Aidoc

1.1 Identity, footprint, and product architecture

Aidoc’s public materials consistently frame the company as a clinical AI vendor rather than a narrow radiology-algorithm shop. The official about page says the company is focused on helping healthcare teams optimize patient treatment and economic value, while the aiOS and CARE pages show how that positioning has widened from image triage into a broader enterprise workflow layer. aiOS is described as the control plane that runs, orchestrates, and governs clinical AI across a health system, connecting PACS, EHR, mobile, and care tools. CARE, in turn, is positioned as the multimodal foundation model that lets Aidoc move beyond single-condition tools toward broader disease coverage, automated measurement, and eventually draft-report generation. This matters because it changes the company’s investment case: Aidoc is no longer selling only acute-finding alerts to radiologists, but an operating layer for multi-algorithm deployment, care coordination, and model governance. Fetched sources support a clear New York commercial posture through the 2026 financing announcements, while Guy Reiner’s official bio still explicitly identifies him as general manager of the Tel Aviv branch. The resulting picture is a company with Israeli technical roots, a visible New York-facing capital-markets presence, and a strategy built around enterprise deployment rather than departmental point solutions.[CO001, CO006, CO007, CO008, CO009, CO010]

1.2 Founders, leadership, and key-person concentration

Aidoc’s founder set is unusually well corroborated for a private company. The official leadership bios identify Elad Walach as co-founder and CEO, Michael Braginsky as co-founder and CTO, and Guy Reiner as co-founder, chief architect, and general manager of the Tel Aviv branch. CTech’s July 2025 reporting matches those roles and dates the founding to 2016. Functionally, the trio still appears to anchor the core business model: Walach is the external spokesperson and financing voice, Braginsky appears in both technical and funding coverage as the foundation-model architect, and Reiner’s remit over the Tel Aviv branch implies deep continuity in product and engineering execution. Publicly visible non-founder executives exist, but they are much less comprehensively disclosed; the record surfaced here is strongest for founder leadership and much thinner for the broader management bench. That makes key-person concentration a real diligence item rather than a generic startup caveat. Aidoc’s public materials do show growing organizational maturity—such as explicit discussion of governance, drift detection, and enterprise monitoring—but the public record still does not provide a fully exhaustive leadership roster or a refresh-to-refresh clean view of how authority is distributed below the founders.[CO002, CO003, CO004, CO005, CO034, CO035]

1.3 Funding history, investor mix, and valuation visibility

Aidoc’s capital path has accelerated materially in the last two years. The official 2022 Series D announcement recorded a $110 million round that brought cumulative funding to $250 million. By July 2025, PR Newswire, Calcalist, and Globes all reported a new $150 million growth financing led by General Catalyst and Square Peg, with NVentures and several major U.S. health systems also participating; both PR Newswire and Globes referenced an additional $40 million revolving credit facility and a cumulative total of $370 million. Then, in April 2026, Aidoc and Goldman Sachs Alternatives each announced a $150 million Series E led by Goldman, with General Catalyst, SoftBank Vision Fund 2, and NVentures joining. Those April 2026 materials state only that total funding is now “over $500 million,” not the exact figure. That distinction matters because it is enough to establish serious capitalization but not enough to settle valuation. The strongest post-2024 source on that point is actually Globes’ July 2025 interview, where Michael Braginsky explicitly said the company would not disclose the valuation of the financing round. As a result, the company’s private-unicorn status remains a genuine evidence gap in this run: high funding and rising scale are clear, but no fetched source dated after 2024-05-20 disclosed a valuation that would let this report confirm a current unicorn mark.[CO015, CO016, CO017, CO018, CO019, CO020]

1.4 Scale signals, named customers, and regulatory milestones

Aidoc’s public scale claims are now large enough that the company should be analyzed as an enterprise deployment story rather than a pilot-stage vendor. The 2025 CARE financing release said Aidoc supported more than 45 million patients a year across 150+ health systems and expected that figure to reach 100 million within three years. The 2026 Series E materials moved the disclosed scale higher still, claiming more than 60 million annual patient cases, more than 110 million patient cases analyzed cumulatively, and deployment across nearly 2,000 hospitals. Those numbers are self-reported, but the supporting customer proof is stronger than it was in earlier rounds. Asklepios disclosed a 28-hospital rollout across Germany with roughly 35,000 CT and X-ray images analyzed monthly, while Hartford HealthCare said Aidoc’s 17-algorithm deployment reached go-live in three weeks and spanned millions of patient exams. The regulatory record is also tangible rather than purely narrative. FDA K213721 shows a 2022 clearance for brain aneurysm triage, and K231631 shows a 2023 CAC quantification clearance. Aidoc’s 2025 and 2026 press releases then move from indication-by-indication clearances toward foundation-model framing: first a rib-fractures clearance on CARE1, then a comprehensive body CT triage clearance that groups 11 new indications with three existing ones into one workflow. That combination of customer proof and regulatory throughput is the core reason later chapters can treat Aidoc as a scaled clinical-AI infrastructure company instead of a single-product startup.[CO022, CO023, CO024, CO025, CO026, CO027]

1.5 Adverse signals and unresolved evidence gaps

The strongest negative evidence fetched in this run is not a lawsuit or warning letter, but a combination of technical and market constraints that limit how aggressively the company’s claims should be read. Aidoc’s own model card is more candid than its marketing pages about what remains unobserved: important demographic attributes are not available inside DICOM, so bias monitoring has to rely on proxies and post-market performance tracking rather than full direct measurement. External evidence also reinforces the need for caution. A May 2026 real-world pulmonary-embolism study summary said Aidoc’s algorithm matched radiologists in 97.8% of 32,501 scans, but also missed 15% of confirmed PE cases and lost most adjudicated disagreements, preserving the need for human oversight. Separately, reimbursement literature on generalist radiology AI says current payment frameworks still fit these products poorly, which means Aidoc’s ROI story likely depends more on error avoidance, throughput, and downstream care management than on direct payment codes. Most importantly for company-level diligence, four issues remain unresolved: current valuation/unicorn status, a stable current FDA-clearance count, a supportable current headcount, and a sourced 170+ model count. The report therefore treats Aidoc as a well-capitalized, high-scale private company with meaningful regulatory proof points—but not as a fully disclosed business whose valuation and coverage metrics can be accepted without qualification.[CO036, CO039, CO040, CO043, CO044, CO045]

2.1 Market boundary and included spend

Aidoc should be analyzed inside the radiology AI workflow market first, then extended into adjacent care-coordination and cross-specialty orchestration layers. The official radiology page is explicit that the company’s value starts with triage, prioritization, quantified findings, and follow-up activation inside imaging workflows. The care-coordination page then broadens the lens from radiologist productivity to multidisciplinary decision support, especially for urgent cases like pulmonary embolism where timing, communication, and follow-up management matter as much as image interpretation. This boundary excludes generic EHR vendors, imaging hardware OEM revenue, billing software, and most consumer-facing digital-health tools, even though those systems remain integration dependencies. It also means broad “AI in healthcare” TAM claims overstate relevance unless they can be translated into radiology or acute-workflow budgets. MarketsandMarkets and Emergen Research are still useful because they confirm that hospitals are the largest radiology-AI buyers and that CT-centric, workflow-heavy use cases dominate current spending. But the chapter treats those reports as category ceilings, not as direct proxies for Aidoc’s capture. The important analytical move is to define Aidoc’s real market as enterprise workflow software attached to imaging, triage, and downstream care activation—not all diagnostic AI, and definitely not all healthcare IT.[CM001, CM002, CM003, CM010, CM011, CM016]

2.2 Evidence-constrained sizing lenses

Public evidence is good enough to build market bounds, but not good enough to turn Aidoc’s market into a single precise TAM/SAM/SOM figure. The top-down lens starts with the size of U.S. healthcare delivery: CMS reports $5.3 trillion of national health expenditure in 2024, including $1.63 trillion of hospital spending and $1.11 trillion of physician and clinical services. The hospital-count lens adds 6,100 U.S. hospitals and more than 900,000 staffed beds from AHA. The category lens then narrows to radiology AI specifically: MarketsandMarkets puts the global market at $0.76 billion in 2025 rising to $2.27 billion in 2030 at 24.5% CAGR, while Emergen says hospitals represented about 55% of global demand in 2024. Finally, the installed-base lens looks at Aidoc itself: 150+ health systems in the 2025 CARE financing release, nearly 2,000 hospitals in the 2026 Series E materials, and verified enterprise deployments at Hartford and Asklepios. Those figures show that the market is real and that Aidoc already occupies meaningful share within the subset of enterprise-ready buyers. What they do not reveal is price per hospital, module, or study. That missing pricing data is why the report uses low/base/high proxy ranges instead of pretending to know a clean revenue pool for every hospital or service line.[CM006, CM007, CM017, CM018, CM037, CM038]

2.3 Buyer map and adoption path

The strongest public deployment evidence suggests Aidoc sells into large provider organizations rather than through direct payer reimbursement or consumer channels. Hartford HealthCare framed its deal around cross-department collaboration, AI governance, and reduced care delays across radiology, cardiology, vascular, neurology, and the emergency department. Asklepios framed its rollout around radiology shortages, support for smaller sites, and 24/7 emergency coverage. Those examples imply a buyer map in which radiology leaders, CMIOs, CIOs, and health-system operations teams all matter, with budget ownership shifting based on whether the product is purchased as a departmental radiology tool or an enterprise platform. Aidoc’s own marketing sharpens that point: the company emphasizes one integration, multi-algorithm orchestration, and the extension of radiology signals into patient management. That combination makes adoption less about winning a single algorithm bake-off and more about proving that the platform reduces friction across an imaging workflow. The public rollout evidence also suggests the adoption path follows a standard enterprise-software pattern: problem recognition, integration design, pilot validation, then system-wide rollout. Aidoc’s disclosed installed base is already large enough that the commercial question is not whether the product can leave pilot mode, but whether buyer budgets and governance structures will support broader enterprise standardization.[CM004, CM005, CM019, CM032, CM033, CM034]

2.4 Drivers, constraints, and reimbursement reality

Three demand drivers recur across the evidence base. First, workforce pressure is real: AAMC still projects a broad physician shortage into 2036, while Neiman HPI and ACR say the radiologist shortage persists and that attrition accelerated materially after 2020. Second, imaging demand keeps rising: Neiman’s demand study projects another 16.9% to 26.9% increase in imaging utilization by 2055, and the older population is a disproportionate consumer of imaging. Third, category growth remains strong: MarketsandMarkets still projects 24.5% CAGR for the global radiology-AI market through 2030. Against those tailwinds, the clearest constraint is reimbursement. The peer-reviewed reimbursement paper and Radiology Business article both make the same point from different angles: most imaging AI tools do not have separate paid CPT codes, and only a tiny number of newer imaging-AI procedures have reached Category I status. That forces ROI to come from throughput, avoided misses, reduced repeats, or better downstream care management rather than a clean one-to-one payment stream. Regulatory burden is the second constraint. FDA’s action-plan and PCCP materials show that adaptive medical AI still needs disciplined change control, monitoring, and documentation. The third constraint is integration friction: hospitals need systems that fit old PACS/RIS/EHR environments, not just high standalone algorithm accuracy.[CM012, CM013, CM020, CM021, CM022, CM023]

2.5 Adverse evidence and what it means for adoption timing

The most useful adverse evidence in this market is practical rather than sensational. Aidoc’s real-world pulmonary-embolism study summary is a good example: 97.8% agreement sounds strong, but the algorithm still missed 15% of confirmed PE cases and lost most adjudicated disagreements, which means hospitals cannot responsibly treat AI as a radiologist substitute. That is not a failure of the product so much as a reminder that the value proposition is triage and workflow support, not autonomous reading. The reimbursement evidence points in the same direction. If most acute radiology AI use cases will not earn separate paid CPT codes, buyers will continue to justify purchases through operational savings and clinical-risk reduction rather than direct line-item reimbursement. That slows adoption because enterprise software budgets, IT resources, and clinical governance committees all have to believe the efficiency and safety story before the product reaches scale. Aidoc’s current installed base shows this hurdle is surmountable for sophisticated systems, but it also implies that future growth depends on winning central workflow budgets, proving integration ROI, and showing that foundation-model governance can be trusted over time. In other words, the category has crossed the credibility threshold, but not the frictionless budget threshold.[CM025, CM026, CM027, CM028, CM029, CM030]

3.1 Landscape and status-quo alternatives

Aidoc does not compete in a single neatly bounded category. Its product stack starts with regulated radiology triage, extends into aiOS orchestration, and then reaches into care coordination and downstream follow-up. That means the real comparison set includes direct clinical-AI peers like Viz.ai, infrastructure-first imaging vendors like Enlitic and Nanox.AI, pathology-adjacent foundation-model players like PathAI and Paige, and incumbent imaging-IT vendors that already own PACS, archiving, routing, and security review. Independent industry coverage reinforces that hospitals are increasingly buying for workflow integration and enterprise value rather than for isolated point algorithms. The status quo remains powerful. ACR's ARCH-AI program frames imaging-AI deployment as an ongoing governance and quality-assurance discipline, not just a software purchase. Radiology Business likewise argues that many health systems still struggle to justify AI unless it improves enterprise operations beyond radiology. In practice, that means existing PACS, worklists, and in-house governance remain real substitutes for Aidoc, especially when a hospital can add AI incrementally on top of incumbent infrastructure instead of replacing its workflow layer. [CP001, CP002, CP003, CP004, CP031, CP032]

3.2 Direct clinical-AI peers

Viz.ai is the clearest like-for-like competitor because it combines image-triggered detection with downstream care-team mobilization and has already scaled into large health-system deployments. It markets more than 50 FDA-cleared algorithms, positions itself as a care coordination platform rather than a single-use tool, and has public evidence of unicorn-scale fundraising. That makes Viz.ai the most direct benchmark for whether Aidoc can keep acute-care coordination differentiated as more vendors add algorithm breadth. Enlitic and Annalise/Harrison.ai compete differently. Their messaging emphasizes imaging-data normalization, workflow efficiency, and broad finding coverage rather than systemwide care coordination. Nanox.AI/Zebra sits in another adjacent lane again: it combines AI and software with a broader imaging-network and hardware agenda. These peers matter because they can pressure Aidoc on modality breadth, algorithm count, or hospital workflow embedding even if they do not fully replicate Aidoc's cross-specialty operating-layer story. [CP005, CP006, CP008, CP009, CP010, CP011]

3.3 Pathology entrants and workflow incumbents

PathAI and Paige are not radiology-first, but they matter because they compete for the same enterprise AI budget and governance attention. Both pitch pathology foundation models, workflow tools, and large data assets; both also became acquisition targets in 2025-2026, underscoring how strategic the workflow-plus-model layer has become. Their relevance to Aidoc is not that a pathology suite can replace acute radiology triage, but that hospital executives increasingly evaluate AI as an operating-layer decision spanning service lines. The more durable distribution threat comes from incumbent imaging-IT vendors. Nuance/Microsoft, Sectra, Fujifilm, AGFA, GE, and Intelerad all market AI deployment as an extension of existing imaging infrastructure, often emphasizing single-contract onboarding, cloud PACS, or embedded orchestration rather than individual algorithms. Once hospitals believe the workflow layer should host multiple third-party models, incumbents gain leverage from installed base, procurement familiarity, and control over integrations that Aidoc still must win account by account. [CP017, CP018, CP019, CP020, CP021, CP022]

3.4 Commercialization, switching costs, and moat durability

Aidoc still has a real competitive asset bundle. Official materials show breadth across radiology triage, enterprise orchestration, and care coordination, while the 2025 financing materials indicate customers already run third-party models on aiOS. That supports an argument that Aidoc is moving from single-algorithm vendor toward operating system, which is strategically important because hospitals increasingly want fewer vendors and more governance consistency. Still, the moat is not absolute. Multi-vendor orchestration is becoming table stakes, not unique. Fujifilm markets open orchestration with 50-plus validated algorithms, Nuance/Microsoft sells easier deployment through existing workflow contracts, and Sectra plus GE/Intelerad can bundle AI with broader imaging infrastructure. Independent coverage also makes clear that reimbursement and direct ROI remain narrow for much of imaging AI. Aidoc can therefore win when a health system values integrated follow-up, incidentals, and enterprise AI governance, but it is vulnerable wherever procurement is dominated by incumbent workflow control or where hospitals treat AI as a feature inside PACS rather than a new operating layer. [CP006, CP007, CP021, CP023, CP026, CP029]

3.5 Exhibits

4.1 Revenue model and pricing signals point to enterprise software, not transparent transaction pricing

Aidoc's public commercial posture looks enterprise-led from every angle. The product pages describe deep integration into PACS, EHR, scheduling, reporting, and care-coordination workflows, while the AWS Marketplace listing says the offering is available only by private offer. That is the classic signature of negotiated enterprise contracts rather than self-serve software or a standardized per-study checkout flow. Aidoc's monetization surface also appears broader than a single radiology triage module: aiOS governs multiple models, the radiology product sits inside existing imaging workflows, and the care-coordination layer extends value into patient follow-up and cross-specialty activation. What is missing is just as important. Official pages do not publish list prices, per-study rates, or segment revenue mix. The only concrete public pricing signal found in this run is a low-reliability third-party review page that estimates $50,000 to $150,000-plus of first-year cost and says pricing starts around $50,000 per installation. That is directionally useful as a packaging proxy, but not reliable enough for underwriting. The defendable conclusion is that Aidoc behaves like enterprise SaaS / site-license software with implementation work and workflow-stickiness, while actual realized pricing and revenue recognition remain undisclosed. [CI001, CI002, CI003, CI004, CI005, CI006]

4.2 Capital history is visible, but valuation support remains incomplete and internally inconsistent

Aidoc's latest financing record is far more current than its disclosed operating metrics. The company said in 2023 that a $110 million Series D brought total funding to $250 million. In 2025, a PRNewswire release described a $150 million financing plus a $40 million revolving credit facility and said total funding was $370 million. Then, in April 2026, Aidoc and Goldman Sachs said a $150 million Series E brought total funding to over $500 million. Taken individually, each announcement is clear on the new round. Taken together, the cumulative totals do not reconcile cleanly, which means the safest way to model capital raised is round-by-round, not by trusting headline lifetime-funding tallies. Valuation evidence is even thinner. Calcalist reported that the 2025 round was raised at a valuation higher than the prior round, while Globes said the company would not disclose either revenue or the round valuation. That leaves the post-2024 record supportive of momentum but insufficient for a numeric fair-value call. In practical terms, public evidence does not currently confirm a precise Aidoc valuation or current unicorn status, even after the 2026 Series E. [CI010, CI011, CI012, CI013, CI014, CI015]

4.3 Unit economics logic is believable, but public numbers stop at customer-value proof and external benchmarks

Aidoc has a credible public value story. Official materials and investor commentary tie the product to shorter lengths of stay, radiology efficiency, reduced diagnostic delay, and measurable financial returns for health systems. That matters because imaging AI still struggles to justify itself on reimbursement alone. Independent market coverage says only a small number of imaging-AI applications have meaningful reimbursement and that ROI often fails unless the benefit accrues across the broader hospital network. Aidoc's own pitch lines up with that reality: the product is sold as workflow infrastructure and care coordination, not merely as a reimbursable image-analysis widget. Where public evidence runs out is inside the vendor P&L. There is no disclosed gross margin, CAC, payback, NRR, or cash conversion. To anchor the economic logic, the best public benchmark in this run is GE's planned Intelerad acquisition, which implies that mature workflow-layer imaging software can generate roughly 90% recurring revenue and more than 30% adjusted EBITDA margins. Nanox provides the opposite cautionary example: a public imaging-AI company can still produce very small software revenue, losses, and ongoing financing needs. Aidoc likely sits somewhere between those poles, but public evidence does not show where. [CI019, CI020, CI021, CI022, CI023, CI026]

4.4 Verdict: strong demand signal, weaker underwriting file

On current evidence, Aidoc's financial profile is stronger than that of an early-stage algorithm vendor but weaker than that of an underwritable late-stage software company. The 2025 financing, 2026 Series E, and broad deployment claims suggest meaningful market pull and reduce immediate distress risk. Post-Series-E news flow also stayed active, with Aidoc highlighting new customer deployments, European product expansion, and a nationally prominent CMO hire rather than signaling retrenchment. The product footprint, integration depth, and enterprise packaging are all consistent with recurring, high-switching-cost software. Public comp transactions in adjacent AI diagnostics—Viz.ai's unicorn financing, Roche's PathAI deal, and Tempus's acquisition of Paige—also show that strategic buyers and growth investors are willing to pay for workflow plus data plus model assets. But the file is still incomplete on every metric that matters most to a financial buyer. Revenue, ARR, gross margin, burn, cash, debt, CAC, payback, and NRR are all undisclosed. Public pricing is opaque. The company's own total-funding figures conflict across official releases. And the post-2024 record still does not publicly confirm a numeric valuation or current unicorn status. The right financial stance is therefore constructive on commercialization quality and platform relevance, but explicitly cautious on margin path, valuation support, and capital adequacy until management opens the books. [CI014, CI015, CI030, CI031, CI034, CI035]

4.5 Exhibits

5.1 aiOS platform architecture and operating model

Aidoc’s core product is not just a library of point algorithms; it is aiOS, an enterprise operating system for clinical AI. The aiOS page explicitly describes the platform as the layer that runs, orchestrates and governs clinical AI across a health system, with one integration feeding multiple workflows. In practice, that means a health system does not need to deploy each algorithm as a standalone widget. Instead, aiOS structures image and metadata inputs, determines which models should run, delivers output in the patient and study context, and exposes governance, monitoring and outcomes tracking at the system level. The architecture is workflow-first. Aidoc says aiOS uses textual data, scan metadata and pixel analysis to decide which scans should receive which AI processing, and then pushes suspected findings into PACS, EHR, mobile and care-team workflows. Aidoc also positions aiOS as an open ecosystem layer rather than a closed product silo: external-model hosting is a stated design goal, and multiple independent 2025 sources say the platform already hosts non-Aidoc models for a majority of customers. That makes aiOS closer to an AI control plane for hospital IT than a single-purpose radiology application. Public deployment examples support the platform framing. Hartford HealthCare reached initial go-live in three weeks, Mercy standardized aiOS across 50 facilities, and Asklepios completed rollout across 28 hospitals. Those examples suggest the product has moved beyond pilot-stage departmental tooling into enterprise implementation patterns. The remaining caveat is that Aidoc’s official site still markets the “largest portfolio of FDA-cleared algorithms” without publishing a current official numeric portfolio total, so breadth must be described with care. [CE001, CE002, CE003, CE004, CE005, CE006]

5.2 CARE foundation model and solution breadth

CARE (Clinical AI Reasoning Engine) is Aidoc’s bid to move from single-condition algorithms to a reusable clinical foundation model. Aidoc’s foundation-model page says CARE is trained on real-world multimodal data, spanning not just imaging but also text, EHR fields, labs and vitals. The January 2026 body-CT clearance is the strongest public proof point that this model is no longer conceptual. Aidoc’s official release and an independent Diagnostic Imaging summary both state that the clearance combined 11 newly cleared indications with three previously cleared indications into one workflow, creating a 14-indication body-CT triage product. The FDA’s K252970 summary adds the most concrete technical detail. The cleared device is BriefCase-Triage: CARE Multi-triage CT Body, a workflow-triage and notification product for contrast and non-contrast CT of the chest, abdomen and pelvis. The 11 new findings listed in the FDA documentation are diverticulitis, abdominal-pelvic abscess, appendicitis, intestinal ischemia/pneumatosis, obstructive renal stone, small bowel obstruction, large bowel obstruction, spleen injury, liver injury, kidney injury, and pelvic fracture. Aidoc’s release also reported mean sensitivity of 97% and mean specificity of 98% across the 11 new indications in its FDA-reviewed study. Outside the comprehensive body-CT clearance, Aidoc’s product pages show the clinical breadth of the existing stack. The neuro page bundles vessel occlusion, CT perfusion, brain aneurysm, intracranial hemorrhage, C-spine fracture and vertebral compression fracture workflows. The VTE page goes beyond pulmonary embolism into incidental PE, DVT and IVC-filter management. The aortic page combines acute aortic dissection triage with aneurysm follow-up. Together those pages support a real workflow portfolio around PE, ICH, aortic dissection, fractures and cervical-spine findings. What remains less clear is the current official aggregate portfolio count: public sources support breadth, but Aidoc’s own site does not publish a reconciled total. [CE007, CE008, CE009, CE010, CE011, CE012]

5.3 Integration, orchestration and enterprise deployment

Aidoc’s product claims consistently emphasize integration into the systems where clinicians already work. The aiOS, VTE and aortic pages all describe PACS, EHR and mobile delivery, with care coordination layered on top of imaging triage. The VTE page is especially explicit: aiOS ingests multimodal data, routes alerts into PACS, EHR and mobile workflows, and exposes labs, vitals, image review and PERT communication in the same operating layer. The aortic page makes a similar claim for cross-department chat, mobile review and EHR-connected care activation. The FDA summary shows the regulatory boundary around this design. K252970 is a workflow triage and notification product, not diagnostic software; users remain responsible for reading full images in PACS and using professional judgment. That boundary matters because it explains how Aidoc has been able to embed AI deeply into operational workflows while still positioning the software as a prioritization and coordination layer rather than an autonomous diagnostic agent. Deployment evidence indicates Aidoc can support more than one operating model. The FDA summary describes a Linux-based server in a cloud environment for the CARE body-CT clearance. Aidoc’s AWS page markets a HIPAA- compliant enterprise deployment on AWS, while the Asklepios rollout says the implementation followed a secure, cloud-based approach with an on-prem aiOS platform to satisfy GDPR and existing radiology-system constraints. Taken together, the public record supports a configurable cloud-plus-enterprise-integration model rather than a single immutable deployment architecture. [CE004, CE006, CE014, CE015, CE018, CE020]

5.4 Trust, safety, security and compliance

Aidoc’s trust posture looks meaningfully more mature than a typical startup AI vendor’s public surface. The quality page says Aidoc operates under a QMS certified to MDSAP and ISO 13485, and that it is compliant with FDA QSR, EU MDR, ISO 14971 and IEC 62304. Just as important, the page explicitly reiterates that the company’s triage and notification solutions are not diagnostic software and are not intended to replace a clinician’s full image review. That is a useful public acknowledgement of the boundary conditions around the product. On the security side, Aidoc says it aligns with the NIST Cybersecurity Framework and runs on AWS and Azure using controls such as EDR, encryption, SIEM and CSPM. The AWS partnership page separately markets the product as HIPAA-compliant and enterprise-ready. None of those statements substitute for a customer trust-center audit or a reviewed SOC report, but they do show that Aidoc is presenting itself as a hospital-grade, regulated software vendor rather than a lightweight departmental AI app. The BRIDGE framework adds a governance lens beyond controls and certifications. Aidoc positions BRIDGE as a roadmap for resilient, responsible AI adoption built with NVIDIA, health systems and industry partners. That is useful not because it proves technical superiority, but because it suggests Aidoc understands that multi-model deployment success is as much about governance and operating discipline as algorithm accuracy. The public gap is that BRIDGE describes principles and collaboration, not the precise customer-by-customer onboarding criteria for third-party models. [CE021, CE022, CE023, CE024, CE027, CE036]

5.5 Maturity, ecosystem direction and technical risks

Aidoc’s strongest technical signal is not just regulatory activity but platform maturity in live health systems. By late April 2026, the company and independent media were describing 60+ million patient cases analyzed per year and deployment across nearly 2,000 hospitals. At the customer-story level, Mercy, Hartford and Asklepios provide direct evidence that Aidoc can move from initial agreement to operational rollout without being trapped in one-site pilots. Careers postings for AI algorithms, ML platform, backend, DevOps, cloud platforms and infrastructure product roles further suggest active investment in a sustained enterprise product roadmap. The roadmap is ambitious. Aidoc says CARE will expand across CT and X-ray workflows and that automated draft report creation is a near-term target. Those directions fit the company’s repeated “pixel to draft report” language and its claim that foundation models compress years of roadmap into a shorter period. But those roadmap items are still prospective rather than broadly evidenced in public production deployments. The main diligence risks are therefore concentrated in proof depth, not in surface breadth. Public sources do not give a reconciled official portfolio count, independent multi-site benchmarking for CARE remains thin, and the company has not published a detailed third-party model onboarding or governance rubric for the 69%-external-model ecosystem it describes. The product appears enterprise-grade and real, but the market-facing narrative is ahead of the publicly documented reference architecture and benchmark corpus. [CE025, CE026, CE027, CE028, CE029, CE030]

5.6 Exhibits

6.1 Scale, geography and customer footprint

Aidoc’s public customer story has two layers: a broad scale narrative and a narrower set of directly named proof points. At the broadest level, 2025 independent coverage from Healthcare IT Today, Fierce Healthcare, MedCity News and HLTH repeated company-reported figures of more than 45 million patients annually across 150+ health systems. Aidoc’s April 2026 official and PRNewswire releases then described nearly 2,000 hospitals, more than 60 million cases analyzed per year and more than 110 million total cases analyzed. Those are clearly large numbers, but they are not interchangeable denominators: hospitals, health systems, annual patients and annual cases all measure different aspects of adoption. The geography of named proof tilts toward large U.S. integrated delivery networks and one major European private operator. Fetched public sources directly confirm Hartford HealthCare, Mercy, Sutter Health, WellSpan, Yale New Haven, Renown Health / Carson Tahoe, Temple Health, Advocate Health and Asklepios. Independent articles also name Mount Sinai, Northwell and University of Miami as partners or users, but those appear more as scale references than as fully documented deployment case studies. The best reading is that Aidoc has broad platform reach and meaningful reference-customer density, but a much thinner publicly disclosed customer list than the raw 150+ or 2,000+ scale metrics might imply. [CU001, CU002, CU003, CU004, CU018, CU019]

6.2 Named deployments and reference quality

The strongest named enterprise deployment evidence in fetched sources is Hartford HealthCare. Aidoc’s official launch announcement and HIT Consultant both say Hartford reached initial go-live in three weeks, deployed 17 FDA-cleared algorithms across millions of exams and expanded across radiology, cardiology, vascular, neurology and emergency workflows. Mercy is the clearest multi-site implementation reference: Aidoc’s public stories describe aiOS running across all 50 Mercy facilities, with over a dozen use cases launched simultaneously and millions of images already analyzed. Beyond those two, Sutter Health publicly described a multi-year strategic collaboration that embeds aiOS across its care system and designates Sutter as Aidoc’s West Coast hub. Asklepios gives Aidoc an important Europe proof point, with 28 hospitals live and about 35,000 CT and X-ray images analyzed monthly. Yale New Haven is one of the best quantified service-line case studies because the pulmonary embolism response workflow is described with specific advanced-therapy and missed-activation metrics. Renown/Carson Tahoe and Temple Health show that Aidoc’s public proof is not limited to radiology-only messaging: the company is selling hub-and-spoke stroke orchestration and enterprise AI operating-system value to executive buyers. [CU005, CU006, CU007, CU008, CU009, CU011]

6.3 Quantified outcomes and workflow adoption evidence

Aidoc’s best public outcome evidence comes from workflow acceleration and care-escalation metrics rather than from contract or revenue metrics. Yale New Haven’s PE response story is especially useful because it ties Aidoc to a clinically consequential outcome: about 40% more appropriate advanced therapy use and roughly 70% of potential activations surfaced that would otherwise have been missed. Cedars-Sinai appears on Aidoc’s VTE page as another quantified site, with a 7-hour (41%) reduction in time-to-treatment and a 26% reduction in length of stay for PE care. Renown Health / Carson Tahoe add a stroke-network example: a 32-minute reduction in door-in-door-out time and roughly 30% faster large-vessel-occlusion transfer after around six months. Mercy’s public deployment story adds operational scale metrics rather than care-episode percentages, including 2.4 million images analyzed, 249,000 flagged studies and a 90% reduction in outpatient time-to-diagnosis. WellSpan’s CEO also cited more than 200,000 cases analyzed in one year. These are all meaningful workflow proofs, but they are heterogeneous. Some are company-run case studies, some are site-provided internal data, and none substitutes for standardized cross-customer cohorts. [CU010, CU013, CU014, CU016, CU017, CU019]

6.4 Expansion signals, stickiness and durability gaps

Public customer evidence suggests Aidoc often lands as a platform layer and then expands across additional service lines. Sutter’s announcement emphasizes multi-year infrastructure and co-development rather than a one-off algorithm. Hartford’s announcement talks about a 12-month path toward fuller enterprise implementation. Mercy’s own story says platform standardization is what let it scale beyond a few isolated AI tools. Healthcare IT Today and Fierce both quote Aidoc’s claim that 69% of customers already run non-Aidoc models on aiOS, implying that buyers may increasingly be treating Aidoc as the governance and orchestration layer for a multi-vendor AI estate. That platform-stickiness story is directionally positive, but public durability proof is still thin. There is no disclosed net revenue retention, gross retention, logo retention or renewal cohort data in the fetched sources. There is likewise no public concentration disclosure showing whether a small number of enterprise systems represent outsized ARR or case volume. So while the public footprint suggests Aidoc can expand inside complex systems, the classic SaaS durability metrics needed for underwriting customer quality remain absent. [CU003, CU024, CU025, CU029, CU030, CU031]

6.5 Customer adoption caveats and adverse signals

The main caution is that scale claims are partly real and partly narrative-dependent. “150+ health systems,” “45M patients,” “nearly 2,000 hospitals” and “60M cases per year” all sound like customer-count statements, but they are not measuring the same thing. Public sources also differ in how directly they prove customer relationships. Hartford, Mercy, Asklepios, Sutter and Yale are strong fetched references; Cedars-Sinai is a workflow study site; Temple is an executive buyer testimonial; and media lists of Mount Sinai, Northwell or University of Miami provide weaker proof than a detailed case study. The clearest adverse public source in the fetch set is ITQlick, which flags expensive enterprise implementation, limited pricing transparency and integration burden for smaller practices. That source is low-reputation and itself uses AI assistance, so it should be treated as a weak cautionary signal rather than a decisive negative. More important than the directory review is what public sources do not show: no renewal cohorts, no concentration table, and no direct confirmation in fetched sources for some user-requested names such as NYU Langone, Mayo Clinic or the University of Rochester Medical Center. Those are diligence gaps, not disproofs. [CU020, CU021, CU022, CU023, CU024, CU025]

6.6 Exhibits

7.1 Regulatory, legal, and clinical-safety risk

Aidoc’s most serious risk remains the combination of regulated-device burden and clinical-liability exposure. The company has real regulatory assets: its quality page claims MDSAP, ISO 13485, EU MDR, and FDA-quality-system alignment, and Aidoc announced a new FDA clearance in January 2026 for a comprehensive triage product built on its CARE foundation model. Those are meaningful barriers to entry, but they also move Aidoc deeper into a regime where model updates, new indications, and documentation standards matter more every year. FDA’s January 2025 lifecycle guidance and August 2025 PCCP guidance both point toward tighter evidence expectations for AI-enabled device software, especially when models evolve after launch. Aidoc’s own BRIDGE framework acknowledges that trust, compliance, and workflow discipline are prerequisites for adoption, which is directionally reassuring but not a substitute for product-specific change-control evidence. Clinical liability is the corollary risk. Aidoc’s value proposition is faster escalation of critical findings in emergency and acute workflows, so a miss, delay, or biased prioritization failure could become a patient-harm event rather than a simple software defect. Peer-reviewed literature still flags bias and generalizability as unresolved problems in medical imaging AI, particularly when products move across sites and populations. The right underwriting posture is therefore not “Aidoc is unregulated,” but “Aidoc is regulated, scaling, and therefore increasingly exposed to the cost of doing regulation and post-market safety well.”[CR001, CR002, CR003, CR004, CR005, CR006]

7.2 Security, privacy, reimbursement, and procurement risk

Aidoc’s public materials point to a serious but only partially disclosed control environment. The security page references NIST CSF and AWS hosting, and the data-transfer notice says Aidoc is certified to the EU-U.S., UK, and Swiss DPF frameworks. That is useful baseline evidence, but it is not the same as seeing current SOC 2, HITRUST, penetration-test, incident-response, or BAA detail. In healthcare procurement, missing assurance detail can matter almost as much as a known weakness because it elongates security review and makes breach downside harder to price. HHS OCR’s 2026 ransomware settlements underline that HIPAA enforcement is ongoing and operational, not theoretical. Reimbursement risk compounds procurement risk. Independent radiology sources and the ACR both note that AI reimbursement still lags the volume of FDA-cleared tools because most imaging AI products remain tied to Category III tracking codes instead of durable payment-linked Category I economics. Aidoc can still sell on ROI without direct reimbursement—as Advocate Health and Novant show—but that makes each enterprise sale more dependent on local budget ownership, workflow redesign, and proof of measurable throughput or outcome gains. This is survivable, but it makes adoption slower, stickier, and more sensitive to CFO scrutiny than a clean reimbursement-backed standard of care would be.[CR013, CR014, CR015, CR016, CR017, CR026]

7.3 Partner, platform, and competition risk

Aidoc’s operating model is deeply intertwined with outside platforms. That is partly a strength: the company says aiOS is vendor-agnostic, integrated across PACS/VNA/EHR workflows, and available inside Epic App Orchard, which is exactly the kind of workflow credibility hospitals want. Aidoc also has an AWS collaboration backed by multiyear investment to develop its CARE foundation model. These relationships reduce go-to-market friction and make Aidoc look more enterprise-grade than a narrow algorithm vendor. The same relationships create concentration risk. If Epic changes its posture, if OEM or reseller channels underperform, or if AWS economics shift, Aidoc’s deployment motion weakens quickly. Competitive intensity is also rising from inside the incumbent stack: Microsoft markets Dragon and Azure AI for Epic environments, Oracle markets a clinical AI agent for documentation and workflow, and Epic itself positions AI as embedded throughout its software. Aidoc can still win if hospitals value acute-imaging triage and cross-team orchestration enough to buy a dedicated layer, but that is a pricing-power question, not a guaranteed moat. The practical conclusion is that Aidoc’s partner posture is an important mitigation, yet it should be treated as an exposure map first and a moat second.[CR018, CR019, CR020, CR021, CR022, CR023]

7.4 Execution, leadership, and investment kill criteria

Aidoc’s execution risk is rising with its scale. By January 2026, the company said aiOS had surpassed 100 million analyzed patient cases; by February and April it was publicizing multi-site rollouts in Germany, the Netherlands, Southern California, and Advocate Health. That is evidence of product-market fit, but it also means implementation quality, post-market surveillance, and customer-success discipline matter more than in an earlier algorithm phase. On leadership, Aidoc improved its bench materially in 2026 by adding former AMA president Jesse Ehrenfeld as chief medical officer, and the leadership page shows a more developed executive team than a pure founder-centric startup. Even so, Elad Walach remains central to the company’s public narrative and financing posture, so key-person risk is reduced, not eliminated. The most useful investor stance is to define kill criteria in advance. This thesis should be reconsidered if Aidoc encounters a named patient-harm or enforcement event, loses privileged interoperability access, fails to convert pilots because reimbursement and budget ownership stay ambiguous, or proves unable to keep model-change governance aligned with FDA expectations. The company’s large 2026 Series E funding softens runway risk, but it also means future rounds and exit expectations will be judged against a much higher proof bar.[CR028, CR032, CR033, CR034, CR035, CR036]

8.1 Recommendation and valuation stance

Aidoc looks like a real category participant with real traction, but public evidence still stops short of an investable price call. The company can point to a new FDA clearance, named enterprise deployments, over 100 million analyzed patient cases, and strategic backing from AWS, Goldman Sachs Alternatives, General Catalyst, SoftBank Vision Fund 2, and NVentures. That is enough to rule out the dismissive view that Aidoc is merely a concept-stock story. It is not enough to justify paying a premium private valuation blindly. The core problem is valuation opacity. Public post-2024 financing coverage confirms more capital and more investor support, but not a clean current post-money number. That means there is no public way to translate today’s implied private price into a revenue multiple, margin-adjusted multiple, or preference-adjusted outcome. My stance is therefore research-more / track. The company may deserve a premium if internal metrics are strong, but public evidence today supports curiosity, not conviction. Because a post-2024-05-20 current valuation is not disclosed, Aidoc’s unicorn status should be treated as an unconfirmed label rather than a verified fact.[CV003, CV011, CV012, CV038, CV039, CV043]

8.2 Financing history and what public sources actually confirm

The last fully publicized Aidoc round with a disclosed dollar amount is the $110 million Series D announced in June 2022. That announcement clearly disclosed size and cumulative capital, but not post-money valuation. Public later evidence becomes less transparent rather than more transparent. Aidoc’s April 2026 Series E release disclosed a new $150 million financing and total capital above $500 million. Independent 2025-2026 press coverage also described additional or higher financing, but still did not pin down a current post-money value. That asymmetry matters. Investors often assume later rounds automatically clarify price discovery; in Aidoc’s case, the public record does the opposite. It proves that capital kept arriving, that the investor set improved, and that the company is still in motion. It does not prove what new investors actually paid, how the cap table changed, or whether the headline mark would survive a public-market or strategic-acquirer reality check. In other words, Aidoc’s financing history today is strong evidence of sponsorship and weak evidence of price.[CV001, CV002, CV003, CV004, CV009, CV010]

8.3 Comparable valuation anchors

The cleanest public anchor for Aidoc is not another opaque private round but the gap between premium public AI multiples and mainstream healthcare-IT multiples. Tempus is the premium outlier at roughly 9.0x market-cap-to-revenue on the cited 2026 anchors. RadNet and GE HealthCare are closer to 2x-3x, while Phreesia and Health Catalyst sit much lower. The message is straightforward: public markets will pay up for differentiated healthcare AI and data scale, but the premium only becomes durable when revenue scale and quality are visible. Private comparables tell a similar but more nuanced story. Viz.ai and Abridge show that category-leading healthcare AI companies can secure rich funding support. PathAI’s 2026 Roche deal and Paige’s 2025 sale to Tempus show that not every clinically credible AI asset gets priced like a standalone platform. For Aidoc, this means the upside case is real, but the burden of proof is also real. Without disclosed economics, the rational approach is to bracket plausible value against these comparable bands rather than treat private scarcity as self-validating.[CV013, CV014, CV015, CV017, CV020, CV023]

8.4 Bull, base, and bear valuation ranges

Because the current private price is undisclosed, the right scenario frame is exit value rather than current intrinsic value. In a bull case, Aidoc converts its current enterprise traction and foundation-model story into several hundred million dollars of recurring revenue while retaining a premium healthcare-AI multiple. That can support a mid-single-digit-billion exit range. In a base case, Aidoc proves itself as a valuable but not exceptional healthcare workflow platform, which points to a lower single-digit-billion outcome. In a bear case, reimbursement friction, EHR-native competition, or weak monetization from current rollouts push Aidoc down toward the 1x-4x public-healthcare-IT corridor. The scenario logic is not that Aidoc must fail to deserve a lower valuation; it is that good companies can still underperform premium private pricing if capital markets normalize faster than operating proof arrives. That is especially true in a 2025-2026 digital-health market that Rock Health and Cooley both describe as concentrated and selective. The right question is not “Can Aidoc be worth a lot?” It is “What revenue and quality bar must Aidoc clear before a premium mark is defensible?”[CV033, CV034, CV035, CV036, CV040, CV041]

8.5 Diligence asks and thesis-break triggers

The most important diligence item is the simplest one: what is the current price? Without that, even the best comparable analysis remains only a plausibility test. The second blocker is revenue quality—ARR, margin, retention, paid usage conversion, and preference stack. If those metrics are strong, Aidoc could plausibly deserve a premium. If they are mediocre, the public comparator band would compress the valuation case quickly. That leads directly to thesis-break triggers. A flat or down round against the opaque 2025-2026 context would be a red flag that earlier private marks outpaced fundamentals. Failure to convert marquee rollouts into paid expansion would tell the same story more slowly. And if reimbursement or incumbent workflow vendors keep hospitals from assigning durable budget to Aidoc, the company will look more like a useful feature layer than a premium standalone platform. Until those questions are answered, the right closing posture is disciplined curiosity rather than eagerness.[CV037, CV039, CV043, CV045, CV046]