Iterative Health

1.1 Identity, positioning, and operating model

Iterative Health was founded in 2017 and is now presented as a healthcare technology and services company headquartered in Cambridge, Massachusetts, and New York, New York. The company’s current homepage and about page frame the business less as a single-product AI startup and more as a multispecialty clinical research infrastructure platform. Its operating focus now spans GI, hepatology, obesity, and cardiology, with research embedded directly into community care settings rather than concentrated only in academic centers. That positioning marks a clear expansion from the earlier Iterative Scopes identity, which emphasized precision gastroenterology, computer vision, and products such as SKOUT. The current model connects provider organizations and research sites to centralized operations, sponsor access, and AI-enabled workflow tools. In practice, Iterative sells speed, activation reliability, and enrollment performance to sponsors and CROs while giving provider groups research infrastructure, staffing support, and patient-identification tooling. SKOUT still matters because it shows the company retains a real commercial device footprint rather than operating only as a services intermediary. But the core 2026 investment narrative is broader: a scaled research network, proprietary workflow technology, and tighter integration between provider sites and sponsor demand. That combination is what later chapters should treat as the company’s ground-truth business model.[CO001, CO002, CO003, CO004, CO005, CO008]

1.2 Founders, leadership, and governance signals

Founder and CEO Jonathan Ng remains the central strategic figure. Public materials consistently identify him as founder/CEO, and the earlier Iterative Scopes financing history ties the company’s origin to an MIT spinout built around Ng’s physician-entrepreneur profile and MIT/Harvard training context. That gives the company credible founder-market fit at the intersection of medicine, AI, and clinical trial operations. Leadership depth improved materially in late 2025. Bill Kayser joined as President and Chief Financial Officer after senior finance roles at GI Alliance, Prospero Health, and McKesson, which is notable because it adds operator experience from physician-practice infrastructure and healthcare M&A rather than pure software finance. Public team materials also show Dana Feuchtbaum as COO and Nadege Gunn as Chief Medical Officer for Hepatology and Obesity, reinforcing the company’s operating emphasis on scaling sites and entering adjacent therapeutic areas. Governance visibility is still partial rather than complete. Public financing announcements disclose a progression of investor-linked board seats or observer roles — including Nan Li and Lotus Mallbris around Series A, Lonne Jaffe around Series B, and Ajay Agrawal plus Anthony Philippakis around Series C — but they do not provide a fully reconciled current board roster or ownership map. The governance takeaway is therefore directional: founder-led, increasingly institutionally supervised, but still opaque on full control rights.[CO006, CO007, CO009, CO010, CO013, CO014]

1.3 Funding history, scale signals, and stakeholder map

Publicly disclosed financing history establishes a credible late-stage company, even though exact valuation and cap-table detail remain partly opaque. Iterative raised a $30 million Series A in 2021 after an earlier $13.5 million seed, then a $150 million Series B in January 2022 co-led by Insight Partners and Clearlake. The 2026 Series C added another $77 million led by Intrepid Growth Partners and GV, with EDBI as a new investor and Insight and Obvious Ventures participating again. That sequence supports a durable institutional syndicate and suggests the company successfully repositioned itself from an AI-enabled GI startup into a broader clinical research platform. Scale claims also strengthened over 2025-2026. Official releases now describe more than 100 research sites across multiple continents, more than 40 pharmaceutical, biotech, medical-device, and CRO partners, and a roughly 250-person organization. Those are meaningful signals because the business model depends on operational density and repeatability, not just algorithmic novelty. GI Alliance, One GI, US Heart & Vascular, and the NextStage cardiology site acquisition show that provider-network partnerships are now as strategically important as financial investors. The important caveat is disclosure quality. TechCrunch, Forge, AI2.work, and Tracxn all imply total funding around $270 million and a latest valuation in the roughly $1.3-1.4 billion range, but those figures remain secondary estimates because the company did not publicly disclose a Series C valuation. Headcount is also inconsistent across third-party trackers. Investors can therefore underwrite momentum and syndicate quality, but not yet precise economics or ownership structure.[CO009, CO011, CO012, CO015, CO016, CO017]

1.4 Milestones, regulatory proof points, and key risks

The milestone record shows two distinct company phases. The first phase runs from the 2017 founding through the 2022 Series B and emphasizes AI for gastroenterology, including the acquisition of CRSG/Precision and the 2022 FDA clearance for SKOUT. The second phase, most visible in 2025-2026, emphasizes the buildout of a multispecialty clinical research network through GI Alliance, One GI, US Heart & Vascular, and the acquisition of three cardiology sites from NextStage. SKOUT remains the clearest regulatory and product proof point. Official materials and FDA documentation support that the device is an FDA-cleared, real-time computer-aided detection aid for colonoscopy, and company reporting cites a 27% increase in adenomas per colonoscopy and a 44% increase in proximal 5-9 mm polyp detection. Those facts matter because they show Iterative can translate its AI work into regulated clinical products, not just workflow software. The main adverse context is not a company-specific enforcement event but category risk. Independent literature reviewing CADe systems still highlights false positives, automation bias, workflow integration burden, generalizability limits, and uncertain long-term outcome data. Combined with the lack of public revenue, debt, and cap-table disclosure, that means the company’s execution narrative is stronger than its public financial transparency. Later chapters should therefore treat Iterative as operationally credible but still diligence-heavy on economics and adoption durability.[CO005, CO024, CO025, CO026, CO027, CO028]

1.5 Exhibits

2.1 Market boundary and demand base

Iterative Health is exposed to two markets that overlap clinically but clear through different buyers. One is AI-assisted colonoscopy quality tooling: software such as SKOUT that sits inside the endoscopy room and tries to improve lesion detection against a rising quality bar. The other is GI/community clinical-trial site-network services: operating infrastructure sold mainly to sponsors and CROs that need faster activation, better enrollment, and access to community patient populations. Treating those as one TAM would hide the fact that one market is budgeted like provider operating or technology spend, while the other is outsourced trial-execution spend. The included spend for the quality-tool market is narrow. It covers lesion-detection software, deployment/support, and quality-workflow fit inside screening or surveillance colonoscopy. It excludes the broader CRC screening universe, which now also includes blood-based and at-home stool tests, and it excludes generic endoscopy software that does not change detection performance. The included spend for the site-network market is equally specific: site startup, contracting, training, patient-identification support, and sponsor access to high-performing community GI sites. It does not include the full CRO stack or every piece of GI research spend. Demand is real but uneven. ACS says more than 20 million eligible Americans are still not up to date on CRC screening, and Medicare policy still routes positive stool or blood tests into follow-up colonoscopy. But the newly age-expanded 45-49 cohort remains far from saturated: only 33.7% were up to date in 2023, with adoption concentrated among better-insured and better-educated adults. Iterative's colonoscopy-facing market is therefore anchored to a large prevention gap, but adoption still depends on access, outreach, and whether direct-visual screening remains attractive versus non-invasive alternatives.[CM001, CM002, CM003, CM004, CM005, CM006]

2.2 Evidence-constrained sizing lenses

The public record supports market importance but not a clean dollar TAM for either of Iterative's two exposures. For colonoscopy AI, the most useful outer-shell lens is not revenue but quality pressure applied to a still-under-screened population. The 2024 ACG/ASGE update broadened ADR measurement to essentially all screening, surveillance, and diagnostic colonoscopies in adults 45 and older and raised the benchmark to 35% overall, with a 6% SSLDR target. That materially increases the relevance of tools that promise more consistent detection. Yale's summary of the evidence and Olympus' EAGLE trial both support the core proposition that CADe can improve adenoma detection without a meaningful time penalty. But buyers still have to underwrite around uncertainty. The same literature says improvements in advanced neoplasia are smaller, long-term interval-cancer and real-world cost-effectiveness evidence remain limited, and CADe commercialization has arguably moved faster than post-market consensus. The right sizing lens for the procedure-room business is therefore a constrained wedge: a subset of colonoscopy programs that feel quality pressure strongly enough to buy despite ambiguous reimbursement. For the site-network business, the strongest public lens is sponsor pain rather than market-research TAM. Iterative's IBD data point to 74-day activation versus 122-171-day benchmarks, 0.34 patients per site per month versus roughly 0.10 benchmark productivity, and similar 0.33-0.32 performance at larger DDW 2026 scale. Those are meaningful service-market indicators, but they are not a publishable SAM/SOM because the public record omits pricing, contracted volume, backlog, and site-level revenue.[CM007, CM008, CM011, CM012, CM013, CM017]

2.3 Buyer, user, and payer map

The buyer map splits cleanly across the two businesses. For CADe, the everyday user is the endoscopist, but the economic conversation usually runs through practice administrators, endoscopy medical directors, hospital service-line leaders, or VA and health-system quality owners. The payer is generally the provider organization itself. The reviewed payment sources did not surface dedicated Medicare reimbursement for AI-assisted colonoscopy, so budget owners have to justify spend through quality improvement, competitive differentiation, training support, or workflow efficiency. Olympus' cloud-procurement framing and Yale's button-press implementation both suggest that vendors know this market must fit existing room economics rather than wait for a new payment code. For the site-network business, the economic buyer is much more sponsor-side. Iterative's own materials and Clinical Trials Arena both describe centralized access for sponsors and CROs to higher-performing, more diverse community sites. Provider organizations such as GI Alliance matter because they supply investigators, coordinators, and patient relationships, but the core spend is closer to outsourced trial execution than to endoscopy reimbursement. Community practices also receive a co-benefit: research can be embedded into ordinary GI care, which gives physicians and patients access without forcing everyone back into academic-center workflows. This split matters for valuation. The colonoscopy-tool business is constrained by provider budgeting and room-level ROI. The site-network business is constrained by sponsor demand, site performance, and the ability to keep community practices activated across multiple studies. Calling them one market would miss the fact that they clear through different procurement motions and different payers.[CM016, CM033, CM040, CM041, CM042, CM043]

2.4 Growth drivers and adoption constraints

The biggest shared tailwinds are straightforward: CRC screening now begins at age 45, the U.S. still has a large unscreened pool, colonoscopy quality benchmarks are tighter, and sponsors remain desperate for faster, more reliable recruitment. On paper that is a strong setup for both SKOUT and the site-network platform. Community GI care is especially important because it concentrates ongoing relationships with chronic-disease patients who matter to both screening volume and GI trial enrollment. The constraints are equally important. On the CADe side, the literature still flags false positives, automation bias, generalizability, and limited long-term outcomes; the evidence is good enough to support interest, but not strong enough to eliminate buyer skepticism. Just as importantly, the reimbursement backdrop is hostile. GI societies say the 2026 CMS proposal would cut ASC and hospital-based endoscopy payments, while Becker's and ACG describe a longer-running gap between Medicare payments and practice costs. MACRA-style cost measurement further sensitizes providers to site of service, anesthesia, pathology, and follow-on utilization. In that context, an AI tool must win on quality or workflow, not on obvious reimbursement arbitrage. The site-network business faces a different but related constraint set. It benefits from community-site scale and sponsor urgency, but public proof is still concentrated in IBD. Expansion into hepatology, obesity, and cardiology is strategically promising yet less evidenced. The diligence takeaway is therefore asymmetric: the markets are real, and Iterative appears to be selling into genuine pain points, but precise TAM/SAM/SOM math should stay deliberately constrained until management discloses pricing, installed base, sponsor contract economics, and non-GI replication data.[CM029, CM030, CM033, CM034, CM035, CM036]

3.1 Endoscopy AI rivals and OEM channel power

In the procedure room, Iterative is not competing against another small GI software startup so much as a mix of large medtech incumbents and one focused pure-play. Medtronic markets GI Genius as an AI-assisted colonoscopy module, while Olympus now spans CADDIE, OLYSENSE, and ENDO-AID inside its broader EVIS X1 ecosystem. Fujifilm positions CAD EYE inside its own processors and scope stack, and Odin Vision remains the closest pure-play cloud competitor with CADDIE. Those products address the same clinical job as SKOUT: real-time polyp detection support during colonoscopy. The strategic difference is distribution power. Olympus and Fujifilm can bundle AI into branded endoscopy equipment pathways, service relationships, and upgrade cycles that already exist inside hospitals and ASCs. Medtronic looks more modular because GI Genius is sold as a stand-alone intelligent endoscopy module, but it still brings a scaled commercial organization and a familiar device-buying motion. Iterative therefore enters the room with less OEM leverage than the installed-base incumbents even when its clinical positioning is credible. Independent evidence keeps the category attractive but not settled. Yale's review and the AGA guideline both support higher adenoma detection with CADe, yet they also preserve uncertainty on advanced lesions, long-term cancer outcomes, cost-effectiveness, and universal adoption. That matters competitively because it leaves a durable status-quo substitute: standard colonoscopy plus conventional quality improvement. Buyers who are not yet convinced on long-horizon outcome lift can still defer AI entirely, or trial it only where quality pressure is most acute.[CP001, CP003, CP004, CP005, CP006, CP008]

3.2 GI and IBD trial-network rivals

On the sponsor-facing side, Iterative runs into much larger competitors than it does in procedure-room AI. IQVIA and ICON both market GI or IBD-specific development expertise on top of global site-network and patient access infrastructure. Alimentiv attacks from the opposite direction: it is narrower than the global CROs but markets itself as a deeply specialized GI operator with endoscopy, histology, and IBD workflow depth. All three rival classes can sell against Iterative when the buyer cares more about sponsor breadth, geographic reach, or established outsourcing relationships than about Iterative's GI-first operating narrative. The scale gap is public. IQVIA markets more than 1,100 high-performing sites across 100 countries and claims outsized share in outsourced phase II and III IBD work. ICON markets 114-plus GI studies in the last five years plus the Accellacare network's 8.1 million patient reach, 50-plus sites, and faster site-initiation metrics. Alimentiv markets 5,000-plus site relationships across 60-plus countries and says it supports more than 70% of IBD compounds in development. Iterative's own materials show meaningful momentum, but public sponsor-access scale is still smaller than those incumbent networks. Provider-controlled site supply is the other competitive pressure. GI Alliance and One GI are partners today, not adversaries in the narrow sense, but their disclosed networks show why large GI groups matter strategically: they control investigators, coordinators, patient flow, and local trial access. If those relationships are not durably exclusive, the same asset that differentiates Iterative can be contested, internalized by provider groups, or redirected toward another CRO or operator.[CP024, CP025, CP026, CP027, CP028, CP029]

3.3 Packaging, pricing opacity, and buyer alternatives

Public commercial packaging is much clearer than public pricing. Iterative, Medtronic, Olympus, Fujifilm, IQVIA, ICON, and Alimentiv all present contact-led enterprise selling rather than transparent list pricing. In practice that means buyers evaluate these offerings through bundled equipment, services statements of work, site-network access, or custom enterprise contracts rather than through a published per-procedure or per-site-rate card. The opacity itself is strategically relevant because it favors scaled sellers with heavier field organizations and more room for negotiated bundling. The substitute set is also broader than a normal one-category comparison table suggests. A GI provider can buy a CADe module from Medtronic, Olympus, or Fujifilm without buying Iterative's site-network services. A sponsor can buy GI or IBD trial execution from IQVIA, ICON, or Alimentiv without buying Iterative's procedure-room AI. And a buyer can still remain on the status quo: no AI in the endoscopy room, or fragmented direct-to-site trial operations supported by internal teams and generic CRO processes. Those alternatives make multi-homing uneven across the two arenas. Trial services are more multi-homeable because sponsors already split work by protocol, geography, or function. Endoscopy AI is harder to multi-home once a site standardizes on a hardware ecosystem, workflow, and service partner. That asymmetry is why hardware bundling and provider-network access matter more to moat durability than any single public efficacy chart.[CP023, CP030, CP041, CP042, CP043, CP044]

3.4 Moat durability and adverse competitive risk

Iterative's best public competitive story is not that it has the single best CADe product or the single largest trial network. It is that it appears to combine two usually separate capabilities: AI tooling that can help in GI workflows, and embedded community-site operations that can help sponsors recruit and activate studies faster. That hybrid positioning is real enough to matter, and it is unusual in the public set reviewed for this chapter. But the durability of that moat is still conditional. On the AI side, Olympus and Fujifilm can deepen channel lock-in by shipping more intelligence inside installed endoscopy ecosystems, while Medtronic can use modular distribution to compete wherever buyers want a lighter add-on. On the trial side, IQVIA and ICON can out-scale Iterative in sponsor coverage and geographic reach, while Alimentiv can challenge the idea that Iterative alone offers GI-specific depth. Provider networks such as GI Alliance and One GI further remind investors that site supply is negotiated, not owned. The adverse evidence base sharpens that risk rather than eliminating it. AGA still withholds a recommendation for or against CADe-assisted colonoscopy, the pilot literature still shows extra non-neoplastic polyp detection, and public materials still do not disclose installed SKOUT share, sponsor win rates versus named rivals, or actual pricing across either arena. The investment conclusion is therefore nuanced: the integrated model is different, but it is not yet protected by public proof strong enough to dismiss incumbent retaliation or commoditization.[CP019, CP020, CP022, CP024, CP041, CP043]

3.5 Exhibits

4.1 Revenue model, monetization surfaces, and pricing opacity

Iterative Health’s public materials support a multi-surface monetization model, but not a publicly auditable revenue mix. The company now sells a sponsor- and site-facing research infrastructure proposition: centralized access to a growing network of community research sites, operational support, trial management, and AI-enabled workflow tools. GI Alliance, One GI, and US Heart & Vascular all describe Iterative as more than a software vendor. In each case the company contributes trial-management know-how, operational backbone, site enablement, and sponsor access rather than just a dashboard or algorithm. That makes sponsor/CRO services and provider-network operating partnerships the clearest observable revenue streams. SKOUT adds a second commercial surface. It proves Iterative still has a real regulated product that can be sold into GI practices and endoscopy settings, but the public record does not show list pricing, reimbursement schedules, or installed-base economics. The same opacity applies to the services side: no rate cards, no disclosed contract terms, no revenue split between sponsor budgets, provider partnerships, or device sales. The strongest conclusion is therefore directional rather than numeric. Iterative almost certainly monetizes across sponsor services, provider-network economics, AI-enabled workflow, and SKOUT, but the company has not disclosed the pricing architecture needed to distinguish recurring software-like revenue from project-based or service-led revenue.[CI001, CI002, CI003, CI004, CI005, CI006]

4.2 Cost structure, reimbursement friction, and unit-economics proxies

The model looks operationally powerful but financially heavier than pure software. Iterative’s own language emphasizes centralized operations, expert staffing, business development, site support, regulatory functions, and active research-site management. The NextStage acquisition release is especially revealing because it describes a 250-person organization spanning site staff and centralized functions across multiple therapeutic areas. That implies meaningful labor cost, onboarding cost, and corporate overhead even before considering product R&D or future acquisitions. Public reimbursement context adds another layer of friction. Medicare does cover screening colonoscopy, but the reviewed sources do not show a durable standalone payment pathway for SKOUT-like colonoscopy AI. Meanwhile, AGA and Becker’s describe worsening GI reimbursement pressure, especially for ASC and hospital-based endoscopy, and Boston Scientific’s guide underscores how strongly economics differ by site of service and by negotiated private-payer contract. Those facts matter because they can limit what provider customers are willing to pay for add-on technology or ancillary services. Unit-economics disclosure remains partial. Iterative’s best public proxies are site count, partner count, activation speed, and enrollment performance. Those suggest the product-service bundle is operationally valuable. But without CAC, LTV, payback, sponsor repeat-rate, or revenue per site, investors cannot convert those operating wins into a reliable underwriting model.[CI015, CI016, CI017, CI018, CI019, CI020]

4.3 Capital adequacy, funding dependency, and public-comparable context

Iterative is not obviously capital-starved, but it is still publicly opaque. The one hard primary capital fact is the $77 million Series C, which management explicitly linked to expansion into cardiology, obesity, geography, and deeper provider-network partnerships. Secondary sources converge around a post-money valuation of roughly $1.3-1.4 billion and total funding of roughly $270-273 million, but those figures remain triangulations rather than company disclosures. Official sources still do not publish current cash, burn, debt, or runway. That omission matters more because the company is expanding operationally, not just digitally. Site acquisitions, multispecialty buildout, and a service-heavy delivery model usually consume cash before investors can observe attractive margin profiles. Public CRO comparables show what good looks like once the model reaches scale. IQVIA, ICON, and Medpace each report billions of revenue, explicit liquidity, and margin/backlog data; their reported operating or EBITDA margins cluster around the low teens to low twenties. But those same public comps also demonstrate that research-services accounting is complex, with working-capital swings and, in ICON’s case, a revenue-recognition restatement and material weaknesses. So the capital question is less “Can Iterative still raise money?” and more “Has Iterative already built a self-funding machine?” Publicly, that answer is still unknown.[CI025, CI026, CI027, CI028, CI029, CI030]

4.4 Financial diligence blockers and underwriting verdict

The public record is good enough to map the business model, but not good enough to underwrite it. Investors can reasonably conclude that Iterative has real commercial demand, real sponsor and provider relationships, and at least one regulated product surface in SKOUT. They can also conclude that the company has enough financing access to keep expanding. What they cannot conclude is how much of the model is recurring, what direct margins look like, how quickly site economics pay back, or how much cash remains after recent expansion. The most material blockers are straightforward: no revenue or ARR disclosure; no revenue split by stream; no gross-margin or direct-cost split; no cash, burn, runway, or debt schedule; no public price cards; and no sponsor retention, backlog, or customer concentration disclosure. Conflicting headcount and valuation proxies reinforce the point that secondary sources are only approximations. The financial verdict is therefore cautious but not negative. Iterative looks commercially credible and strategically well financed, yet still behaves like a private company that expects investors to trust operating momentum more than disclosed economics. For diligence, that means Chapter 4 should treat upside as plausible, but capital adequacy and unit-economics confidence as unproven until management shares actual financial statements and contract metrics.[CI014, CI024, CI026, CI028, CI032, CI045]

4.5 Exhibits

5.1 Product stack and customer workflow

Iterative Health does not present itself as a single-product medtech vendor. Across its homepage, about page, and current hiring description, the company describes a blended healthcare technology and services platform that combines community-embedded research infrastructure, AI-enabled disease-assessment tooling, and at least one regulated point-of-care device. In customer workflow terms, the stack serves two connected user groups. Sponsors and CROs use Iterative to activate sites faster, reduce operational burden, and improve enrollment in GI-adjacent trials. Providers and endoscopists use SKOUT during screening and surveillance colonoscopy as an assistive device inside the procedure room. The result is a company whose commercial promise is not just algorithm performance, but a wider operating model that links patient identification, site operations, endoscopic assessment, and sponsor reporting. The publicly visible modules reinforce that view. SKOUT is the most concrete product surface because it has an FDA-cleared indication, a GUDID record, and a published workflow description. But the broader platform is also visible in the company’s repeated emphasis on a global site network embedded into community care, operationally supported activation and prescreening workflows, AI-enabled endoscopy and histology analysis for ulcerative colitis, and a hybrid AI-human central-reading paradigm for trial endpoints. That breadth matters strategically. It means Iterative is trying to own more of the GI research workflow than a single detection algorithm can capture, which is an important differentiator against category peers that mainly sell one CADe module.[CE001, CE002, CE003, CE004, CE005, CE018]

5.2 Architecture, integration, and operating model

SKOUT’s technical architecture is more concretely documented than the rest of the stack. The FDA clearance and GUDID record describe a real-time software layer that receives endoscopy video over SDI, analyzes the feed during colonoscopy, and overlays a blue rectangular outline when a potential colorectal polyp is detected. The same sources show that the product is designed to fit directly into procedure workflow rather than forcing a separate review environment: users can switch between a bypass endoscopic feed and the SKOUT feed, and the system pauses detection when tools enter the field or lighting is inadequate. Compatibility is currently bounded by specific processor ecosystems, notably Olympus EVIS EXERA II/III and Fujifilm Eluxeo VP-7000 outputs, which makes integration quality and installed equipment mix an immediate adoption dependency. The broader operating architecture is more services-heavy. Iterative’s site-network publication describes activation, budgeting, prescreening, training, and regulatory documentation as the operational bottlenecks it tries to centralize or support. The IBD endpoint tooling then sits on top of that workflow: AI-enabled endoscopy and histology analysis adds structured disease-assessment output, while the BMJ-described 2M+1H framework inserts two independent ML models and a gastroenterologist adjudicator into central reading rather than replacing human review outright. The same architecture makes distribution and documentation partnerships material. Iterative’s FDA-clearance announcement ties SKOUT commercialization to Provation, whose GI workflow footprint can reduce go-to-market friction. Put simply, Iterative’s technical moat is not a single model alone; it is the combination of device integration, operational enablement, human oversight, and workflow adjacency across research sites and endoscopy rooms.[CE006, CE007, CE008, CE009, CE010, CE018]

5.3 Trust, evidence, and competitive context

Iterative’s trust story starts with regulatory and workflow guardrails rather than pure AI novelty. SKOUT now has 510(k) clearance as a Class II gastrointestinal lesion software detection system and is listed as commercially distributed in AccessGUDID. The device labeling is explicit that it assists rather than replaces trained gastroenterologists, and the user interface includes status signaling plus a paused-detection state when tools enter the field. Those controls matter because they position SKOUT as an augmentation layer inside a regulated procedure, not an autonomous diagnostic decision-maker. Public guidance is directionally supportive: Yale’s summary of draft AGA guidance says CADe receives a conditional recommendation in adult colonoscopy and does not materially increase procedure time, which broadly supports the adoption case for the category that SKOUT serves. That said, the evidence base still has nuance. Iterative’s official materials and its partner brief highlight strong efficacy signals such as no procedure-time increase, one extra adenoma resected per 4.5 patients, and better performance on sessile or mid-sized lesions, while the earlier SKOUT study and broader CADe literature support higher ADR/APC. But the literature also warns about false positives, automation bias, operator deskilling, generalizability, and long-term outcomes. Competitive materials from Medtronic, Olympus, and Fujifilm show that real-time AI detection, sessile-lesion emphasis, and workflow-preservation claims are no longer unique to Iterative. That means trust in Iterative’s stack depends not only on category-level AI efficacy, but on how well its combination of distribution, integration, site operations, and human-in-the-loop trial workflows stands up against other CADe vendors and operational alternatives.[CE012, CE013, CE014, CE015, CE016, CE032]

5.4 Roadmap, maturity, and critical dependencies

The public roadmap is visible mainly through milestone evidence rather than a forward product plan. On the device side, Iterative progressed from early SKOUT efficacy data to a larger randomized study cited in its value-analysis materials, then to 2025 FDA clearance and commercial distribution status. On the trial-tech side, the 2025-2026 publication cadence matters: the site-network paper and DDW 2026 update emphasize scaled operational performance, the ulcerative-colitis endoscopy-plus-histology work shows Iterative expanding from procedural assist into endpoint analytics, and the BMJ central-reading paper shows a more regulatory-minded human-plus-model framework for high-stakes trial scoring. Those are meaningful maturity signals because they suggest a roadmap centered on expanding along the GI research workflow rather than merely adding features to SKOUT. The main dependencies are equally clear. SKOUT adoption depends on compatible processor footprints and on Provation-assisted commercial reach. The site-network thesis depends on continued uptake by community sites, sponsor willingness to route studies through a site-centric model, and enough trained operational staff to manage startup, prescreening, and activation. The central-reading and disease-assessment tools depend on curated video and histology data, maintainable model performance, and access to expert human adjudicators. Finally, the public record still leaves some business-critical blanks: there is no disclosed installed base, no public pricing, no exhaustive interoperability matrix, and no detailed forward release schedule for either the device or the IBD toolchain. So while Iterative’s product maturity is stronger than a one-feature startup, it still has material disclosure gaps investors would want to close before underwriting technical scale-up at full confidence.[CE011, CE017, CE022, CE023, CE041, CE042]

5.5 Exhibits

6.1 Customer segmentation by buyer, user, and payer

Iterative Health sells into a linked but non-identical customer stack. On the provider side, the company courts physician groups, management services organizations, principal investigators, and research-site operators that want to build or scale clinical research as a profitable ancillary service. On the sponsor side, Iterative markets directly to sponsors, CROs, and biotech companies that need faster activation, enrollment forecasting, and a performance-managed site network. A third, less transparent buyer group sits around SKOUT: GI endoscopists and provider organizations evaluating AI-assisted colonoscopy equipment. Public proof is strongest for the first two layers and much weaker for the third. That segmentation matters because Iterative is not just selling software seats. It is selling operational leverage, patient identification, and access to sites that can execute consistently. The provider pitch emphasizes centralized business operations, staffing, regulatory help, training, and sponsor access; the sponsor pitch emphasizes high-performing sites and accelerated trial execution. The company now presents the network as multispecialty rather than GI-only, spanning GI, hepatology, obesity, and cardiology, with 100+ sites and 40+ sponsor/CRO/device partners. But named customer evidence remains concentrated in large U.S. provider networks such as GI Alliance, One GI, US Heart & Vascular, Gastro Health, and the site-level case study with Gastro One, while sponsor-side named proof is mostly limited to Takeda and aggregate counts.[CU001, CU002, CU003, CU006, CU007, CU008]

6.2 Adoption trajectory and deployment scale

The public adoption arc is visible through network milestones rather than customer-count disclosures by cohort. In April 2025, GI Alliance said adding its 21 active research sites and 80+ active Phase II-IV trials pushed the Iterative network to more than 70 locations in the U.S. and Europe. By the September 2025 One GI partnership, Iterative had become an embedded operating partner for a 34-clinic network with 13 active research sites. By April 2026, the US Heart & Vascular partnership repositioned the company as a multispecialty network entering cardiology, and by the April 2026 Series C announcement Iterative was citing 100+ research sites across North America, Europe, India, and Australia plus 40+ pharmaceutical, biotech, medical-device, and CRO partners. The strongest deployment proof comes from trial-performance metrics rather than account counts. Iterative’s provider page cites 2x faster activation, 3.4x higher patient enrollments, and AI-powered prescreening that lifts randomizations by 40%. The JCC summary and later DDW 2026 release add more specific operating numbers: 74 days from site selection to activation, 45 days to first patient screened, 83 days to first randomized, roughly 0.33-0.34 patients randomized per site per month, and four completed trials that still delivered about 0.32 patients per site per month at full enrollment. Those are high-quality adoption proxies because they show the network continuing to function as denominator and study count rose, even though Iterative still does not publish standard customer cohort data by sponsor, site, or product line.[CU003, CU004, CU005, CU010, CU014, CU015]

6.3 Named customer proof and evidence quality

Named customer proof is credible but uneven. On the provider side, GI Alliance, One GI, US Heart & Vascular, Gastro Health, and Gastro One all appear in public materials with either executive quotes, site counts, trial counts, or quantified workflow outcomes. GI Alliance and One GI provide the cleanest platform-level evidence because both the company and the customer-side organizations describe the relationship, and both quantify the attached site footprint. Gastro One provides the richest single-site operating case study, including faster activation, randomizations, and the share of enrollments sourced through Iterative. US Heart & Vascular broadens the thesis from GI into cardiology, while Gastro Health demonstrates that Iterative can land through AI-recruitment tooling before broader network rollout. Sponsor-side proof is materially thinner. Takeda is the only named sponsor relationship found in public materials, and even there the evidence is framed around AI-enabled trial design and disease assessment rather than explicit contract scope or repeat-study volume. The company’s 40+ sponsor/CRO/device-partner count is useful for breadth, but it does not substitute for named deployments, study counts per sponsor, or sponsor-level retention data. Public proof is also weak for SKOUT as a customer business. The product page contains anonymous testimonials and efficacy claims but no named health-system or ambulatory-surgery-center deployment references. So the chapter’s evidence quality is best described as strong for provider-network logos, good for trial-performance outcomes, limited for sponsor naming, and weak for named product installations.[CU006, CU007, CU008, CU009, CU010, CU011]

6.4 Retention, repeat usage, and durability proxies

Iterative does not disclose NRR, GRR, logo churn, renewal rate, contract duration, or public review-platform scores, so the retention lens has to rely on indirect signals. The best repeat-usage proxies are longitudinal. Ziad Younes says Iterative has been a trusted partner to his site for several years, which is stronger than a one-off implementation quote. GI Alliance’s chief medical officer described the DDW 2026 results as evidence of what became possible one year into the strategic partnership. The DDW release also calls the 2026 presentation the second major public reporting of aggregated network performance and highlights that four studies completed full enrollment with consistent randomization rates. These are useful signals because they indicate Iterative is not merely announcing logos, but continuing to execute after onboarding. Still, investors should not confuse operating consistency with customer retention proof. Clinical trials are episodic, sponsors often multi-home across site networks, and provider groups may keep several research or technology relationships alive at once. Public evidence emphasizes activation, prescreening, chart-review relief, and within-trial patient retention; it does not show how many sponsors return with new protocols, how many provider groups expand contract scope after year one, or what percentage of revenue renews automatically. The blocked OUP abstract page also means the direct conference record is less accessible than the surrounding company summaries and syndicated press coverage. In short, durability looks directionally positive, but the hard retention dataset remains undisclosed.[CU023, CU024, CU025, CU027, CU028, CU029]

6.5 Expansion paths and concentration risks

Iterative’s expansion logic is clear. It can add sites within existing provider networks, take the same network model from GI into obesity and cardiology, deepen sponsor relationships as trial performance data accumulates, and potentially cross-sell AI tooling such as AIR and disease-assessment workflows alongside site operations. The 2026 move into cardiology with US Heart & Vascular and the Series C message around obesity and broader multispecialty growth both support this view. If the company can keep its activation and enrollment advantage while moving into adjacent therapeutic areas, existing sponsors may have reason to consolidate more trial work inside the network. The risk is that public proof is concentrated in a few large partner ecosystems and still rests heavily on partner-authored announcements. GI Alliance, One GI, US Heart & Vascular, Gastro Health, and Gastro One dominate the named-provider record; sponsor-side naming is sparse; and SKOUT installations are opaque. That leaves unanswered questions around top-account revenue concentration, sponsor renewal frequency, and whether provider groups are exclusive or simply one of several parallel research channels. Procurement friction is another real risk because Iterative’s value proposition depends on staffing, regulatory work, budgeting, and chart-review support; sites or sponsors that internalize those functions may bargain harder on price or split work across multiple vendors. Device-side competition compounds that issue: Medtronic’s GI Genius publicly markets strong detection claims, so SKOUT adoption likely depends on workflow fit and network relationships rather than an uncontested product category.[CU018, CU034, CU035, CU036, CU037, CU038]

6.6 Exhibits

7.1 Regulatory, legal, and reimbursement risk

Iterative’s regulatory and legal risk profile is shaped less by a visible enforcement event today than by an unsettled operating perimeter. SKOUT participates in a category where clinical evidence is directionally positive but still debated on generalizability, false positives, false negatives, deskilling, and long-term outcomes. BMJ’s 2025 review goes further, framing AI endoscopy as a responsibility problem that now spans clinicians, institutions, developers, and regulators. That matters because major bodies still diverge on routine CADe use, which means the category has not yet settled into an uncontroversial standard of care. If adverse events or disappointing real-world evidence emerge, accountability will not sit neatly with a single actor. The reimbursement side is equally important. Medicare clearly covers screening colonoscopy, but reviewed policy and legal sources do not show a stable, device-specific payment pathway for endoscopy AI. Instead, the policy discussion is still at the level of proposed legislation and generic payment frameworks for AI-enabled medical devices. At the same time, GI provider economics are under pressure: AGA and Becker’s describe cuts and longer-term erosion in endoscopy reimbursement that can reduce buyer appetite for adjunct technology and operating overhead. Iterative does have basic compliance infrastructure—privacy policy, DPO, service-provider disclosures, and general security language—but its public security posture remains high-level. In practical diligence terms, the regulatory question is not whether Iterative has zero compliance process; it is whether that process is robust enough to defend a data-heavy AI and site-operations model if scrutiny increases.[CR010, CR011, CR013, CR014, CR016, CR017]

7.2 Operational, quality, and security risk

The core operating bet is that Iterative can preserve trial-performance outperformance while scaling a high-touch network past 100 sites and into new specialties. Public data support the upside case: the company cites 2x faster activation, 3.4x higher enrollment, and DDW 2026 results that held roughly 3x benchmark randomization performance at larger scale. That is meaningful mitigation against the idea that the model is merely a collection of pilot wins. But the same sources also reveal why the model is fragile. Iterative is not just shipping a device or SaaS dashboard; it is providing budgets, legal and regulatory support, training, sponsor engagement, study startup, and staff management. That means quality drift can come from labor, process, or site-governance failure even if the underlying software remains stable. On the AI side, the operational risk is not simply whether SKOUT or related CADe tools work in a study. It is whether performance stays reliable in day-to-day use across diverse users, equipment, case mixes, and clinical cultures. Reviews cite false positives, false negatives, dataset bias, operator over-reliance, and weak real-world reproducibility in some analyses. The company’s own provider materials emphasize ongoing PI and CRC training, while external interview coverage says the network took years to build and still needs constant improvement. That combination argues for a balanced view: operational evidence is better than many startups have, but the model remains execution-heavy and therefore highly sensitive to staffing quality, training discipline, and process control. Data security belongs in the same bucket because research operations, sponsor coordination, and AI workflows all increase data-handling complexity, while public security disclosures remain thin.[CR001, CR002, CR006, CR011, CR013, CR014]

7.3 Partner, dependency, and people risk

Iterative’s public traction is real, but it is concentrated. GI Alliance, One GI, and US Heart & Vascular dominate the named-provider record, and each relationship is strategically meaningful enough that any one of them could matter disproportionately to growth perception, sponsor access, and pipeline density. GI Alliance brings 21 active sites and 80+ trials; One GI brings 34 clinics and 13 active sites; USHV opens cardiology and reinforces multispecialty ambitions. Those are strengths, but they are also dependencies. If a major partner underperforms, reprices, delays expansion, or simply proves less economically important than headline site counts suggest, the external story can deteriorate quickly because so much of the proof set lives inside a few ecosystems. There is a second dependency layer at the product and workflow level. Medtronic and Olympus are not only alternative AI vendors; they are large incumbents that can bundle training, hardware compatibility, and installed-base leverage. Olympus explicitly ties CADDIE to its hub and white-light workflow, while Medtronic markets GI Genius on sensitivity and low false positives. Iterative therefore has to win not just on model quality but on channel access and workflow fit. People risk sits on top of all of this. The management bench is clearly improving—there is now a President/CFO, COO, medical leadership, and legal leadership—but the bench still looks recently assembled rather than battle-tested over a long period at current scale. Founder dependence on Jonathan Ng remains meaningful because the strategy still requires synchronized execution across AI, provider partnerships, sponsor GTM, and multispecialty expansion.[CR003, CR004, CR005, CR007, CR008, CR021]

7.4 Financial/model risk, mitigations, and thesis-break triggers

Financial-model risk is high mainly because the operating story is much more visible than the economics. Public materials support network scale, partner count, and performance metrics, and third-party coverage suggests strong capitalization after the Series C. But there is still no public revenue mix, margin profile, burn, renewal rate, top-account concentration, or SKOUT installed-base disclosure. That leaves investors underwriting a complex services-plus-AI model with limited direct visibility into whether value accrues mostly to sponsor services, provider-network operations, SKOUT sales, or some combination. The reported valuation backdrop only sharpens this problem: a company can be well capitalized and still be economically hard to underwrite if the key monetization levers remain opaque. The practical investment question is therefore not whether Iterative has risks—every clinical-infrastructure business does—but whether those risks are monitorable and whether the existing mitigations are strong enough to justify residual exposure. The best mitigations are tangible: real partner networks, evidence that performance held at larger scale, a broader executive bench, and at least baseline privacy/compliance infrastructure. The main thesis-break triggers are equally tangible: operating metrics revert toward industry norms, a major provider ecosystem stalls or churns, or AI reimbursement and customer economics never support durable pricing. If any of those happen, the same features that make Iterative attractive today—its integrated model and dependence on partner networks—can transmit risk quickly into growth, gross margin, and valuation.[CR016, CR018, CR019, CR020, CR031, CR032]

7.5 Exhibits

8.1 Investment thesis, anti-thesis, and recommendation call

Iterative Health has earned a serious valuation conversation because the operating proof is better than the average AI-health startup. Official materials show a research network of more than 100 sites, over 40 sponsor and CRO relationships, and repeated claims of 2x faster activation plus 3x enrollment outperformance versus industry benchmarks. The DDW 2026 update is especially important because it suggests the thesis held at larger scale rather than only inside a pilot cohort. That combination supports the positive side of the case: Iterative appears to have built a differentiated, site-centric clinical research network with real workflow and sponsor relevance, not just a narrow endoscopy algorithm story. The anti-thesis is that the public record prices the upside much more aggressively than it explains the economics. Third-party sources cluster around a $1.3 billion to $1.4 billion mark, but the company did not disclose revenue, ARR, gross margin, or retention in the latest financing or performance updates. Public evidence therefore supports company quality more than it supports underwriting precision. In today’s market, that distinction matters: premium multiples still exist for AI and data assets, but only when revenue quality, durability, and monetization clarity are visible. Because Iterative’s price looks fair only in a strong execution case and stretched in a more CRO-like case, the most defensible public-evidence recommendation is research-more, with medium confidence, high risk, and a stretched valuation stance.[CV003, CV004, CV005, CV014, CV017, CV018]

8.2 Current valuation context and comparable set

Current valuation context is visible enough to frame discipline but not enough to justify aggressive conviction. Forge shows a $1.3 billion Series C valuation in April 2026, while TechCrunch listed Iterative at $1.4 billion on its 2026 unicorn tracker; ai2.work also reported an approximately $1.3 billion mark. Official company materials confirm the $77 million round, but they do not confirm the post-money value. Forge adds only partial financing-term clues, including a 1.0x non-participating preference and 8% dividend rate, so downside protection and dilution remain incomplete in the public record. Public comparables argue for caution. IQVIA and ICON trade near roughly 1x-2x revenue, while Medpace trades near roughly 5x revenue and also shows the best publicly visible growth-and-margin combination of the set. Meanwhile, current 2026 M&A commentary suggests that quality healthtech names clear around 4x-6x revenue and only premium AI/data assets push into 6x-8x-plus territory. That means Iterative’s current mark can be defended only if the business is already much closer to premium AI/data economics than to a services-heavy CRO profile. Because revenue is undisclosed, that cannot be established from public evidence alone.[CV009, CV010, CV011, CV014, CV015, CV016]

8.3 Bull, base, and bear scenario range

All scenario work here is explicitly assumption-based because current revenue and margin are undisclosed. The bull case assumes Iterative converts network proof into multi-specialty revenue, keeps performance materially above benchmark, and earns an upper-band AI/data multiple. Under that path, an exit around $1.8 billion to $2.6 billion is possible, but even that outcome does not create a classic late-stage venture return from a $1.3 billion to $1.4 billion starting mark. The base case is more conservative: Iterative scales as a strong CRO/data hybrid, lands around $160 million to $220 million of revenue, and exits around $0.9 billion to $1.5 billion on a 5x-7x multiple. The bear case matters because today’s private-mark environment is less forgiving than in 2021. If reimbursement frictions, partner concentration, or execution slippage push Iterative toward CRO-like economics, the mark could compress into roughly $0.4 billion to $0.9 billion. That downside is not a remote fantasy; PitchBook and Finro both describe a market that now reprices companies when monetization is thin or stale marks finally get tested. The probability-weighted picture is therefore closer to “already mostly priced” than to “obvious bargain,” which is why the chapter lands on research-more and stretched rather than fair and buy.[CV023, CV024, CV025, CV027, CV036, CV037]

8.4 Final diligence asks and thesis-break triggers

The remaining blockers are not cosmetic. To move from research-more toward track or buy, investors need current revenue, gross margin, retention, concentration, and preference-stack data—not another partner logo or throughput anecdote. Reimbursement also matters more than the company narrative implies. The legal and policy sources reviewed here still describe Medicare reimbursement for AI-enabled devices as an evolving policy project, while GI payment sources describe real pressure on endoscopy economics. That does not kill SKOUT or the broader platform thesis, but it does mean the valuation case should assume monetization friction until management proves otherwise. The monitoring framework is straightforward. The thesis breaks if actual economics prove far weaker than the current mark requires, if network performance falls back toward benchmark, or if multispecialty expansion fails to convert into real revenue. Conversely, the call could improve if diligence shows current revenue already supports premium AI/data multiple territory, sponsor and provider concentration are acceptable, and the cap table does not contain unusually punitive terms. Until then, the right IC posture is disciplined curiosity rather than conviction.[CV030, CV031, CV032, CV033, CV034, CV035]