7 per 1,000 requests across Brave, You.com, Exa, and Perplexity Search API. | 高 | SP001, SP014, SP017, SP020 |
| CP032 | Research-grade workflows are priced materially above basic search across You.com, Exa, and Perplexity, implying that retrieval alone is becoming commoditized while packaged research is where vendors seek margin. | 高 | SP001, SP006, SP014, SP020 |
| CP033 | Glean and the cloud incumbents have structural distribution advantages because they sit closer to enterprise data, identity, embedded workflow surfaces, and standard procurement paths. | 中 | SP011, SP012, SP021, SP023, SP025 |
| CP034 | You.com's moat is strongest where buyers need fresh web grounding, citations, model choice, and configurable enterprise workflows in one product. | 中 | SP002, SP005, SP006 |
| CP035 | Public sources do not reveal You.com's realized enterprise pricing, direct win rates against Glean or hyperscalers, or channel revenue concentration. | 低 | |
| CP036 | Google Agent Search and Azure AI Search both explicitly market grounded answers and citations on enterprise data, so grounded-answer messaging alone is not a defensible moat. | 中 | SP021, SP025 |
| CP037 | AWS Kendra and Google Agent Search are easier to justify when buyers already accept cloud-style capacity or query metering for internal knowledge retrieval. | 中 | SP024, SP026 |
| CP038 | Glean's public packaging includes embedded experiences across ServiceNow, Zendesk, Service Cloud, Slack, Zoom, Teams, GitHub, and Miro. | 中 | SP012 |
| CP039 | You.com's enterprise page claims connectors to systems such as SharePoint, Databricks, S3, Notion, and Slack, showing an attempt to meet incumbents on private-data breadth. | 中 | SP002 |
| CP040 | The core underwriting risk is three-sided at once: direct API price compression, enterprise distribution power from incumbents, and build-your-own substitution enabled by open models and hosted tools. | 中 | SP017, SP020, SP025, SP031 |
| CP041 | You.com says its API now handles more than 1 billion queries per month, which is an important scale signal but remains company-authored. | 中 | SP035 |
| CI001 | You.com's current official pricing surface centers usage-based API monetization rather than only a chat subscription. | 中 | SI001 |
| CI002 | The Web Search API is listed at $5 per 1,000 calls. | 中 | SI001 |
| CI003 | The Contents API is listed at $1 per 1,000 pages. | 中 | SI001 |
| CI004 | Public pricing text shows a research tier priced at $12 per 1,000 queries. | 中 | SI001 |
| CI005 | The Finance Research API is publicly priced at $110 per 1,000 queries. | 中 | SI001, SI004 |
| CI006 | You.com's pricing page explicitly offers volume discounts and annual savings, so list rates are not the same as realized net price. | 中 | SI001 |
| CI007 | New accounts receive $100 of free credits on the developer platform. | 高 | SI001, SI010 |
| CI008 | You.com offers a free MCP search profile with no signup or credentials and a 100-query daily limit. | 中 | SI010 |
| CI009 | The Series B post said You.com introduced a Team plan and described enterprise pricing as consumption-based. | 中 | SI002 |
| CI010 | Official enterprise and Series C materials show You.com also sells custom agents, private-data integrations, and end-to-end enterprise builds. | 高 | SI003, SI012 |
| CI011 | The September 2025 roundup introduced Express and Contents APIs as new productized endpoints. | 中 | SI007 |
| CI012 | The November 2025 roundup introduced Deep Search early access and broader distribution through MCP, SDKs, Zapier, and Databricks. | 中 | SI008 |
| CI013 | The December 2025 roundup added evals-as-a-service, vertical indexes, and a unified platform home and playground. | 中 | SI009 |
| CI014 | The Finance Research API is augmented with licensed structured data from S&P Global and other providers. | 中 | SI004 |
| CI015 | The finance-research documentation says the product indexes structured company fundamentals, market data, private-company metrics, macro data, SEC filings, and financial news. | 中 | SI011 |
| CI016 | ARI's initial launch said the research agent could analyze up to 400 sources for analyst or consultant-style workflows. | 中 | SI005 |
| CI017 | The later ARI enterprise launch said ARI could process more than 500 sources, added enterprise-only features, and temporarily offered unlimited reports to new enterprise customers. | 中 | SI006 |
| CI018 | Multi-step research products that read hundreds of sources and incorporate licensed datasets are likely more expensive to serve than base search. | 中 | SI004, SI005, SI006, SI011 |
| CI019 | The Series C post claims You.com now processes more than 1 billion API queries per month. | 高 | SI003, SI024, SI025 |
| CI020 | The Series B post claimed You.com had already served 1 billion queries cumulatively and had grown ARR 500% since January 2024. | 中 | SI002 |
| CI021 | The Series C post says more than 100,000 agents were built on You.com in the prior 12 months. | 中 | SI003 |
| CI022 | A You.com customer story says NIH accelerated research 1000x with You.com. | 低 | SI013 |
| CI023 | A You.com customer story says Maryville reduced nursing dropouts by 78% with You.com. | 低 | SI014 |
| CI024 | A You.com customer story says Mimecast used the product to drive GTM productivity and practical AI adoption. | 低 | SI015 |
| CI025 | Official and independent Series C coverage names customers such as DuckDuckGo, Harvey, and Windsurf and cites exposure to multiple industries, but it does not disclose customer concentration or contract sizes. | 中 | SI003, SI024, SI025 |
| CI026 | TechCrunch reported a $20 million round in 2021. | 中 | SI020 |
| CI027 | TechCrunch reported a $25 million round in 2022 led by Radical Ventures. | 中 | SI021 |
| CI028 | You.com's official Series B announcement disclosed $50 million of new funding in 2024. | 中 | SI002 |
| CI029 | You.com's official Series C announcement and independent coverage align on a $100 million round at a $1.5 billion valuation. | 高 | SI003, SI024, SI025 |
| CI030 | The disclosed 20/25/50/100 sequence sums to roughly $195 million of visible company-level capital. | 高 | SI020, SI021, SI002, SI003 |
| CI031 | Verdict reported in June 2025 that You.com was seeking new funding at a $1.4 billion valuation to bolster its AI assistant offerings. | 中 | SI023 |
| CI032 | The SEC Form D filing for GSBackers You.com Fund I reported total offering and sold amounts of $10,012,500. | 中 | SI016 |
| CI033 | The SEC Form D filing for GSBackers You.com Fund II reported total offering and sold amounts of $5,072,500. | 中 | SI017 |
| CI034 | Two August 2025 You.com-named Form D vehicles imply at least $15.085 million of pooled investment capital around the 2025 financing, although the filings do not prove direct proceeds to the operating company. | 高 | SI016, SI017, SI018, SI019 |
| CI035 | TechCrunch wrote in 2024 that it would have been futile for You.com to compete with Google on simple queries, helping explain the shift toward deeper productivity agents. | 中 | SI022 |
| CI036 | Base web-search list prices cluster tightly across peers, with You.com at $5 per 1,000 calls, Brave at $4 per 1,000 requests, and Exa at $7 per 1,000 requests. | 中 | SI001, SI026, SI027 |
| CI037 | OpenAI's API pricing page lists web-search tool calls at $10 per 1,000 calls, above You.com's base list rate even though the product packaging is not identical. | 高 | SI001, SI028 |
| CI038 | Enterprise materials emphasize zero data retention and private-data integration, features that can support higher-value contracts but also add implementation complexity. | 高 | SI006, SI012 |
| CI039 | The Series B announcement stressed customer success and real adoption rather than just proofs of concept, implying a hands-on enterprise motion. | 中 | SI002 |
| CI040 | Docs, SDK, marketplace, and playground investments are designed to lower developer onboarding cost and shorten time to first value. | 高 | SI008, SI009, SI010 |
| CI041 | Reviewed public materials do not disclose current revenue, revenue mix, or revenue-recognition policy across API, Team, and bespoke enterprise work. | 中 | SI001, SI002, SI003, SI012 |
| CI042 | Reviewed public materials do not disclose gross margin, infrastructure cost per query, CAC payback, or net revenue retention. | 中 | SI001, SI004, SI010, SI011 |
| CI043 | Reviewed public materials do not disclose cash on hand, monthly burn, or runway. | 中 | SI002, SI003, SI023, SI024 |
| CI044 | Because You.com sells software and APIs, its capital intensity should be lighter than a hardware startup's, but compute, licensed data, and enterprise deployment work can still absorb meaningful cash. | 中 | SI004, SI006, SI012 |
| CI045 | With roughly $195 million of disclosed company-level capital plus at least $15.085 million of You.com-named SPV filings, capital access looks credible even though runway remains undisclosed. | 高 | SI020, SI021, SI002, SI003, SI016, SI017 |
| CI046 | Revenue quality likely depends on mix shift from commodity search into premium research, finance, and enterprise workflow products. | 中 | SI001, SI004, SI006, SI022 |
| CI047 | Base-layer search price compression makes published list pricing a weak proxy for realized economics or durable margin. | 中 | SI001, SI026, SI027, SI028 |
| CI048 | The official pricing and enterprise pages show no minimums for self-serve experimentation but reserve custom QPS and tailored economics for enterprise buyers. | 高 | SI001, SI012 |
| CI049 | Radical's 2022 portfolio description already framed You.com as an orchestration layer across private data sources, internet search, and LLMs, suggesting the enterprise-intelligence thesis predates the 2025 Series C. | 中 | SI029 |
| CI050 | Unlimited ARI reports for new enterprise customers are a promotional acquisition tactic, not evidence of durable unit economics. | 中 | SI006 |
| CI051 | The absence of customer-count, net-retention, concentration, and contract-size data prevents public usage signals from being translated into dependable recurring-revenue quality. | 中 | SI003, SI024, SI025 |
| CI052 | Public evidence supports a strategic case for premium AI-search infrastructure, but not a clean financial underwriting case on margin path or cash sufficiency. | 中 | SI001, SI003, SI016, SI017 |
| CE001 | You.com's current public API surface includes Search, Contents, and Research endpoints, with broader agentic access exposed through its docs and platform. | 中 | SE001 |
| CE002 | You.com offers a free MCP profile with up to 100 search queries per day and no signup, while paid accounts add credits and access to deeper endpoints. | 中 | SE001 |
| CE003 | The public pricing page lists the Web Search API at $5 per 1,000 calls. | 中 | SE002 |
| CE004 | The public pricing page lists the Contents API at $1 per 1,000 pages. | 中 | SE002 |
| CE005 | The public pricing page lists the Research API at $12 per 1,000 queries and describes it as source-backed with citations and a multi-step search and synthesis pipeline. | 中 | SE002 |
| CE006 | The public pricing and finance-doc pages list Finance Research at $110 per 1,000 deep queries and emphasize cited, traceable financial answers. | 高 | SE002, SE011 |
| CE007 | The enterprise page describes a layered platform with data, reasoning and tooling, agent, and application layers. | 中 | SE003 |
| CE008 | You.com publicly positions its enterprise stack as model-agnostic across GPT, Claude, Gemini, and more. | 中 | SE003 |
| CE009 | You.com publicly says its platform connects to enterprise data sources including Google Drive, SharePoint, Databricks, S3, Notion, and Slack. | 中 | SE003 |
| CE010 | You.com says most enterprise teams can go live in days because it handles infrastructure, private RAG setup, onboarding, and optimization. | 中 | SE003 |
| CE011 | You.com frames agentic search as a shift away from keyword snippets toward structured, context-aware passages that models can reason over directly. | 中 | SE004 |
| CE012 | You.com's AI search infrastructure narrative explicitly spans ready-made Search APIs, vertical indexes, private RAG, custom agents, and tailored indexes. | 中 | SE005 |
| CE013 | You.com describes RAG as a two-step retrieve-then-generate pattern that grounds answers in retrieved sources rather than only in model weights. | 中 | SE006 |
| CE014 | You.com's own AI-search explainer breaks the pipeline into content gathering, indexing, vectorization, retrieval, and RAG synthesis. | 中 | SE006 |
| CE015 | The 2024 YOU API launch post said future work would expose intermediate modules such as query rewriting and broader agent-like capabilities through the API surface. | 中 | SE007 |
| CE016 | The September 2025 release introduced Express API as a one-call grounded answer surface that returns an answer, citations, and underlying web results. | 中 | SE012 |
| CE017 | The September 2025 release introduced Contents API as a way to return clean markdown or HTML plus metadata, including via Search API livecrawl parameters. | 中 | SE012 |
| CE018 | The November 2025 release described Deep Search as retrieving live webpages, extracting answer passages, and verifying that the quoted text really appears on those pages. | 中 | SE013 |
| CE019 | You.com positions Deep Search between generic snippets and full-page extraction so downstream models receive tighter context with less token overhead. | 中 | SE013 |
| CE020 | By late 2025 You.com had expanded its MCP server, Python SDK, Zapier integration, and Databricks marketplace footprint around Search, Content, and Express. | 中 | SE013 |
| CE021 | The December 2025 release added evals-as-a-service, vertical indexes, an API playground, a TypeScript SDK, and integrations with the Vercel AI SDK, OpenAI SDK, and Claude Agent SDK. | 中 | SE014 |
| CE022 | You.com says Finance Research is built on top of the Research API and augments it with licensed structured data from S&P Global and other providers. | 中 | SE010 |
| CE023 | The Finance Research docs say the endpoint searches a finance-optimized index and returns markdown answers with inline citations and a sources array. | 中 | SE011 |
| CE024 | The ARI launch post says ARI analyzes up to 400 sources, combines public and private data, and packages results as professional-grade research outputs. | 中 | SE008 |
| CE025 | The later ARI enterprise update says ARI can process more than 500 sources and now supports custom data integrations plus branded deliverables for enterprise customers. | 中 | SE009 |
| CE026 | You.com has publicly released a GitHub repository with DeepConsult datasets and pairwise evaluation scripts for deep-research benchmarking. | 中 | SE022 |
| CE027 | OpenAI's current web-search docs separate fast search, agentic search, and deep research, reinforcing the same product segmentation You.com is pursuing. | 中 | SE023 |
| CE028 | OpenAI's GPT-OSS materials describe open-weight reasoning models as tool-using systems that can run on user-controlled infrastructure, which raises demand for external grounding providers. | 高 | SE024, SE032 |
| CE029 | You.com says OpenAI integrated You.com as a core search provider for GPT-OSS, giving You external platform validation as a grounding layer for open-weight models. | 中 | SE016 |
| CE030 | Google Agent Search packages out-of-the-box RAG, connectors, citations, privacy controls, and DIY grounding APIs for enterprise retrieval. | 中 | SE025 |
| CE031 | Azure AI Search packages vector, keyword, and hybrid search with knowledge bases and multiple data connections for enterprise RAG use cases. | 中 | SE026 |
| CE032 | Elastic's enterprise-search materials describe crawl, index, query, connectors, security, and authorization as core building blocks of enterprise retrieval systems. | 中 | SE028 |
| CE033 | Exa exposes separate search, contents, answer, and research endpoints, showing that endpoint specialization is now a common design pattern among agent-search vendors. | 中 | SE029 |
| CE034 | Tavily exposes search, extract, crawl, map, and research tasks, reinforcing that adjacent API vendors are also bundling multiple retrieval modes for agents. | 中 | SE030 |
| CE035 | Brave packages search answers, citations, an independent web index, and enterprise zero-data-retention options at the API layer, increasing competitive pressure on baseline agentic search. | 中 | SE031 |
| CE036 | The NIH customer story says You.com met research workflows that required verifiable citations and expanded from a 30-35 user pilot toward 120 licenses. | 中 | SE019 |
| CE037 | The Maryville case study says the school built more than 600 custom agents and saw 76% faculty adoption, using You.com as a multi-model academic workflow platform. | 中 | SE020 |
| CE038 | The Mimecast story links You.com's custom-agent sharing and training support to a reported 20% increase in sales-enablement content. | 中 | SE021 |
| CE039 | You.com's privacy policy says the service uses third-party LLM providers including OpenAI, Anthropic, and Google, hosts content on AWS, and does not use connected third-party data to train third-party AI models. | 中 | SE017 |
| CE040 | You.com's terms say parts of the service depend on third-party APIs remaining available on reasonable terms and that features may be modified, suspended, or discontinued. | 中 | SE018 |
| CE041 | You.com's terms permit automated access via API key or explicit permission but prohibit unapproved scraping, rate-limit circumvention, and security bypasses. | 中 | SE018 |
| CE042 | You.com publicly markets zero-data-retention, SOC 2, DPA-ready contracting, and custom QPS support, but its public materials do not expose detailed SLOs, incident history, or control mappings. | 中 | SE002, SE003 |
| CE043 | You.com's 2025 benchmark report claims 77.84% SimpleQA accuracy, FreshQA F1 of 0.44, and sub-445ms standard-search latency, but the methodology is company-authored. | 中 | SE015 |
| CE044 | Across the benchmark report and finance launch materials, You.com is increasingly segmenting products by query complexity, using fast search for quick grounding and deeper research surfaces for slower multi-step analysis. | 中 | SE010, SE015, SE023 |
| CE045 | TechCrunch's 2024 reporting said competing with Google on easy questions was futile, independently corroborating the product shift toward deeper productivity and research agents. | 中 | SE033 |
| CE046 | AWS Kendra's positioning as a managed enterprise search and RAG service shows that You.com competes not just with web-search APIs but also with incumbent enterprise retrieval platforms. | 中 | SE027 |
| CE047 | You.com’s 2026 API-architecture essay argues that in production the model becomes an orchestration layer while APIs handle live system interaction, latency, and failure management. | 中 | SE034 |
| CE048 | You.com’s tool-design essay says its research stack centers on simple web_search and read_page primitives, goal-conditioned extraction, compact link encoding, and inline citation markers to preserve token efficiency. | 中 | SE035 |
| CE049 | A May 2026 company benchmark claimed that Claude Sonnet 4.6 using You.com search achieved 61% fewer tokens, 60% lower cost, and 40% faster responses than Claude’s built-in web search on 47 completed queries. | 中 | SE037 |
| CE050 | You.com’s context-rot essay identifies stale retrieval data, silent model-behavior drift, and prompt or schema drift as three distinct ways AI integrations can degrade without throwing hard errors. | 中 | SE036 |
| CE051 | The June 2026 MindStudio Remy integration post presents You.com as a compile-in live web layer spanning web search, live news, page-content extraction, research, and financial intelligence. | 中 | SE038 |
| CU001 | You.com is selling both self-serve web/research APIs for developers and higher-touch enterprise AI solutions for organizations. | 高 | SU001, SU002 |
| CU002 | The public customer evidence spans institutional buyers, workflow users, and channel partners rather than a single homogeneous seat-based segment. | 中 | SU003, SU004, SU005, SU012, SU007, SU008, SU009, SU010, SU011 |
| CU003 | Named verticals in public materials include public-health research, higher education, cybersecurity GTM, insurance, health publishing, media/news, developer tools, and AI-native software. | 中 | SU001, SU003, SU004, SU005, SU012, SU013 |
| CU004 | Public proof quality is strongest in detailed workflow case studies and weaker in logo references that do not disclose deployment size, timing, or commercial value. | 中 | SU001, SU003, SU004, SU005 |
| CU005 | NIH says it chose You.com because other evaluated AI tools produced unreliable or inconsistent outputs for research work. | 中 | SU003 |
| CU006 | NIH began with a 30-35 user pilot and expanded the deployment toward 120 licenses across the institution. | 中 | SU003 |
| CU007 | NIH uses You.com for research synthesis, document review, verification support, drafting, and contractor workflows. | 中 | SU003 |
| CU008 | NIH's IT leader estimates that saving 2-3 hours per week across 50-60 contractors creates more than $1 million of annual productivity value. | 中 | SU003 |
| CU009 | Because NIH describes itself as the world's largest biomedical research agency with more than 20,000 researchers and clinicians, a 120-license expansion is meaningful departmental deployment proof rather than casual consumer usage. | 中 | SU003, SU018 |
| CU010 | Maryville says You.com reduced nursing-course withdrawals from 45% to 10% in one semester, a 78% reduction. | 中 | SU004 |
| CU011 | Maryville says 115 additional students stayed on track to graduate and that the deployment generated over $3 million in additional revenue. | 中 | SU004 |
| CU012 | Maryville reports 76% faculty adoption, 600+ custom AI agents, and more than 50% of students using You.com multiple times weekly. | 中 | SU004 |
| CU013 | Maryville's rollout separates the buyer and payer from the end user: the institution and faculty set policy while students consume tutoring and custom agents. | 中 | SU004 |
| CU014 | Mimecast says it selected You.com for multi-model access, shareable custom AI agents, and competitive pricing versus alternatives. | 中 | SU005 |
| CU015 | Mimecast reports a 20% increase in sales-enablement content and faster GTM processes after deployment. | 中 | SU005 |
| CU016 | Mimecast is itself a scaled enterprise software vendor with 42,000+ customers and 300+ integrations, supporting the plausibility that its internal GTM tooling requirements are sophisticated. | 中 | SU005, SU019 |
| CU017 | The dpa partnership gives newsrooms AI-generated discovery and curation workflows and gives dpa access to Team Plan plus custom AI agents. | 中 | SU012 |
| CU018 | The enterprise solutions page adds HanseMerkur and Wort & Bild Verlag as named references, but it does not disclose deployment scale, seat count, or quantified ROI for those logos. | 中 | SU001 |
| CU019 | You.com now distributes through MCP, cloud marketplaces, and framework integrations rather than relying only on direct web-seat sales. | 高 | SU006, SU007 |
| CU020 | The partnerships page claims marketplace availability on AWS, Azure, and Databricks and frames these channels as procurement and deployment surfaces. | 中 | SU006 |
| CU021 | The integrations overview says the same Search, Contents, and Research APIs travel across MCP, SDKs, Claude Skills, and third-party frameworks with one API key. | 高 | SU007, SU010, SU006 |
| CU022 | You.com's GPT OSS integration docs say You.com is the default browser-tool backend for search, open, and find actions in OpenAI's open-weight models. | 高 | SU008, SU024 |
| CU023 | OpenAI's gpt-oss repository explicitly says You.com is the default browser backend and names Exa as an alternative implementation. | 中 | SU024 |
| CU024 | The Zapier integration connects You.com Search, Content Extraction, and News APIs to more than 8,000 apps, broadening reach into no-code and business workflows. | 中 | SU009 |
| CU025 | The Vercel AI SDK plugin packages youSearch, youContents, and youResearch as drop-in tools for application teams already building on Vercel's stack. | 中 | SU010 |
| CU026 | The MindStudio/Remy partnership positions You.com as a compiled-in live web layer for generated apps, shifting some customer acquisition from direct enterprise selling to partner-led embedding. | 中 | SU011, SU027 |
| CU027 | You.com says it now processes over 1 billion API queries monthly and names DuckDuckGo, Windsurf, and Harvey as API customers. | 中 | SU013 |
| CU028 | The DuckDuckGo, Windsurf, and Harvey references imply You.com can serve both end-user products and developer platforms, but public materials do not quantify spend, seats, or contract length for any of those logos. | 中 | SU013, SU020, SU021, SU022 |
| CU029 | Public adoption evidence is much richer for deployment anecdotes and partner breadth than for classical SaaS KPIs such as active customer count, GRR, NRR, or logo churn. | 中 | SU003, SU004, SU005, SU013 |
| CU030 | The clearest repeat-use proxy in public sources is NIH's pilot-to-120-license expansion and Maryville's increase from 38% initial engagement to 76% in follow-up courses. | 中 | SU003, SU004 |
| CU031 | Third-party review surfaces are mixed: G2 shows a 4.4/5 rating from 19 reviews, while Trustpilot shows a 2.4 score from 36 reviews with a strongly negative review mix. | 中 | SU014, SU015 |
| CU032 | G2's extracted pros and cons highlight ease of use and efficiency but also flag expense, cost, slow performance, and limited features. | 中 | SU014 |
| CU033 | Trustpilot complaints center on billing and support issues, team activation failures, provider-imposed model limits, deteriorating output quality, and privacy distrust. | 中 | SU015 |
| CU034 | Because You.com sells multi-model access and research workflows, part of customer satisfaction depends on third-party model availability and cost ceilings that You.com does not fully control. | 中 | SU015, SU026 |
| CU035 | The privacy policy and enterprise page promise zero data retention and say connected third-party system data is not used to train third-party AI models, which materially helps enterprise expansion conversations. | 高 | SU001, SU025 |
| CU036 | The terms say some functionality depends on third-party APIs remaining available on reasonable terms and reserve the right to modify or discontinue features, which adds renewal risk for customers embedding You.com deeply. | 中 | SU026 |
| CU037 | Customer concentration is opaque because You.com does not disclose customer count, top-customer share, ARR per logo, or seat-count distribution. | 中 | SU001, SU013 |
| CU038 | Public procurement proof is thinner than the partnership marketing suggests because the captured AWS Marketplace search page does not itself reveal a named You.com listing or pricing detail in readable text. | 中 | SU006, SU016 |
| CU039 | Proof quality varies sharply by logo: NIH and Maryville show quantified outcomes, Mimecast shows a narrower GTM gain, dpa describes workflow and access, and DuckDuckGo/Windsurf/Harvey are only named in a financing post. | 中 | SU003, SU004, SU005, SU012, SU013 |
| CU040 | Expansion and renewal could be attractive where teams share custom agents and embed You.com into workflows, but the absence of contract length, cohort, or NRR data keeps this as a hypothesis rather than a verified moat. | 中 | SU004, SU005, SU012 |
| CU041 | The enterprise page says deployment is fast and low-lift and that most teams go live in days with white-glove onboarding, implying a consultative deployment motion for larger accounts. | 中 | SU001 |
| CU042 | The customer base likely mixes direct enterprise accounts and embedded API/platform customers, creating concentration risk if a small set of high-query partners drive disproportionate volume. | 中 | SU013, SU006 |
| CU043 | MindStudio, Zapier, Vercel, and GPT OSS integrations reduce top-of-funnel friction but also increase dependence on external ecosystems for distribution and support expectations. | 中 | SU008, SU009, SU010, SU011 |
| CU044 | After deployment proof, the main unresolved diligence gaps are renewal metrics, true concentration, marketplace procurement evidence, and independent verification for several named logos. | 中 | SU014, SU015, SU016, SU013 |
| CR001 | You.com’s privacy policy says the service uses multiple LLM providers, including OpenAI, Anthropic, and Google. | 中 | SR001 |
| CR002 | You.com says data collected through connected third-party services is not used to train third-party AI or ML models. | 中 | SR001 |
| CR003 | The privacy policy tells users not to submit restricted data subject to regimes such as HIPAA, GLBA, or GDPR in prompts or uploads. | 中 | SR001, SR002 |
| CR004 | You.com says cross-border transfers may occur and it remains responsible under Data Privacy Framework onward-transfer principles for third-party agent processing. | 中 | SR001 |
| CR005 | The privacy policy says You.com uses vendors including AWS, Stripe, and Google for parts of its security and operating stack while disclaiming any guarantee of perfect security or privacy. | 中 | SR001 |
| CR006 | The terms place third-party services, outputs, and linked content outside You.com’s control and at the user’s risk. | 中 | SR002 |
| CR007 | The terms say third-party API or program withdrawal on unreasonable terms can lead You.com to stop providing affected functionality and also reserve broad rights to modify, suspend, or discontinue services. | 中 | SR002 |
| CR008 | The terms allow fee increases, additional charges, and generally non-refundable payments except where law requires otherwise. | 中 | SR002 |
| CR009 | The public legal hub lists a DPA template, MSA/SOW, security overview, security addendum, and terms of service. | 中 | SR003 |
| CR010 | The enterprise page claims zero data retention, SOC 2 certification, DPA readiness, and custom QPS limits for enterprise buyers. | 中 | SR004, SR005 |
| CR011 | You.com’s governance kit says uncontrolled enterprise AI adoption raises data-exposure, unreliable-output, and compliance risks and says its templates align with NIST AI RMF, ISO/IEC 42001, and GDPR/CCPA. | 中 | SR008 |
| CR012 | The European Commission says GPAI obligations under the AI Act became applicable in August 2025 and transparency rules take effect in August 2026. | 中 | SR022 |
| CR013 | The European Commission says high-risk AI systems in areas including education, employment, and essential services must meet strict obligations on risk management, logging, documentation, oversight, cybersecurity, and accuracy. | 中 | SR022 |
| CR014 | HanseMerkur says it chose You.com partly to compare multiple models in one secure environment and because citations matter in regulated workflows. | 中 | SR013 |
| CR015 | HanseMerkur describes unmanaged employee experimentation and concern about company data leaving approved channels before adopting a secure, compliant AI platform. | 中 | SR013 |
| CR016 | You.com’s GPT OSS documentation says You.com is the default backend for the browser tool and handles search, page open, and content-find actions under the hood. | 中 | SR023 |
| CR017 | OpenAI’s GPT OSS launch emphasizes open-weight models, flexible deployment, and strong tool use, which reduces switching friction for developers who can choose between self-hosted and platform-attached workflows. | 中 | SR024 |
| CR018 | You.com’s hallucination guide says current architectures cannot completely eliminate hallucinations and still require monitoring and human oversight in high-stakes use cases. | 中 | SR009 |
| CR019 | The same guide says source quality, retrieval accuracy, and citation verification are the key trust variables in production deployments. | 中 | SR009 |
| CR020 | You.com’s latency guide says third-party dependencies, queuing, and rate limiting are major drivers of latency spikes and recommends tracking P95 and P99 instead of averages. | 中 | SR010 |
| CR021 | You.com’s benchmarks page says the Research API is #1 in DeepSearchQA and markets the finance product as cited, source-reconciled intelligence. | 中 | SR007 |
| CR022 | Because the benchmark and latency materials are company-authored, they are directionally useful but do not provide independent proof of reliability or performance. | 中 | SR007, SR010 |
| CR023 | Trustpilot shows 36 reviews for You.com and says the company has replied to 29% of negative reviews. | 中 | SR020 |
| CR024 | A Trustpilot reviewer described paying for a Team migration but being unable to add team members while support and engineering failed to resolve the issue quickly. | 中 | SR020 |
| CR025 | Another Trustpilot reviewer quoted support saying GPT-4.5 Preview and Claude 3.7 Sonnet had limited requests because of high costs and provider limitations. | 中 | SR020 |
| CR026 | G2 shows a 4.4 out of 5 rating across 19 You.com reviews and says the reviews are authentic and verified. | 中 | SR021 |
| CR027 | TechCrunch reported in 2023 that You.com needed to switch focus from growth to revenue and was experimenting with private ads. | 中 | SR018 |
| CR028 | TechCrunch reported in 2024 that ads had not really been figured out for chat and that You.com was emphasizing paid productivity and enterprise use cases. | 中 | SR019 |
| CR029 | The same 2024 TechCrunch report said large enterprise unit economics were positive and some companies were using You.com millions of times per day. | 中 | SR019 |
| CR030 | You.com’s pricing page lists $5 per 1,000 Search API calls, $1 per 1,000 Contents pages, a $12 Research tier, and a $110 Finance Research tier while also advertising free credits, discounts, and annual savings. | 中 | SR005 |
| CR031 | OpenAI prices web search at $10 per 1,000 calls, showing that a major model platform can bundle grounded retrieval into a broader API stack. | 中 | SR027 |
| CR032 | Google’s Agent Search markets grounded search, citations, connectors, privacy controls, and compliance features from a hyperscaler platform. | 中 | SR025 |
| CR033 | Brave Search API markets grounded answers with citations, 50 qps search capacity, enterprise support, and zero-data-retention positioning. | 中 | SR026 |
| CR034 | You.com’s AWS Marketplace post says purchases can count toward AWS Enterprise Discount Program commitments and show up on consolidated AWS invoices. | 中 | SR011 |
| CR035 | The DPA customer story says the deployment reached 20 departments, 1,000 journalists, and a full rollout in under eight weeks. | 中 | SR012 |
| CR036 | The Wort & Bild case says research time fell from 1.5 days to 3 hours and the publisher reaches more than 16 million monthly readers. | 中 | SR014 |
| CR037 | NIH, Maryville, Mimecast, DPA, HanseMerkur, and Wort & Bild provide real public proof points, but they still represent a curated reference set rather than a disclosed customer base. | 中 | SR012, SR013, SR014, SR015, SR016, SR017 |
| CR038 | Across the reviewed public materials, You.com does not disclose total customer count, top-customer revenue share, GRR, NRR, or renewal rates. | 中 | SR012, SR013, SR015, SR016, SR017, SR019, SR030 |
| CR039 | The API, pricing, and AWS Marketplace materials show that You.com is simultaneously running self-serve API, marketplace, and higher-touch enterprise motions. | 中 | SR005, SR006, SR011 |
| CR040 | The About page still presents Richard Socher and Bryan McCann as the two leading public founders, while the March 2026 CTO update elevates Saahil Jain after Bryan McCann moved to an advisory role. | 中 | SR028, SR029 |
| CR041 | The public trust package includes DPA-ready language, security addenda, onboarding, governance templates, and private-data integration claims, but most control detail remains outside the public web. | 中 | SR003, SR004, SR008 |
| CR042 | The highest residual risks are model and platform dependence, enterprise trust and compliance burden, concentration opacity, and pricing competition rather than lack of product relevance. | 中 | SR001, SR002, SR004, SR013, SR019, SR020, SR022, SR025, SR026 |
| CR043 | Thesis-break triggers include a material trust or regulatory incident, upstream provider repricing or deprecation without a credible fallback, inability to evidence customer diversification, or sustained paid-plan reliability failures. | 中 | SR001, SR002, SR020, SR022, SR025, SR026 |
| CR044 | Mandatory private diligence includes provider mix and fallback design, incident and SLA history, top-customer concentration and renewals, gross margin by product, and full DPA, indemnity, and security schedules. | 中 | SR003, SR004, SR010, SR020, SR022 |
| CR045 | The adverse record is meaningful but not fatal: negative reviews and monetization pivots challenge the company narrative, while G2 and named customer stories prevent a one-sided bear case. | 中 | SR018, SR019, SR020, SR021, SR012, SR013, SR014 |
| CR046 | The Series C post says You.com processes more than 1 billion API queries per month for companies including DuckDuckGo, Windsurf, and Harvey, but it does not reveal workload mix or customer concentration. | 中 | SR030 |
| CR047 | The APIs page offers free MCP access with 100 queries per day and $100 in complimentary credits for new accounts, which broadens top-of-funnel adoption but does not by itself prove durable monetization quality. | 中 | SR006 |
| CV001 | Official and independent 2025 coverage agrees that You.com raised a $100 million Series C at a $1.5 billion valuation. | 高 | SV001, SV013, SV014 |
| CV002 | The publicly disclosed 2021, 2022, 2024, and 2025 rounds sum to roughly $195 million of visible company-level funding. | 高 | SV040, SV041, SV003, SV001 |
| CV003 | Verdict reported in June 2025 that You.com was seeking financing at roughly a $1.4 billion valuation before the later $1.5 billion Series C close. | 中 | SV012, SV001 |
| CV004 | Two August 2025 SEC Form D vehicles named GSBackers You.com Fund I and II totaled about $15.085 million of sold capital, but the filings do not prove direct operating-company proceeds. | 高 | SV016, SV017 |
| CV005 | You.com's current pricing page shows public list prices of $5 per 1,000 Search API calls, $1 per 1,000 Contents pages, $12 per 1,000 Research queries, and $110 per 1,000 Finance Research queries. | 高 | SV002, SV004 |
| CV006 | The same pricing surface explicitly advertises volume discounts and annual savings, so public sticker prices are not the same as realized net pricing. | 中 | SV002, SV004 |
| CV007 | The Series C post claims that You.com now processes more than 1 billion API queries per month. | 中 | SV001, SV013 |
| CV008 | The Series B post claimed that You.com had already served 1 billion cumulative queries and grown ARR 500% since January 2024. | 中 | SV003 |
| CV009 | The Series C post also says more than 100,000 agents were built on You.com in the prior 12 months. | 中 | SV001 |
| CV010 | Official product and enterprise pages show that You.com sells both low-friction self-serve APIs and higher-touch enterprise or private-data deployments. | 高 | SV002, SV005, SV006, SV042 |
| CV011 | Public customer proof spans NIH, Maryville, DPA, HanseMerkur, and Mimecast, which supports real enterprise relevance but still represents a curated reference set rather than a disclosed customer base. | 中 | SV007, SV008, SV009, SV010, SV011 |
| CV012 | Mimecast plus the Series C mention of DuckDuckGo, Windsurf, and Harvey suggests You.com is serving both enterprise workflow buyers and high-usage API customers. | 中 | SV009, SV001, SV013 |
| CV013 | Public materials still do not disclose customer count, top-customer revenue share, net revenue retention, gross revenue retention, or gross margin. | 高 | SV001, SV002, SV006 |
| CV014 | Trustpilot reviews show billing, support, and output-quality complaints that matter for valuation discipline even if the sample is small and noisy. | 中 | SV037 |
| CV015 | You.com's privacy and terms surfaces market zero-retention and enterprise controls while also disclaiming responsibility for third-party services, outputs, and some service changes. | 中 | SV006, SV038, SV039 |
| CV016 | OpenAI's hosted web-search pricing and gpt-oss release lower switching costs for builders who can assemble their own agent stack instead of committing to one search vendor. | 中 | SV026, SV027 |
| CV017 | Direct search API pricing already clusters tightly across You.com, Brave, Exa, and Perplexity, indicating commodity pressure on base retrieval. | 高 | SV002, SV023, SV024, SV025 |
| CV018 | OpenAI prices web search at $10 per 1,000 calls, showing that a major model platform can bundle grounded retrieval inside a broader API stack rather than as a stand-alone search company. | 中 | SV026 |
| CV019 | Glean announced a $150 million Series F in June 2025 at a $7.2 billion valuation, and both Glean and TechCrunch said the company had surpassed $100 million in ARR. | 高 | SV020, SV021 |
| CV020 | Glean's much higher valuation is paired with public ARR disclosure, 100-plus connector positioning, and enterprise-seat packaging that You.com has not matched publicly. | 中 | SV020, SV021, SV022 |
| CV021 | As of June 2026, Elastic's public market cap was about $6.3 billion, and StockAnalysis listed a price-to-sales ratio around 3.6x. | 中 | SV031, SV032 |
| CV022 | As of June 2026, Datadog's public market cap was about $81.8 billion, and StockAnalysis listed a price-to-sales ratio around 22.3x. | 中 | SV033, SV034 |
| CV023 | As of June 2026, Snowflake's public market cap was about $80.7 billion, and StockAnalysis listed a price-to-sales ratio around 16.0x. | 中 | SV035, SV036 |
| CV024 | Public software infrastructure comps command large values only with disclosed revenue, margins, and liquid public-market evidence, so their multiples cannot be transferred mechanically to You.com. | 中 | SV032, SV034, SV036 |
| CV025 | BVP said private AI-cloud valuations had rebounded even while the public cloud index traded around historical norms, which supports AI premium narratives but also warns against carrying marks forward without proof. | 中 | SV018 |
| CV026 | a16z reported that enterprise leaders were nearly tripling generative-AI budgets, moving spend into recurring software lines, and increasingly prioritizing multi-model optionality and data control. | 中 | SV019 |
| CV027 | Those market signals help explain why You.com can win funding and enterprise attention even though the company has not published standard software underwriting metrics. | 中 | SV001, SV018, SV019 |
| CV028 | Those same market signals do not solve You.com's missing public disclosure on revenue, margin, retention, or concentration. | 中 | SV001, SV018, SV019 |
| CV029 | Google, AWS, and Azure all offer enterprise retrieval or search packaging, reinforcing that You.com competes against heavyweight procurement and distribution channels. | 高 | SV028, SV029, SV030 |
| CV030 | The right valuation lens is not consumer search alone; it is a hybrid of search infrastructure, research workflow software, and enterprise AI tooling. | 中 | SV001, SV005, SV006, SV019 |
| CV031 | The $1.5 billion Series C sits far below public software leaders and below Glean's $7.2 billion private mark, but that does not automatically make it cheap because You.com's ARR is undisclosed. | 中 | SV001, SV020, SV021, SV021 |
| CV032 | The main upside thesis is that premium research, finance, and enterprise deployments monetize well above commodity search while query volume continues to scale. | 中 | SV001, SV002, SV004, SV005 |
| CV033 | The main downside thesis is that base retrieval commoditizes while concentration, support burden, and provider dependence cap realized margins and renewal durability. | 中 | SV017, SV014, SV015, SV037, SV038, SV039 |
| CV034 | Public evidence is strong enough to support product-market relevance, but not strong enough to underwrite durable economics at a premium growth multiple. | 中 | SV001, SV007, SV008, SV009, SV010, SV011, SV013 |
| CV035 | A scenario-based framework is more defensible than a single multiple because revenue, gross margin, and retention are not public. | 中 | SV013, SV020, SV021, SV032, SV034, SV036 |
| CV036 | A bear case of roughly $0.8 billion to $1.2 billion fits a world where You.com is mostly a price-pressed retrieval layer with opaque concentration and little proof that premium SKUs drive high-margin recurring revenue. | 低 | SV002, SV017, SV028, SV037 |
| CV037 | A base case of roughly $1.2 billion to $1.8 billion fits a world where the Series C mark broadly holds because enterprise workflows are real, but public economics remain too thin for a richer premium. | 中 | SV001, SV010, SV011, SV019, SV020, SV021 |
| CV038 | A bull case of roughly $1.8 billion to $2.6 billion only makes sense if private diligence confirms strong ARR, healthy gross margins, durable retention, and diversified large-account usage behind the volume claims. | 低 | SV001, SV003, SV019, SV020 |
| CV039 | At the current public evidence level, a buy recommendation would be false precision; research-more or track is the more defensible posture. | 中 | SV013, SV018, SV020, SV021, SV032, SV034, SV036 |
| CV040 | The valuation stance is stretched at the current $1.5 billion public mark, but it could move toward fair if management discloses enough economic proof. | 中 | SV001, SV013, SV018, SV019, SV020 |
| CV041 | The highest-value diligence unlocks are ARR and revenue mix, gross margin by product, concentration and retention, the preference stack, and current cash or runway. | 高 | SV001, SV002, SV016, SV017, SV020 |
| CV042 | IPO-style exit readiness looks limited because You.com still lacks public financial history; a next private round or strategic transaction is the more plausible near-term valuation checkpoint. | 中 | SV001, SV020, SV032, SV034, SV036 |
| CV043 | Thesis-break triggers include a flat or down round, failure to evidence premium-SKU monetization, major trust or support failures, or concentration that makes the >1 billion-query headline less valuable than it sounds. | 中 | SV012, SV014, SV037, SV038, SV039 |
| CV044 | A credible upgrade from research-more toward track or buy would require disclosed ARR scale, retention, and margin data plus evidence that premium workflows materially change realized pricing. | 中 | SV002, SV003, SV004, SV020, SV022 |
| CV045 | The spread between Elastic's ~3.6x sales multiple and much richer Datadog or Snowflake multiples shows how sharply public software valuations widen or compress with growth and quality. | 中 | SV032, SV034, SV036 |
| CV046 | Customer proof in regulated and citation-sensitive workflows supports the thesis that You.com may own a valuable wedge outside generic chat. | 中 | SV004, SV007, SV010, SV011 |
| CV047 | AWS Marketplace positioning and enterprise QPS customization reduce procurement friction for larger deployments even though public marketplace detail is still thin. | 中 | SV006, SV042 |
| CV048 | OpenAI, Perplexity, Exa, and Brave all publish transparent retrieval pricing, so long-run margin capture is more likely to come from differentiated workflow depth than from raw web access alone. | 高 | SV023, SV024, SV025, SV026 |
| 编号 | 出版方 | 标题 | 引文 |
|---|---|---|---|
| SO001 | You.com | You.com Raises $100M Series C at $1.5B Valuation | Today, we process over 1 billion queries monthly with our API for companies like DuckDuckGo, Windsurf, Harvey. |
| SO002 | You.com | About Us | You.com | |
| SO003 | You.com | Privacy Policy | You.com | SuSea, Inc. d/b/a you.com |
| SO004 | You.com | Terms & Conditions | You.com | |
| SO005 | You.com | Meet Richard Socher | You.com | |
| SO006 | You.com | You.com | Saahil Jain as Next You.com CTO | |
| SO007 | You.com | How Richard Socher Built a $1.5B AI Search Unicorn | |
| SO008 | You.com | You.com raises $50M Series B | |
| SO009 | TechCrunch | Former Salesforce chief scientist announces new search engine to take on Google | TechCrunch | |
| SO010 | TechCrunch | You.com lands $20M Seed to reimagine the search engine | TechCrunch | |
| SO011 | TechCrunch | You.com raises $25M to fuel its AI-powered search engine | |
| SO012 | TechCrunch | How You.com plans to combat Google's search dominance | TechCrunch | |
| SO013 | TechCrunch | With $50M in new funding, You.com thinks its AI can beat Google on hard questions | TechCrunch | |
| SO014 | Radical Ventures | You.com - Radical Ventures | |
| SO015 | arXiv | The Natural Language Decathlon: Multitask Learning as Question Answering | |
| SO016 | YouTube | How Prompt Engineering Inventor Built $1.5B in 3 Years | You.com, Richard Socher - YouTube | |
| SO017 | You.com | Careers | You.com | |
| SO018 | You.com | You.com Partners with dpa for AI-Powered News Discovery | |
| SO019 | You.com | How NIH Accelerated Research 1,000× with You.com | |
| SO020 | You.com | How Maryville University Reduced Nursing Dropouts by 78% with You.com | |
| SO021 | You.com | How Mimecast Uses You.com to Drive GTM Productivity | |
| SO022 | DuckDuckGo | About DuckDuckGo | |
| SO023 | Harvey | Harvey – Professional Class AI | |
| SO024 | OpenAI | Introducing gpt-oss | |
| SO025 | GitHub | GitHub - openai/gpt-oss: gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI | |
| SO026 | Maryville University | Maryville University | St. Louis, Missouri | |
| SO027 | You.com | Enterprise AI Solutions | You.com | |
| SO028 | You.com | Our Pricing Plans | You.com | |
| SO029 | Built In San Francisco | AI Infrastructure Company You.com Raises $100M Series C | Built In San Francisco | |
| SO030 | Investing.com | You.com secures $100M in Series C funding at $1.5B valuation By Investing.com | |
| SM001 | You.com | What is AI Search? A Developer Guide | The You.com Web Search API provides this foundation for developers building AI applications. The system processes over one billion queries monthly with 99.9 percent uptime, delivering real-time web data with citations through a model-agnostic architecture. |
| SM002 | You.com | AI Search Infrastructure: The Foundation for Tomorrow's Intelligent Applications | AI search infrastructure is a new category of search technology, built for the age of AI. |
| SM003 | You.com | How Leading APIs Stack Up in the Agentic Era | You.com achieved 77.84% answer accuracy (95% CI: 76.60% - 79.08%) using shorter, more relevant snippets delivered in 445ms (p50). |
| SM004 | You.com | 2025 API Benchmark Report | The YOU API comes out ahead on every benchmark. |
| SM005 | You.com | Introducing the YOU API: Web Scale Search for LLMs | The YOU API empowers developers to overcome the limitations of LLMs by providing them with the tools to ground their outputs in the most recent, accurate, and relevant information. |
| SM006 | You.com | Web Search API Pricing | $5.00 /1k calls |
| SM007 | Elastic | Enterprise search definition | Enterprise search, or corporate search software, is a solution for finding data and information within an enterprise organization. |
| SM008 | Elastic | Why is vector search important? | Vector search powers semantic or similarity search. |
| SM009 | Microsoft Azure | Azure AI Search | Azure AI Search (Foundry IQ) is an enterprise knowledge system that powers sophisticated retrieval-augmented generation (RAG) applications and enterprise search engines. |
| SM010 | Google Cloud | Agent Search on Gemini Enterprise Agent Platform | Agent Search functions as an out-of-the-box RAG system for information retrieval. |
| SM011 | OpenAI | Web search | Web search allows models to access up-to-date information from the internet and provide answers with sourced citations. |
| SM012 | Amazon Web Services | What is Amazon Kendra? | Receive highly accurate answers with an easy-to-use retrieval augmented generation or enterprise search service powered by Machine Learning (ML). |
| SM013 | Glean | Work AI that understands your company | AI only works in the enterprise when answers are secure, explainable, and permission-aware. |
| SM014 | Glean | Glean press | With over 100 connectors, LLM options, and no need for costly professional services, data training, or manual fine-tuning, Glean delivers turnkey implementation of a complex AI ecosystem on one centralized platform. |
| SM015 | Exa | Getting started | Exa finds the exact content you’re looking for on the web, with four core functionalities: /search, /contents, /answer, /research. |
| SM016 | Exa | Pricing | Base price, with up to 10 results (per 1k requests) $7. |
| SM017 | Tavily | Welcome | Your journey to state-of-the-art web search starts right here. |
| SM018 | Brave | Brave Search API | Power your agents & chatbots with the world's largest independent index of the Web. |
| SM019 | Andreessen Horowitz | 16 Changes to the Way Enterprises Are Building and Buying Generative AI | In 2023, the average spend across foundation model APIs, self-hosting, and fine-tuning models was $7M. |
| SM020 | Bessemer Venture Partners | State of the Cloud 2024 | Players such as OpenAI, Anthropic, Mistral, Cohere, among others raised $23 billion at an aggregate market cap of $124 billion. |
| SM021 | Microsoft WorkLab | Work Trend Index | As AI and agents take on execution, our own agency expands. |
| SM022 | arXiv | Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks | RAG models generate more specific, diverse and factual language than a state-of-the-art parametric-only seq2seq baseline. |
| SM023 | arXiv | Retrieval-Augmented Generation for Large Language Models: A Survey | Retrieval-Augmented Generation (RAG) has emerged as a promising solution by incorporating knowledge from external databases. |
| SM024 | SiliconANGLE | You.com raises $100M to help developers build AI applications | You.com has evolved from its initial focus on connecting large language models (LLMs) to live web data with verifiable citations to becoming an AI infrastructure company that processes over 1 billion queries monthly through its API. |
| SM025 | You.com | You.com Raises $100M Series C at $1.5B Valuation | Today, we process over 1 billion queries monthly with our API for companies like DuckDuckGo, Windsurf, Harvey, and others across technology, e-commerce, financial services, media, manufacturing, hospitality and more. |
| SP001 | You.com | Web Search API Pricing | $5.00 /1k calls |
| SP002 | You.com | Enterprise AI Solutions | Our platform connects seamlessly with any and all data sources, including GDrive, SharePoint, and Databricks. |
| SP003 | TechCrunch | With $50M in new funding, You.com thinks its AI can beat Google on hard questions | it would be futile to try and compete with Google over simple queries. |
| SP004 | You.com | You.com Trust Center | You.com Trust Center |
| SP005 | You.com | Introducing ARI, the first professional-grade research agent for business | ARI makes research faster and more trustworthy. |
| SP006 | You.com | Introducing the You.com Finance Research API: Agentic Research, No Infra Required | The Finance Research API uses a per-request pricing model with two tiers based on research depth. |
| SP007 | OpenAI GitHub | ydc-deep-research-evals | This repository contains evals metrics and scripts for evaluating Deep Research reports. |
| SP008 | Glean | Glean – Work AI that Works | Agents, Assistant & Search | Glean connects knowledge, systems, and context so AI can actually work. |
| SP009 | Glean | Glean press | With over 100 connectors, LLM options, and no need for costly professional services, data training, or manual fine-tuning, Glean delivers turnkey implementation of a complex AI ecosystem on one centralized platform. |
| SP010 | Glean | Work AI you can trust | Secure AI at work with a private-by-design platform, sensitive data protection, and guardrails that keep AI agents safely in scope. |
| SP011 | Glean | Leading enterprise AI solution | Glean’s enterprise AI brings together your emails, documents, chats, and business systems. |
| SP012 | Glean Docs | Glean Enterprise Flex | Enterprise Flex Seats are designed to be deployed to every employee in your organization. |
| SP013 | Exa | Getting started | Exa finds the exact content you’re looking for on the web, with four core functionalities. |
| SP014 | Exa | Pricing | $7 |
| SP015 | Tavily | Welcome | Search the web |
| SP016 | Tavily | Pricing Overview | Pay-as-you-go: $0.008 per credit. |
| SP017 | Brave | Brave Search API | $4 per 1,000 requests |
| SP018 | OpenAI | Web search | Web search allows models to access up-to-date information from the internet and provide answers with sourced citations. |
| SP019 | OpenAI | API Pricing | Web search $10.00 / 1k calls |
| SP020 | Perplexity Docs | Pricing | Search API $5.00 |
| SP021 | Microsoft Azure | Azure AI Search | Azure AI Search also supports data connections in knowledge bases, including SharePoint, the public web, and Azure Blob storage. |
| SP022 | Microsoft Azure | Pricing - Foundry IQ | Dedicated plans are for enterprise workloads with consistent traffic and high-utilization with an hourly pricing mode. |
| SP023 | Amazon Web Services | What is Amazon Kendra? | Receive highly accurate answers with an easy-to-use retrieval augmented generation or enterprise search service powered by Machine Learning. |
| SP024 | Amazon Web Services | Amazon Kendra Pricing | Base Index (1 Storage Unit + 1 Query unit) $0.32 per hour |
| SP025 | Google Cloud | Agent Search on Gemini Enterprise Agent Platform | Agent Search functions as an out-of-the-box RAG system for information retrieval. |
| SP026 | Google Cloud | Agent Search pricing | Search Enterprise Edition includes Core Generative Answers (AI Mode) $4.00 / 1,000 query |
| SP027 | Elastic | Enterprise search definition | Enterprise search is the practice of making content from multiple enterprise sources searchable to a defined audience. |
| SP028 | Elastic | Why is vector search important? | Vector search uses machine learning models to represent data as vectors. |
| SP029 | Harvey | Harvey – Professional Class AI | Harvey is AI designed for legal and professional services. |
| SP030 | DuckDuckGo | About DuckDuckGo | 3B Monthly Searches |
| SP031 | OpenAI | Introducing gpt-oss | We’re releasing gpt-oss-120b and gpt-oss-20b—two state-of-the-art open-weight language models. |
| SP032 | OpenAI GitHub | gpt-oss | Agentic capabilities: Use the models' native capabilities for function calling, web browsing, Python code execution, and Structured Outputs. |
| SP033 | AWS Marketplace | AWS Marketplace | Top 1 By Brave |
| SP034 | You.com | How Leading APIs Stack Up in the Agentic Era | This 2025 API Benchmarking Report rigorously compares You.com’s Search API with leading alternatives. |
| SP035 | You.com | You.com Raises $100M Series C at $1.5B Valuation | Today, the You.com API handles over 1 billion queries per month. |
| SI001 | You.com | Our Pricing Plans | $100 free credit to get started · SOC 2 certified · No model training |
| SI002 | You.com | You.com has raised a $50M Series B led by Georgian | Since our launch, we've served 1 billion queries, and our ARR has grown by 500% since January 2024. |
| SI003 | You.com | You.com Raises $100M Series C at a $1.5 Billion Valuation to Build the Infrastructure for the Agentic Era | Today, we process over 1 billion queries monthly with our API for companies like DuckDuckGo, Windsurf, Harvey, and others. |
| SI004 | You.com | Introducing the You.com Finance Research API: Agentic Research, No Infra Required | The You.com Finance Research API is purpose-built for developers who need accurate, current, and deeply sourced intelligence in their financial workflows. |
| SI005 | You.com | Introducing ARI, the first professional-grade research agent for business | You.com's newest deep research agent thinks, analyzes up to 400 sources, and creates professional-grade research in under five minutes. |
| SI006 | You.com | Benchmarking ARI: 76% Win Rate Over OpenAI Deep Research, According to OpenAI's Model | We’re also launching new enterprise-only features and, for a limited time, unlimited ARI reports for new enterprise customers. |
| SI007 | You.com | September 2025 API Roundup: Introducing Express & Contents APIs | Express API: Fast, web-grounded answer API with “search + answer” all in one call. |
| SI008 | You.com | November 2025 API Roundup: Introducing Deep Search and New Developer Tooling | Deep Search retrieves live webpages, extracts only the passages that directly answer the query, and verifies that the text genuinely appears on those pages before returning it. |
| SI009 | You.com | December 2025 API Roundup: What's New at You.com | We’re introducing a new evaluation service for teams who need to measure and compare search, LLM model and other AI agentic performance in real workflows. |
| SI010 | You.com Developer Documentation | You.com API Docs | Real-Time Web Intelligence for AI Applications | Connect any MCP-enabled tool ... and use you-search with no signup or credentials. Limited to 100 queries per day. |
| SI011 | You.com Developer Documentation | You.com Finance Research API | Finance-Grade Research with Citations | Instead of the open web, the Finance Research API searches a purpose-built financial index — covering structured company fundamentals, equity and commodity price data, private company metrics, alternative signals, macroeconomic indicators, SEC filings, and financial news. |
| SI012 | You.com | Enterprise AI Solutions | Combine private data and real-time web results with the best model for the job, enabling fast, accurate reasoning grounded in your context. |
| SI013 | You.com | How NIH Accelerated Research 1000x with You.com | How NIH Accelerated Research 1000x with You.com |
| SI014 | You.com | How Maryville University Reduced Nursing Dropouts by 78% with You.com | How Maryville University Reduced Nursing Dropouts by 78% with You.com |
| SI015 | You.com | How Mimecast Uses You.com to Drive GTM Productivity | How Mimecast Uses You.com to Drive GTM Productivity |
| SI016 | Securities and Exchange Commission | Form D - GSBackers You.com Fund I a Series of CGF2021 LLC | GSBackers You.com Fund I a Series of CGF2021 LLC |
| SI017 | Securities and Exchange Commission | Form D - GSBackers You.com Fund II a Series of CGF2021 LLC | GSBackers You.com Fund II a Series of CGF2021 LLC |
| SI018 | Securities and Exchange Commission | Filing a Form D Notice | Form D is used to file a notice of an exempt offering of securities with the SEC. |
| SI019 | Securities and Exchange Commission | Form D Data Sets | The data sets will be updated quarterly. |
| SI020 | TechCrunch | You.com lands $20M series A to reimagine the search engine | You.com lands $20M series A to reimagine the search engine |
| SI021 | TechCrunch | You.com raises $25M to fuel its AI-powered search engine | You.com raises $25M to fuel its AI-powered search engine |
| SI022 | TechCrunch | You.com refocuses from AI search to deeper productivity agents with new $50M round | It would have been futile to compete with Google on easy questions. |
| SI023 | Verdict | You.com seeking $1.4bn valuation in new funding round | AI startup You.com is in discussions to secure new funding at a valuation of $1.4bn. |
| SI024 | Investing.com | You.com secures $100M in Series C funding at $1.5B valuation | You.com has raised $100 million in Series C funding at a $1.5 billion valuation. |
| SI025 | Built In San Francisco | You.com Raises $100M Series C | You.com ... provides developers and enterprises with over 1 billion deeply researched, AI-generated answers a month through its APIs. |
| SI026 | Exa | API Pricing | Base price, with up to 10 results (per 1k requests) $7 |
| SI027 | Brave | Brave Search API | $4 per 1,000 requests |
| SI028 | OpenAI | OpenAI API Pricing | $10.00 / 1k calls |
| SI029 | Radical Ventures | You.com portfolio page | You’s AI conductor orchestrates the flow of queries across multiple private data sources, internet search, and LLMs to collect the most relevant information. |
| SE001 | You.com Developer Documentation | You.com API Docs | Real-Time Web Intelligence for AI Applications | Build agentic apps with real-time web intelligence. |
| SE002 | You.com | Web Search API Pricing | All queries and data can be automatically purged—giving you complete control over what’s stored and for how long. |
| SE003 | You.com | Enterprise AI Solutions | Combine private data and real-time web results with the best model for the job, enabling fast, accurate reasoning grounded in your context. |
| SE004 | You.com | Search API for the Agentic Era | They're making intelligent tool calls that require structured, high-signal, context-aware passages to reason and act effectively. |
| SE005 | You.com | AI Search Infrastructure: The Foundation for Tomorrow's Intelligent Applications | Whether you want to leverage ready-made Search APIs and vertical-specific indexes for immediate high-accuracy retrieval, or create custom solutions through private RAG, custom agents, or tailored indexes... |
| SE006 | You.com | What Is AI Search? | RAG works in two steps: first, it retrieves relevant documents from a search index, then generates an answer by feeding those documents to an LLM. |
| SE007 | You.com | Introducing the YOU API: Web Scale Search for LLMs | Intermediate modules like query rewriting will be exposed as new endpoints, enabling developers to customize their usage of the YOU API. |
| SE008 | You.com | Introducing ARI, the first professional-grade research agent for business | You.com's newest deep research agent thinks, analyzes up to 400 sources, and creates professional-grade research in under five minutes. |
| SE009 | You.com | Benchmarking ARI: 76% Win Rate Over OpenAI Deep Research, According to OpenAI's Model | ARI now offers 4x greater depth and breadth compared to the beta version of ARI. |
| SE010 | You.com | Introducing the You.com Finance Research API: Agentic Research, No Infra Required | Built on the You.com Research API, and augmented with licensed structured data from S&P Global and others... |
| SE011 | You.com Developer Documentation | You.com Finance Research API | Finance-Grade Research with Citations | Instead of the open web, the Finance Research API searches a purpose-built financial index. |
| SE012 | You.com | September 2025 API Roundup: Introducing Express & Contents APIs | Express packages “search + answer” in a single API call. |
| SE013 | You.com | November 2025 API Roundup: Introducing Deep Search and New Developer Tooling | Deep Search retrieves live webpages, extracts only the passages that directly answer the query, and verifies that the text genuinely appears on those pages before returning it. |
| SE014 | You.com | December 2025 API Roundup: What's New at You.com | We're introducing a new evaluation service for teams who need to measure and compare search, LLM model and other AI agentic performance in real workflows. |
| SE015 | You.com | 2025 API Benchmarking Report | You.com’s standard search endpoint delivers precise answers in under 445ms. |
| SE016 | You.com | You.com Powers OpenAI's New GPT OSS Models: New Era of Open, Accurate AI | OpenAI has integrated You.com as a core search provider, bringing real-time, high-quality web search directly to advanced language models. |
| SE017 | You.com | Privacy Policy | You.com | This data will not be used for developing, improving, or training third party AI and/or ML models. |
| SE018 | You.com | Terms & Conditions | You.com | Some features and/or functions of the Services that interoperate with third parties ... depend on the continuing availability of such third parties’ respective APIs. |
| SE019 | You.com | How NIH Accelerated Research 1000x with You.com | Researchers can surface hundreds of relevant sources in minutes, with citations that tie directly back to peer-reviewed material. |
| SE020 | You.com | How Maryville University Reduced Nursing Dropouts by 78% with AI Agents from You.com | 600+ custom AI agents created by faculty and students. |
| SE021 | You.com | How Mimecast Uses You.com to Drive GTM Productivity | The team selected You.com for its multi-model AI access, ability to create and share custom AI agents between teams, and competitive pricing compared to alternatives. |
| SE022 | GitHub | youdotcom/ydc-deep-research-evals | This repository contains evals metrics and scripts for evaluating Deep Research reports. |
| SE023 | OpenAI | Web search | Deep research is a specialized, agent-driven method for in-depth, extended investigations by reasoning models. |
| SE024 | OpenAI | Introducing GPT-OSS | These models are compatible with our Responses API and are designed to be used within agentic workflows with exceptional instruction following, tool use like web search or Python code execution. |
| SE025 | Google Cloud | Agent Search on Gemini Enterprise Agent Platform | Agent Search functions as an out-of-the-box RAG system for information retrieval. |
| SE026 | Microsoft Azure | Azure AI Search | Hybrid search executes a query for both text search and vector search in one request. |
| SE027 | Amazon Web Services | What is Amazon Kendra? | Receive highly accurate answers with an easy-to-use retrieval augmented generation or enterprise search service powered by Machine Learning. |
| SE028 | Elastic | Enterprise search definition | Enterprise search works by allowing data to be indexed, queried, and displayed to authorized users. |
| SE029 | Exa | Getting Started | Exa finds the exact content you’re looking for on the web, with four core functionalities: /search, /contents, /answer, /research. |
| SE030 | Tavily | Welcome | Search the web. Extract webpages. Crawl webpages. Map webpages. Create a Research Task. |
| SE031 | Brave | Brave Search API | Answers grounded on a single search or multiple searches for better accuracy & reduced hallucinations. |
| SE032 | GitHub | openai/gpt-oss | Agentic capabilities: Use the models' native capabilities for function calling, web browsing, Python code execution, and Structured Outputs. |
| SE033 | TechCrunch | You.com refocuses from AI search to deeper productivity agents with new $50M round | It would have been futile to compete with Google on easy questions. |
| SE034 | You.com | How APIs Became the Connective Tissue of LLMs | The model handles the reasoning but the API handles the real-world interaction—neither one is sufficient alone. |
| SE035 | You.com | Tool Design for Agentic Search | Our agent gathers information through two primary tools: web_search(query: str) and read_page(url: str, goal: str). |
| SE036 | You.com | Context Rot Is Quietly Breaking Your API Integrations | Context rot is the gradual degradation of an API’s usefulness as the context it depends on ... silently drifts out of alignment with the real world. |
| SE037 | You.com | Same LLM, Better Web Search, Better Outcome | With You.com, Claude Sonnet 4.6 reviews compact snippets first and only pulls full pages when it needs them. |
| SE038 | You.com | You.com Web Intelligence Is Now Available in MindStudio’s Remy Apps | Four purpose-built capabilities expose that at different layers of depth and structure: Web Search, Live News, Page Content Extraction, Research, Financial Intelligence. |
| SU001 | You.com | Enterprise AI Solutions | |
| SU002 | You.com | Web Search API Pricing | |
| SU003 | You.com | How NIH Accelerated Research 1000x with You.com | The rollout began with a pilot group of 30–35 users ... leading to an expansion toward 120 licenses across NIH. |
| SU004 | You.com | How Maryville University Reduced Nursing Dropouts by 78% with You.com | Withdrawals dropped from 45% to 10% - 78% reduction. |
| SU005 | You.com | How Mimecast Uses You.com to Drive GTM Productivity | As a result, sales enablement content increased by 20%, while GTM processes showed marked improvement in speed and efficiency. |
| SU006 | You.com | You.com Partnerships | |
| SU007 | You.com Developer Documentation | SDKs & Tools Overview | |
| SU008 | You.com Developer Documentation | OpenAI GPT OSS Integration | |
| SU009 | You.com Developer Documentation | Zapier Integration | |
| SU010 | You.com Developer Documentation | Vercel AI SDK Integration | |
| SU011 | You.com | You.com Web Intelligence Is Now Available in MindStudio’s Remy Apps | |
| SU012 | You.com | You.com Partners with German Press Agency (dpa) for Agentic AI News Discovery and Curation | |
| SU013 | You.com | You.com Raises $100M Series C at a $1.5 Billion Valuation to Build the Infrastructure for the Agentic Era | Today, we process over 1 billion queries monthly with our API for companies like DuckDuckGo, Windsurf, Harvey, and others. |
| SU014 | G2 | You.com Reviews 2025: Details, Pricing, & Features | |
| SU015 | Trustpilot | you.com Reviews | Read Customer Service Reviews of you.com | Been using You.com nearly a year for educational curriculum ... I can't add any team members and they're already paid for. |
| SU016 | AWS Marketplace | AWS Marketplace: Search Results | |
| SU017 | Maryville University | Maryville University | |
| SU018 | National Institutes of Health | National Institutes of Health | |
| SU019 | Mimecast | Mimecast | |
| SU020 | DuckDuckGo | DuckDuckGo About | |
| SU021 | Harvey | Harvey | |
| SU022 | Windsurf | Agent Command Center | |
| SU023 | OpenAI | Introducing GPT-OSS | |
| SU024 | GitHub | openai/gpt-oss repository | |
| SU025 | You.com | Privacy Policy | |
| SU026 | You.com | Terms & Conditions | |
| SU027 | MindStudio | Remy — An AI agent that builds, ships, and runs products | |
| SR001 | You.com | Privacy Policy | |
| SR002 | You.com | Terms & Conditions | |
| SR003 | You.com | Legal | |
| SR004 | You.com | Enterprise AI Solutions | |
| SR005 | You.com | Web Search API Pricing | |
| SR006 | You.com | APIs | |
| SR007 | You.com | Benchmarks | |
| SR008 | You.com | Roll Out Enterprise AI Without Rolling the Dice | |
| SR009 | You.com | AI Hallucination Prevention Guide | |
| SR010 | You.com | API Latency | |
| SR011 | You.com | Maximize Your AWS Committed Spend: Transform Business Productivity With You.com via AWS Marketplace | |
| SR012 | You.com | How DPA Transformed Journalism with AI-Powered News Research | |
| SR013 | You.com | How HanseMerkur Set New Standards in Insurance with AI | |
| SR014 | You.com | How Wort & Bild Verlag Accelerates Research with You.com | |
| SR015 | You.com | How NIH Accelerated Research 1000x with You.com | |
| SR016 | You.com | How Maryville University Reduced Nursing Dropouts by 78% with You.com | |
| SR017 | You.com | How Mimecast Uses You.com to Drive GTM Productivity | |
| SR018 | TechCrunch | You.com wants to build a search engine to take on Google | |
| SR019 | TechCrunch | You.com refocuses from AI search to deeper productivity agents with new $50M round | |
| SR020 | Trustpilot | You.com Reviews | The GPT-4.5 Preview and the Claude 3.7 Sonnet model currently have limited requests available due to high costs and provider limitations. |
| SR021 | G2 | You.com Reviews | |
| SR022 | European Commission | AI Act | |
| SR023 | You.com Documentation | OpenAI GPT OSS Integration | |
| SR024 | OpenAI | Introducing GPT OSS | |
| SR025 | Google Cloud | Agent Search on Gemini Enterprise Agent Platform | |
| SR026 | Brave | Brave Search API | |
| SR027 | OpenAI | API Pricing | |
| SR028 | You.com | About | |
| SR029 | You.com | Today, Saahil Jain becomes You.com CTO | |
| SR030 | You.com | You.com Raises $100M Series C at a $1.5 Billion Valuation to Build the Infrastructure for the Agentic Era | Today, we process over 1 billion queries monthly with our API for companies like DuckDuckGo, Windsurf, Harvey, and others across technology, e-commerce, financial services, media, manufacturing, hospitality and more. |
| SR031 | Amazon Web Services | AWS Marketplace | |
| SR032 | DuckDuckGo | About DuckDuckGo | |
| SV001 | You.com | You.com Raises $100M Series C at a $1.5 Billion Valuation to Build the Infrastructure for the Agentic Era | We raised $100 million in Series C funding at a $1.5 billion valuation. |
| SV002 | You.com | Web Search API Pricing | Volume discounts ... Annual savings. |
| SV003 | You.com | You.com has raised a $50M Series B led by Georgian | Our ARR has grown by 500% since January 2024. |
| SV004 | You.com | Introducing the You.com Finance Research API: Agentic Research, No Infra Required | |
| SV005 | You.com | Introducing ARI, the first professional-grade research agent for business | |
| SV006 | You.com | Enterprise AI Solutions | |
| SV007 | You.com | How NIH Accelerated Research 1000x with You.com | |
| SV008 | You.com | How Maryville University Reduced Nursing Dropouts by 78% with You.com | |
| SV009 | You.com | How Mimecast Uses You.com to Drive GTM Productivity | |
| SV010 | You.com | How DPA Transformed Journalism with AI-Powered News Research | |
| SV011 | You.com | How HanseMerkur Set New Standards in Insurance with AI | |
| SV012 | Verdict | You.com seeking $1.4bn valuation in new funding round | |
| SV013 | Investing.com | You.com secures $100M in Series C funding at $1.5B valuation | |
| SV014 | Built In San Francisco | You.com Raises $100M Series C | |
| SV015 | TechCrunch | You.com refocuses from AI search to deeper productivity agents with new $50M round | |
| SV016 | Securities and Exchange Commission | Form D - GSBackers You.com Fund I a Series of CGF2021 LLC | |
| SV017 | Securities and Exchange Commission | Form D - GSBackers You.com Fund II a Series of CGF2021 LLC | |
| SV018 | Bessemer Venture Partners | State of the Cloud 2024 | |
| SV019 | Andreessen Horowitz | 16 Changes to the Way Enterprises Are Building and Buying Generative AI | |
| SV020 | Glean | Glean raises $150M Series F at $7.2B valuation to accelerate enterprise AI agent innovation globally | Glean today announced it raised $150 million in Series F financing, bringing its valuation to $7.2 billion. |
| SV021 | TechCrunch | Enterprise AI startup Glean lands a $7.2B valuation | |
| SV022 | Glean Docs | Glean Enterprise Flex | |
| SV023 | Perplexity Docs | Pricing | |
| SV024 | Exa | API Pricing | |
| SV025 | Brave | Brave Search API | |
| SV026 | OpenAI | OpenAI API Pricing | |
| SV027 | OpenAI | Introducing gpt-oss | |
| SV028 | Google Cloud | Agent Search on Gemini Enterprise Agent Platform | |
| SV029 | Amazon Web Services | Amazon Kendra Pricing | |
| SV030 | Microsoft Azure | Pricing - Foundry IQ / Azure AI Search | |
| SV031 | CompaniesMarketCap | Elastic NV (ESTC) - Market capitalization | |
| SV032 | StockAnalysis | Elastic (ESTC) Statistics & Valuation | |
| SV033 | CompaniesMarketCap | Datadog (DDOG) - Market capitalization | |
| SV034 | StockAnalysis | Datadog (DDOG) Statistics & Valuation | |
| SV035 | CompaniesMarketCap | Snowflake (SNOW) - Market capitalization | |
| SV036 | StockAnalysis | Snowflake (SNOW) Statistics & Valuation | |
| SV037 | Trustpilot | you.com Reviews | Read Customer Service Reviews of you.com | |
| SV038 | You.com | Privacy Policy | |
| SV039 | You.com | Terms & Conditions | |
| SV040 | TechCrunch | You.com lands $20M series A to reimagine the search engine | |
| SV041 | TechCrunch | You.com raises $25M to fuel its AI-powered search engine | |
| SV042 | You.com | Maximize Your AWS Committed Spend: Transform Business Productivity With You.com via AWS Marketplace |