AI for Health — The Market Radar 250 companies across 21 segments, every row carrying its sourced traction and a citation. Each segment is a layer — open one for where the gaps are. Sources and what this is not
August 2026 vintage 250 of 250 sourced
88
gaps identified— show all
The shape of the market Estimated segment size (USD)
Market size Companies Gaps
What this shows: how big each segment is, per the best published estimate we could source. Consumer hardware is the largest at $99B and 193x the smallest sized segment (Dental). Scopes differ, so treat each bar as its own estimate.
Consumer hardware $99B 25
Senior care $45B 26
Occupational health services $43B 23
Pharmacy software $34B 26
Revenue cycle & coding $21B 24
Payer SaaS $17B 26
Owning care delivery $16B 25
Practice management $13B 25
Devices & diagnostics $9.1B 25
Interoperability & records $4.9B 25
Clinical trials $3.3B 26
Longevity & health checks $2.8B 25
Workplace safety software $2.3B 25
Mental health $1.9B 26
Beauty & wellness $1.6B 25
Veterinary AI $1.5B 25
Ambient scribes $1.2B 25
Dental $516M 25 Bar colour = how crowded the segment is saturated contested open
Bar height is compressed (square-root) because the sourced range spans roughly 193× — on a straight scale most segments would be invisible. Read the printed figure, not the height. 3 of 21 segments carry no bar: no credible published figure exists, and none was invented. Segment market sizes are third-party analyst estimates, each measuring a DIFFERENT scope on a different definition — they are point estimates for context, not a comparable series, and they must not be summed. Every figure carries its scope, year, source and caveat; segments with no credible published figure carry none rather than an invented one.
Who is building what Every company in the corpus, stacked by segment
250 of 250 companies in scope · 21 of 21 segments shown · layers ordered and sized by company count — the corpus carries no market caps, so a box earns attention by its evidence: open one for the sourced traction.
S1 Devices & diagnostics saturated 22 4 gaps identified — open for the gaps, incumbents and entry risks. Volpara Health (→ Lunit) NZ · Breast-imaging AI: density, risk scoring, screening quality software Alimetry NZ · Wearable body-surface gastric mapping + AI for gastric disorder diagnosis Toku Eyes NZ · Retinal-image AI for eye disease and cardiovascular/diabetes risk (THEIA) HeartLab NZ · Cloud echo PACS + AI-assisted echocardiography reporting (Pulse) Formus Labs NZ · AI 3D pre-operative planning for hip/joint replacement Aroa Biosurgery NZ · Soft-tissue regeneration devices (non-AI medtech export comp) Harrison.ai AU · Clinical AI holding co: radiology (Annalise) + pathology (Franklin) Annalise.ai AU · Comprehensive CXR (124 findings) and head-CT (130 findings) triage AI See-Mode Technologies AU · AI ultrasound analysis/reporting: vascular, breast, thyroid EMVision AU · Portable electromagnetic brain scanner for bedside/ambulance stroke diagnosis 4DMedical AU · Quantitative lung-function imaging software (CT:VQ) Pro Medicus AU · Enterprise radiology imaging/viewer software (Visage) — the ASX benchmark comp Eko Health Other · AI-enabled digital stethoscopes for heart/lung disease detection Butterfly Network Other · Handheld chip-based point-of-care ultrasound + AI software Qure.ai Asia · Medical-imaging AI across chest X-ray (qXR), head CT (qER) and CT — including TB screening and lung-nodule detection deployed through national health programmes Bunkerhill Health US · Carebricks platform for building and deploying clinical + operational AI agents inside health systems — imaging review (coronary calcium, lung nodules), follow-up identification, prior authorisation, registry management Aidoc US · Enterprise clinical-imaging AI platform (aiOS + CARE foundation model) for hospital triage workflows Viz.ai US · AI care-coordination platform routing CT triage alerts to on-call specialists across 50+ care pathways Cleerly US · AI coronary-CT plaque quantification (AI-QCT) sold to cardiology practices and health systems GE HealthCare US · Imaging, ultrasound, patient monitoring and contrast agents, with AI image reconstruction and reading RapidAI US · Hospital imaging-AI platform: stroke triage (LVO/ICH, CT/MR perfusion), aneurysm, PE, aortic modules HeartFlow US · AI analysis of coronary CT angiograms (FFR-CT, plaque, RoadMap), billed per case to hospitals + Gap Selling the read TO the insurer, rather than getting the insurer to pay for a provider-ordered one The additions sharpen this gap instead of filling it: Cleerly and HeartFlow prove a plan will reimburse an algorithmic read — a Category I CPT code from 1 Jan 2026 and 86M+ Americans covered across Aetna, UnitedHealthcare, Cigna, Humana and EviCore; HeartFlow billed per case to hospitals for $176.0M in FY2025 — but both sell to cardiology practices and health systems, with the plan as payer of record, never as the customer. Overjet, in this corpus's dental segment, remains the only row selling FDA-cleared image reads to providers AND insurers for claims review, and the payer segment's own AI (Cohere Health at 12M+ PA requests a year, Anterior across 50M lives) does clinical-reasoning prior auth, not the read. Viz.ai shows the adjacent motion pays — its life-sciences arm doubled over 18 months to 14 partnerships (Jul 2026), monetising the alert to pharma rather than to the provider. Checked against the dental, payer and interoperability segments before claiming it. Absence from a 250-company snapshot is not absence from the market. + Gap Post-market surveillance of already-cleared models at the deploying site The six additions make this gap bigger, not smaller: Aidoc runs at ~2,000 hospitals on 110M+ cases, Viz.ai at ~2,000 US hospitals, RapidAI at 2,500+ hospitals across 100+ countries, and no row in the 250 sells the layer that revalidates a cleared model against a given hospital's scanner mix and case mix over time. Canon counts 1,524 authorisations across 781 distinct as-filed filers, 528 of which hold exactly one device. The natural owner is now identifiable rather than hypothetical — Aidoc's aiOS and Bunkerhill's Carebricks (15 health systems incl. Cleveland Clinic, Mayo, Intermountain) are deployment substrates that could absorb audit as a feature — which argues the window is closing, not that it is wide open. Honest caveat: this may be empty because neither the buyer nor the budget line exists yet, not because it was missed. + Gap Clinical dermatology / lesion triage, and digital pathology sold as a standalone product Re-tested against all 250 rows, not just this segment's 24: the corpus's skin AI is still cosmetic — Haut.AI and Perfect Corp license skin analysis to beauty brands and clinics, PatientNow is an aesthetic EMR — and pathology appears only as Franklin inside the Harrison.ai holding company, as dental x-ray detection (Overjet, Pearl), or as a coding input (CodaMetrix). None of the six additions touches either: Aidoc, Viz.ai, RapidAI, Cleerly and HeartFlow are CT-centred, and GE HealthCare's row names imaging, ultrasound, patient monitoring and contrast agents. Also still absent across all 250 rows — by full-text search of the merged file, which returns zero rows for either — any paediatric-specific diagnostic AI and any microbiology / antimicrobial-resistance diagnostic. + Gap A reimbursement-access layer sold to the vendors that do not have one The gap survives but its premise has inverted: it is no longer that the pool does not pay. Cleerly obtained a Category I code effective 1 Jan 2026 and payer-by-payer coverage reaching 86M+ Americans; HeartFlow bills per case and booked $176.0M in FY2025. Both did it themselves, vendor-specifically, and HeartFlow needed a Nasdaq listing to fund the road there. Nothing in the 250 rows sells that capability as a product — the revenue-cycle segment (SmarterDx, Fathom, CodaMetrix, Nym Health, Adonis, Waystar) codes the provider's encounter, not a device vendor's own service line — and the 528 single-device filers canon counts are precisely the population that cannot self-fund a coding-and-coverage campaign. The old framing (81 Medicare OPPS claims worth $76,990.50 across the three flagship AI services in CY2024) still describes the Medicare side; it no longer describes commercial cardiac CT. S2 Owning care delivery saturated 19 4 gaps identified — open for the gaps, incumbents and entry risks. Forward Health (dead) US · $150/mo membership AI clinic + CarePod kiosks; shut down Nov 2024. Babylon Health (dead) UK · AI triage chatbot + virtual GP; took capitated medical risk; bankrupt Aug 2023. One Medical (Amazon) US · Membership hybrid primary care; Amazon-owned; expanding via health-system partnerships and pharmacy kiosks. Carbon Health US · Tech-enabled urgent/primary care clinic chain; Chapter 11 Feb 2026, emerged lender-owned Jun 2026. K Health US · AI symptom-intake + virtual primary care copilot; payer/health-system partnerships. Crossover Health Other · Employer/payer advanced primary-care clinics (Apple, Amazon, Microsoft clients). Akido Labs US · AI-native clinic operator: ScopeAI runs the visit, doctor approves; ~100 clinics, 26 specialties. Hims & Hers US · DTC virtual care + pharmacy; the profitability benchmark for virtual-first care. Eucalyptus (Pilot/Juniper) AU · House of DTC telehealth brands; women's/men's health, GLP-1 weight-loss led. Midnight Health AU · DTC telehealth brands (hub.health, Youly, Stagger); insurer-backed. InstantScripts AU · Online scripts + telehealth consults; now inside Wesfarmers Health retail-pharmacy machine. ForHealth AU · PE-owned bulk-billing medical/dental centre network; facilities-services model to independent GPs. Tend Health NZ · Digital-first GP chain (app + clinics); now NZ's biggest general-practice consolidator. Green Cross Health NZ · NZX-listed pharmacy + GP operator (The Doctors); exiting clinic ownership via sale to Tend. Practice Plus NZ · Virtual GP / after-hours telehealth sold as an extension of existing practices (B2B2C). Midi Health Other · Virtual perimenopause/menopause clinic expanding to broader midlife women's health Nourish US · Insurance-covered virtual nutrition care: telehealth visits with registered dietitians for chronic-condition management, on an AI-native scheduling and documentation platform Sword Health US · AI 'care specialist' (Phoenix) delivering virtual MSK, pelvic and mental health care to employers Transcarent US · Gen-AI benefits navigation (WayFinding) plus owned virtual primary, surgery and pharmacy care + Gap Psychiatry and serious mental illness as an OWNED delivery business The 'zero behavioural rows' version of this gap is gone: Sword Health delivers virtual mental health alongside MSK and pelvic care to employers, at a $4B valuation and ~$240M run-rate. What is still unclaimed is narrower — psychiatric care for severe populations under an AI-native operating model, as the product rather than as a third line bundled into an employer MSK benefit. Across the whole 250-company corpus the only other behavioural entries are Slingshot AI's Ash (a consumer therapy app, not a clinic) and SimplePractice (250,000+ practitioners, tooling for behavioural practices, owns no care). Honest caveat: behavioural delivery is a large, well-populated market in reality; absence from a 250-company snapshot is not absence from the market. + Gap AI-native owned delivery to Medicaid and Medicare complex populations Akido Labs is still the only row working public payers at scale — ~500k patients/yr across ~100 clinics and 26 specialties. The old framing, that everything else is cash-pay or an employer contract, does not survive the new roster: Transcarent's 1,700+ clients are employer AND health-plan, and Nourish bills insurance for dietitian visits. Both are commercial-coverage channels, so the public-payer gap narrows in wording but not in substance. Name the reason before calling it opportunity: canon CM-CAP-01 records that the AI productivity-to-revenue channel is worth ~$0 wherever the clinician is not paid fee-for-service, which is exactly the full-risk arrangement these populations sit in. One player is thin cover for the largest patient pool, but the thinness is probably economics, not oversight. + Gap Clinical capacity sold INTO an existing practice's own panel rather than as a competing front door Both additions cut the other way and make this thinner rather than fuller: Sword's Phoenix delivers the care itself to an employer's members, and Transcarent's WayFinding routes members into virtual primary, surgery and pharmacy care it owns. Practice Plus (Hamilton NZ, Pinnacle-operated) is still the single B2B2C row — nationwide virtual consults sold as an extension of the patient's own practice, notes flowing back to that practice, priced per consult. Everyone else takes the patient relationship (Hims, Midi, One Medical, K Health, Sword, Transcarent) or buys the clinic (Tend Health, Carbon Health, ForHealth, Akido). The adjacent practice-management segment (Tebra at 140,000+ providers, SimplePractice at 250,000+ practitioners, Jane App, Assort Health, Hello Patient) arms independent practices with software but not with clinician hours. + Gap Pediatric and family primary care Zero rows in this segment and none in an adjacent one across 250 companies — every delivery model in the list is adult, and mostly adult-with-a-credit-card. Transcarent's owned virtual primary care reaching 20M+ members is the closest thing on the roster, and its row does not say pediatrics is in scope. Read this as the market pricing children's primary care out rather than overlooking it: the demand is not shoppable, the payer mix is public, and nothing in the corpus suggests an unserved buyer with money. Listed as terrain, not as an opening. S3 Practice management saturated 17 4 gaps identified — open for the gaps, incumbents and entry risks. Jane App (Jane Software) Other · Cloud PMS for allied health/wellness clinics: booking, charting, billing, payments SimplePractice US · PMS/EHR for solo-to-small behavioral and allied health practices Tebra (Kareo+PatientPop) Other · PMS, billing and patient-acquisition platform for US independent practices Cliniko AU · Bootstrapped allied-health PMS (physio, chiro, podiatry); ANZ/UK strongholds Halaxy AU · Free-core PMS (EMR, billing, scheduling) monetizing payments/add-ons Semble UK · Cloud clinical system/PMS for UK+France private healthcare IntakeQ / PracticeQ Other · Digital intake forms plus light PMS for small practices Medtech Global NZ · Dominant NZ GP PMS (Medtech32/Evolution/Artia) + ALEX integration ecosystem indici (Valentia Technologies) NZ · Cloud-native NZ GP PMS challenger, born from Pinnacle PHO collaboration Best Practice Software AU · Dominant AU GP PMS (Bp Premier); 51% founder-owned, 49% Sonic MedicalDirector (Telstra Health) AU · Legacy AU GP PMS; cautionary tale of corporate ownership Assort Health US · Specialty-specific agentic voice AI for patient access/call centers Hello Patient US · Conversational AI (voice/text/chat) front door for medical practices Infinitus Systems US · AI voice agents for payer-facing calls (benefit verification, prior auth) Weave Communications Other · SMB healthcare comms platform (dental/optometry/vet/medical) adding AI receptionist athenahealth US · Cloud EHR, practice management and RCM for ambulatory practices; AI voice agents, ambient scribe Hippocratic AI US · Patient-facing non-diagnostic clinical voice AI agents for outreach, sold to health systems and payors + Gap PMS-to-PMS migration sold as the product Every incumbent here is defended by the cost of moving a practice's schedule, clinical ledger, recall lists, payment tokens and claiming setup, and none of the 17 rows sells that move — athenahealth adds a destination record (cloud EHR, practice management and RCM, 160,000+ providers), not a switching service. The nearest adjacent rows remain query and aggregation layers: Metriport, Particle Health and Zus Health are record-query and shared-data APIs for digital-health builders, and Innovaccer unifies hospital EHR with payer claims for health systems (80M patient records) — reading across systems, not relocating a clinic's live book of business. Caveat, sharper than before: migration is a classic services business that may sit below this corpus's inclusion threshold, and athenahealth's Marketplace carries 500+ integrated solutions across 60+ specialties, which is where such a tool would sit without ever becoming a row here. + Gap An ANZ/UK-native agentic front office for independent practices All six agent rows are US-headquartered — Assort, Hello Patient, Infinitus, Weave, Hippocratic AI, plus athenahealth's own voice agents — while the largest single-country installed bases here are ANZ/UK records: Best Practice (~80% of AU GP desktops), Medtech Global (>85% of NZ practices), Cliniko (65,000+ health professionals), Halaxy, Semble (10,000+ professionals). The gap is narrower than it was: Hippocratic runs across 6 countries on 180M+ patient interactions, so 'no agent here reaches outside the US' is no longer true. What stays unclaimed is the combination — an agent built for the non-US independent-practice channel and integrated into those local records, where Hippocratic's stated buyers are health systems, payors and pharma. The only non-US-built AI in these rows remains indici's 2025 scribing add-on. Caveat: the adjacent scribe segment holds AU-native players (Lyrebird Health, Heidi Health, Splose), so the gap is front-office agents, not clinical AI. + Gap Schedule density sold as the outcome — recall reactivation, waitlist backfill, no-show recovery — to the practice that owns the empty slot. This gap narrowed hardest and survives only in narrowed form: the prior reading that every agent row here is inbound-answering or payer-facing outbound is falsified, because Hippocratic AI's product is patient-facing outreach at 180M+ interactions. What no row does is sell the filled slot to the practice — Hippocratic's buyer is the health system, payor or pharma sponsor; Assort covers patient access and call centres and Hello Patient the conversational front door, both inbound; Infinitus calls payers for benefit verification and prior auth; Weave is SMB comms; athenahealth's voice agents ride its own record. Across all 250 corpus rows the nearest things to a retention engine remain RepeatMD (membership, rewards and patient financing layered on aesthetics practices' existing stacks) and Phorest (salon and medspa software with a marketing and retention focus) — both in the beauty rows, neither selling medical or allied recall. + Gap Outcome-linked pricing in medical and allied PMS — now partly an evidentiary gap rather than a clean market one Every model the rows actually state is per-seat subscription or a payments take-rate: IntakeQ/PracticeQ at $49.90–$74.90/mo plus $25/practitioner, Halaxy giving the core away and monetising payments and add-ons, Cliniko holding prices unchanged since 2011. athenahealth changes the standing of the claim without settling it — its row names an RCM line for ambulatory practices and this corpus does not record how that line is priced, so the segment can no longer be asserted free of collections-linked economics from row evidence. The surviving claim is the narrow one: no row here is documented as paid on recovered appointments or collected revenue at the practice level. Revenue-share and services-linked economics do exist elsewhere in the corpus (Moxie's revenue-share medspa model, Infinx's services-heavy RCM), so this was always an absence from a segment, not from healthtech. S4 Revenue cycle & coding saturated 17 5 gaps identified — open for the gaps, incumbents and entry risks. SmarterDx US · Clinical AI for revenue integrity — pre-bill chart audit finds missed diagnoses/charges Smarter Technologies (rollup) Other · SmarterDx + Thoughtful.ai + Access Healthcare combined AI-RCM platform CodaMetrix US · Autonomous multi-specialty coding (radiology, pathology, surgery) spun out of Mass General Brigham Fathom US · Autonomous medical coding, 90%+ automation rates Anomaly (Anomaly Insights) US · AI payer intelligence — predicts denials/underpayments by decoding payer behavior Adonis US · Agentic AI revenue-cycle operations: denials, claim status, underpayment detection Candid Health US · API-first billing/claims engine for digital-health and multi-site provider groups AKASA US · GenAI/LLM platform for hospital revenue cycle: coding, CDI, prior auth Infinx Asia · Tech-enabled RCM services: prior auth, coding, A/R — services-heavy model Cedar US · Patient financial engagement/billing UX for health systems Nym Health US · Fully autonomous coding engine (no coder in loop), radiology/ED focus RapidClaims US · Autonomous coding + claim scrubbing agents; newest funded entrant Waystar (incumbent, public) Other · Public healthcare-payments/clearinghouse platform buying AI capability Tiplu (EU analog) Other · MOMO AI DRG-coding software for German hospitals Tyro Health (AU analog) AU · AU health claiming/payments rails (Medicare, DVA, health funds); ex-Medipass Luminai US · AI workflow-automation platform purpose-built for health-system operations — moving data between disconnected systems, revenue-cycle and billing workflows, prior authorisation, claims, forms and eligibility checks Commure US · AI agents for health-system back office: intake, coding, notes, claims, denials, appeals, collections + Gap Defending a code after payment, not contesting a denial before it This gap narrowed with Commure, whose row runs intake → coding → claims → denials → appeals → collections, so the post-denial contest is now sold inside one stack rather than left to point tools. What no row describes is retrospective defence — standing behind a code once it has been paid and is then audited. Canon CM-PAY-01 records Cigna's R49 auto-downcoding policy paused after AMA advocacy and the Maryland Insurance Administration fining Cigna $80,000 and extending the prohibition statewide via Bulletin 26-9 (7 Apr 2026), while stating that 'the underlying audit right is untouched, and CMS's statutory MA coding-intensity adjustment is unaffected'. Honest limit: RADV and audit-defence vendors exist outside this corpus, so absence from a 250-company snapshot is not absence from the market. + Gap The provider who is not paid fee-for-service Every proposition here monetises a claim, and the new row does not change that: Commure describes health-system back office — claims, denials, appeals, collections — with named references (HCA, Tenet) that are large hospital operators, and nothing in it describes a delegated or capitated deployment. Canon CM-CAP-01 states the coding-revenue channel is worth ~$0 wherever the clinician is not paid FFS — Kaiser Permanente, the VA, staff-model plans, full-risk MA/ACO arrangements. Checked against the adjacent corpus again: risk adjustment is covered on the plan side (Anterior, in the payer segment), not the provider's, so the provider-under-risk seat is still empty at 250 rows. The one qualification Commure forces: intake and notes retain value where there is no claim to code, so a full-stack vendor could reach this buyer without repricing — none of the seventeen says it does. + Gap A bilateral layer instead of two armouries This list arms providers, and Commure arms them harder — one contract now spanning intake through collections at 130+ large health systems — while the corpus's payer segment arms the other side of the same dispute (Cohere Health, Anterior, Lyric, Rialtic, Machinify), and CMS now pays reviewers on a savings share under WISeR: six performance years from 1 Jan 2026 across six states, participants receiving 'a percentage of the expenditures associated with averted wasteful, inappropriate care' (canon REG-WISR-001, REG-WISR-004). Nobody in either segment sells a neutral pre-adjudication substrate both parties accept; the addition widened the asymmetry rather than closing it. Honest limit: payer-owned clearinghouse utilities of roughly this shape exist in the US and simply are not corpus rows. + Gap Coding markets outside the US claim rail Commure makes fourteen US-only rows; Nym is US/Israel, and the only two non-US entries remain already-consolidated incumbents rather than AI-native challengers — Tiplu's MOMO DRG coder at ~400 German hospitals with LOGEX as sole shareholder, and Tyro Health owning AU claiming distribution across 80+ practice-management integrations. Offshore appears only as delivery capacity behind US companies (Infinx's India ops, RapidClaims' India engineering), never as a local coding product. No UK, Canadian, Japanese or LatAm entrant appears. This stays the weakest of the four claims: the corpus is US-first by construction and coding rails are country-specific, so absence here reports the sampling frame at least as much as the market. + Gap Nothing here is priced for the small hospital The scale references in this list are all large — 650+ hospitals (AKASA), 60+ health systems (SmarterDx), 20+ health systems (Anomaly), 130+ large health systems and 500+ orgs across 3,000+ sites (Commure) — and the newest capital went to the widest stack rather than the smallest buyer. Set against canon CM-KH-05, which puts the 26–99-bed cohort at −11.5% YTD 2026, the steepest of six, the segment's supply and the segment's most distressed demand are pointed in opposite directions. Honest limit: a small-hospital RCM outsourcer would plausibly not clear this corpus's inclusion bar in the first place, so treat this as a gap in the funded AI-native cohort, not proof that nobody serves those beds. S5 Consumer hardware saturated 16 4 gaps identified — open for the gaps, incumbents and entry risks. Oura US · Smart ring + $6/mo-class subscription; sleep/readiness; category leader with >75% smart-ring share Whoop US · Screenless strap, subscription-only membership (~85% of revenue recurring); recovery/strain analytics Ultrahuman Asia · Smart ring (Ring Pro) + CGM app; capital-efficient Oura challenger Eight Sleep US · $3,000 AI temperature-regulating mattress cover (Pod) + Autopilot subscription; expanding into medical Withings EU · Connected scales, BP monitors, watches, BeamO multiscope; consumer + RPM Garmin US · Public wearables giant; launched Connect+ AI subscription 2025 Dexcom (Stelo) US · First OTC wellness CGM (Aug 2024); subscription model; opens glucose data to non-diabetics Abbott (Lingo) US · OTC wellness CGM biowearable; consumer metabolic coaching app Levels US · CGM-based metabolic health app on top of third-party sensors Signos US · AI weight-management platform pairing Dexcom OTC CGM with coaching Bee (Amazon) US · $50-class AI memory wristband; acquired by Amazon Limitless (Meta) US · AI recorder pendant; acquired by Meta, hardware discontinued Friend US · $129 AI companion necklace; emotional-support positioning Movano Health (Evie) US · Women's-health smart ring; FDA 510(k)-cleared EvieMED variant Samsung (Galaxy Ring) Asia · $399 no-subscription smart ring; big-tech challenger to Oura August AI Asia · Consumer AI health companion delivered over WhatsApp and mobile apps, answering health questions against a user's own health profile and history; also sold B2B to hospital chains and telehealth providers + Gap Nobody but the consumer pays Every monetised row here is cash-pay — Oura's ~$6/mo-class subscription, Whoop's membership, Eight Sleep's Autopilot, Abbott's ~€120/month UK Lingo. Withings is the only row that touches RPM and that is device sales into providers, not a payer-funded AI subscription; Signos has Blue Cross Blue Shield of Alabama as an investor, not as a payer of record. Checked against the adjacent segment: Sprinter Health does sell B2B2C to health plans, but sells visits, not a device subscription. + Gap The continuous stream never reaches a chart No row in this list sells wearable or CGM telemetry into an EHR, or into a clinician-readable summary a doctor would act on. The adjacent interoperability rows (Metriport, Particle Health, Zus Health, Fasten Health, Epic Systems, Innovaccer) move claims and document records, not device streams: Epic Systems is now a corpus row and ships the patient surface (MyChart), but its row describes no device-telemetry ingestion — so this is not covered next door either. Absence from a 250-company snapshot is not absence from the market. + Gap Sensing stops at the wrist, the finger, the bed and glucose There is no consumer hearing, respiratory/spirometry, oral-health or cuffless blood-pressure product in this list; Withings sells a BP cuff, but as hardware rather than an AI subscription. Consumer skin is explicitly NOT a gap — Perfect Corp and Haut.AI cover AI skin analysis from the adjacent beauty/wellness segment. + Gap Every row is an adult self-tracker Movano (Evie, women's health) is the only demographic specialisation and its own row describes a rocky launch and a $4.1M working-capital raise; nothing here addresses children, infants, or the family as the subscribing unit, while the adjacent senior segment is served (Sensi.AI, Honor). This is the weakest of the four claims: the corpus samples AI-first venture-backed companies, not the retail monitor shelf, so its silence here is thin evidence. S6 Investors & capital saturated 15 4 gaps identified — open for the gaps, incumbents and entry risks. General Catalyst UK · Largest health-AI investor; incubates companies (Hippocratic AI, Commure), owns provider assets, and runs a capital pool for AI-enabled services rollups. a16z Bio+Health Other · Dedicated bio+health fund; leads clinical- and ambient-AI mega-rounds. Thrive Capital US · Growth fund leading the biggest AI rounds, including health. Oak HC/FT Other · Healthcare + fintech specialist. Khosla Ventures Other · Deep-tech VC; long-time health-AI backer. Bessemer Venture Partners US · Multi-stage VC; publishes the State of Health AI benchmark. Flare Capital Partners US · Healthcare-tech specialist, seed/Series A. Transformation Capital US · Growth-stage digital-health specialist. Y Combinator US · Accelerator; largest early-stage health-AI dealflow; standard NZ→US bridge. Blackbird Ventures NZ · ANZ's largest VC; anchor ANZ health-AI investor; runs an NZGCP-backed NZ fund. Heal Capital EU · Healthtech fund backed by 25+ German private health insurers; €1–5M tickets, seed–Series A. Sofinnova Partners EU · European life-sciences VC with a dedicated Digital Medicine strategy. Calm/Storm Ventures EU · Micro-VC for pre-seed/seed digital health and wellbeing. Icehouse Ventures NZ · NZ's most active seed investor; first-cheque specialist. Pacific Channel NZ · NZ deep-tech VC; science-led health and wellbeing startups. + Gap Credit. All 16 rows are equity or accelerator equity; no lender, private-credit fund or specialty-finance provider appears as a row anywhere in the 250-company corpus — yet debt decides outcomes inside it: Dwelly's £69M was £32M GC-led equity plus £37M Trinity debt, Radiology Partners restructured to ~$2.4B at ~8x leverage, and Carbon Health emerged from Chapter 11 lender-owned in Jun 2026. The nearest thing inside the list is General Catalyst's non-dilutive $200M Customer Value Fund (canon CM-PRIV-CM-02, repaid from a capped share of the acquired customer cohort) — which finances customer acquisition, not acquisition or receivables lending. Absent from this snapshot; specialty health lenders plainly exist in the market. + Gap Liquidity vehicles No secondaries fund, continuation vehicle or crossover public-private vehicle is in the list, and none of the 16 is described as buying existing shares. Demand for one is evidenced: canon PB-024 names eleven 2026-27 digital-health IPO candidates with none priced at this vintage, and PB-065 has the listed cohort shrinking by acquisition faster than the IPO class replaces it. The one secondary visible in the corpus — Jane App's TCV-led C$500M+ secondary at a C$1.8B valuation, May 2025 — was led by a firm that is not a row in this segment. + Gap US payer- and provider-side strategic capital Heal Capital (Berlin, backed by 25+ German private health insurers) is the only insurer-funded vehicle in the list; no US payer or health-system corporate venture arm appears. It matters because the most-repeated winning structure in this corpus is capital plus a captive customer — General Catalyst via HATCo's $515M Summa Health acquisition. US health-system and payer CVCs exist in the market; they are simply absent here. + Gap Geography outside the US / Western Europe / ANZ All 16 are US (10), Western Europe (3 — Berlin, Paris, Vienna) or ANZ-NZ (3). No fund in the list is domiciled in the UK, Canada, Japan, Korea, Singapore, India, LatAm, the Middle East or Africa. This is the item most likely to be a snapshot artifact rather than a market fact — the corpus source note is explicit that absence is not evidence a competitor does not exist — but within this list those geographies have no representation at all. S7 Pharmacy software contested 15 4 gaps identified — open for the gaps, incumbents and entry risks. Toniq (Clanwilliam Group) NZ · Dominant NZ dispensary + retail POS software; acquired by Clanwilliam 2017. RxOne NZ · Independent integrated NZ pharmacy platform; minority challenger to Toniq. Fred IT Group / eRx Script Exchange AU · Fred dispensing software + Australia's sole national e-prescription exchange. Minfos (EBOS Group) AU · POS/dispense platform for AU community pharmacy incl. TerryWhite Chemmart; EBOS division. Corum Health AU · LOTS Dispense + POS for AU pharmacies. MedAdvisor (renamed Adheris Health, ASX:AHE) AU · Medication adherence + patient-engagement software for pharmacies; now US-focused. StrongRoom AI (in administration) AU · Medication-management AI for pharmacies/aged care — collapsed 2025. Outcomes (Transaction Data Systems) Other · US pharmacy management roll-up: Rx30, Computer-Rx, PrescribeWellness + clinical-services network. DrFirst Other · E-prescribing + medication-management network across prescribers, hospitals, pharmacies. Prescryptive Health Other · Prescription-intelligence/pharmacy-benefits platform; verified active (not shut down). Aspen RxHealth (MedImpact) Other · Gig-economy pharmacist network app for clinical services (MTM) at scale. TJM Labs Other · AI agents automating intake, data entry, refills, prior-auth, patient comms inside existing pharmacy systems. Plenful Other · AI workflow automation for pharmacy/healthcare ops (340B, inventory, admin). Foundation Health Other · Developer-first pharmacy API + PAIGE AI pharmacist assistant (Truepill co-founder's second act). Asepha Other · AI agents for pharmacy automation (handwritten-script processing, admin). + Gap No AI-native dispensing system of record All four agent-native companies here are explicitly layers on somebody else's platform — TJM Labs runs 'inside existing pharmacy systems', Foundation Health sells a pharmacy API plus the PAIGE assistant, Plenful and Asepha automate workflow around the dispense. Nobody in this list is rebuilding the record itself. Read this as a warning at least as much as an opening: four separately funded teams all chose the layer, which is evidence the record is a regulated, switching-cost-heavy, low-margin business rather than an oversight. + Gap Nothing here sells the pharmacy a tool for its reimbursement problem Every one of the fifteen attacks labour cost — intake, data entry, refills, prior auth, MTM, adherence, inventory. None attacks what the pharmacy is paid per script: PBM contract analytics, spread and clawback recovery, margin-per-fill decision support. Prescryptive Health is the near-miss and sits on the benefits and manufacturer side (it powers the LillyDirect digital experience), not the dispensing pharmacy's P&L; Mark Cuban Cost Plus Drugs attacks the PBM spread in the adjacent ownership segment but as an operator, not as software sold to pharmacies. + Gap No AI-native automation layer anywhere in ANZ All four agent-native companies are North American (TJM Labs, Plenful and Foundation Health in the US, Asepha in Toronto), while the entire ANZ half of the segment is legacy record-keepers — Toniq, RxOne, Fred IT, Minfos, Corum. The space is empty but not untried: StrongRoom AI was the ANZ attempt and is in administration, and MedAdvisor sold its ANZ business to Jonas Software for A$35M in July 2025. An entrant here inherits a market that has already rejected one seller. + Gap Long-term-care and aged-care medication management is uncovered in this corpus StrongRoom AI was the only row addressing pharmacies and aged care together, and it is in administration; the adjacent Senior care segment holds only Sensi.AI (audio in-home monitoring) and Honor (home-care platform), neither of which touches medication workflow. Stated with the honest caveat: LTC pharmacy is a real vertical with real established vendors, so this is absence from a 250-company snapshot, not absence from the market. S8 Dental contested 14 4 gaps identified — open for the gaps, incumbents and entry risks. Overjet US · FDA-cleared AI x-ray pathology detection sold to providers AND insurers for claims review Pearl US · FDA-cleared AI pathology detection in x-rays; PMS integrations incl. EXACT for UK/APAC VideaHealth US · Dental AI diagnostics + workflow for DSOs DentalMonitoring EU · AI remote monitoring for orthodontics/dentistry via smartphone intraoral scans Tend US · VC-backed 'dental studio' clinic chain — the AI-native-clinic cautionary tale Bola AI US · Voice AI for perio/restorative charting and clinical notes inside dental PMS Dandy US · Digital dental lab (AI design/manufacturing); expanding into PMS via Neem acquisition HeyDonto US · AI dental-medical data exchange + revenue cycle management (Conduit platform) InsideDesk Other · AI revenue-cycle-management software for DSOs Abano Healthcare (Lumino NZ / Maven AU) NZ · Australasia's largest DSO — proof non-dentist corporate ownership works in NZ/AU Pacific Smiles Other · Australian dental-centre chain; contested PE takeover benchmark for ANZ clinic values Heartland Dental Other · Largest US DSO; primary enterprise AI adoption channel Software of Excellence / EXACT (Henry Schein One) NZ · NZ-built dominant PMS in AU/NZ/UK/IE — the ANZ distribution chokepoint Praktika AU · First Australian cloud dental PMS (2007); AU-onshore data residency + Gap Mouth-to-systemic screening No company in this list reads a dental image or exam for non-dental disease and routes the finding into medical care. The corpus contains the eye-shaped analog — Toku Eyes assesses cardiovascular and kidney risk from retinal images (Clinical AI devices & diagnostics) — and an entire preventive/longevity tier (Neko Health, Function Health, Prenuvo, Superpower, Everlab) whose 100+ biomarker and whole-body-imaging programs stop at the mouth. Every dental AI row here points inward at caries, bone loss and tooth movement. + Gap Longitudinal outcome measurement on restorative work Overjet, Pearl and VideaHealth all detect pathology at a point in time; DentalMonitoring tracks tooth movement during an orthodontic course. Nothing in this list measures whether a restoration or endodontic treatment held up over years, which is the evidence a payer or a DSO would need to pay for quality rather than volume. + Gap An audit that sits on the patient's side Overjet is described in the corpus as selling the same FDA-cleared detection to providers AND to insurers for claims review; HeyDonto and InsideDesk build the provider's claims case. No company here is positioned as an independent second read for the patient on whether proposed treatment is warranted — the one seat at this table nobody in the list occupies. + Gap The dental benefit itself Across all 250 corpus rows there is no dental carrier or dental-benefits administrator; the insurer segment is medical (Oscar, Clover, Sidecar, Alan, ottonova, Southern Cross, nib NZ). Dental AI here sells INTO payers (Overjet) but nobody is building or rewiring the plan. Honest caveat: large dental carriers plainly exist outside this snapshot, so what is absent is any AI-native attempt on them in this dataset, not the category. S9 Payer SaaS saturated 14 4 gaps identified — open for the gaps, incumbents and entry risks. Cohere Health US · AI prior-authorization / clinical decisioning platform for health plans Anterior US · LLM clinical-reasoning copilot for health-plan prior auth, payment integrity, risk adjustment Machinify (New Mountain combo) US · AI payment-integrity/healthcare-payments platform merged with Rawlings, Apixio PI, VARIS Lyric (ex-ClaimsXten) US · Pre-pay claims editing / payment accuracy platform for payers Rialtic US · API-first claims-editing and payment-accuracy platform EvolutionIQ US · AI claims guidance for disability, workers-comp and injury insurers Develop Health US · AI benefit-verification and drug prior-authorization automation SuperDial US · AI voice agents automating payer phone calls (claims, eligibility, PA status) HealthEdge US · Payer core-admin SaaS (HealthRules Payer), payment accuracy, care management EIS US · Core insurance platform (policy, billing, claims) incl. health/benefits lines Sapiens International Asia · Insurance core software incl. workers-comp; AI/automation suite Alan EU · Full-stack digital health insurer with AI assistant (Mo); FR/BE/ES ottonova EU · Germany's first fully digital private health insurer (PKV) Honeysuckle Health AU · Health data-science, analytics and care-program arm of nib (ex Cigna JV) + Gap Payer-to-payer exchange — the rail CMS mandates and nobody in this list builds CMS-0057-F (89 FR 8758, published 8 Feb 2024) sets a Payer-to-Payer API compliance date of 1 Jan 2027 alongside Patient Access, Provider Access and Prior Authorization APIs, binding MA organisations, state Medicaid and CHIP FFS, Medicaid and CHIP managed care, and FFE QHP issuers (canon REG-PA-001); CMS-0062-P (91 FR 19890, published 14 Apr 2026, comments closed 15 Jun 2026) would extend electronic prior auth to drugs across the same set (REG-PA-003). Every one of the fourteen rows operates inside a single plan's four walls — decisioning, editing, core admin — and none describes moving prior-auth history or claims between two competing plans when a member switches. Checked against the adjacent Health-data interoperability segment: Metriport, Particle Health and Zus Health all query PROVIDER records via Carequality/CommonWell/FHIR for digital-health builders, and Innovaccer joins hospital EHR to payer claims inside a single plan's book — every one a different rail, with a different counterparty, from plan-to-plan portability. Honest limit: the EDI clearinghouses and payer-IT integrators that would obviously build this are absent from the entire 250-company corpus, so this is a gap in the AI-native cohort, not evidence the rail is unbuilt. + Gap Nobody here sells to the person who was denied The fourteen rows have exactly two buyers — the plan (Cohere Health, Anterior, Lyric, Rialtic, Machinify, HealthEdge, EIS, Sapiens) or the provider chasing the plan (SuperDial's voice agents working payer phone lines, Develop Health's benefit verification and drug PA). The member holding an adverse determination is a customer of no one. The right exists in federal law: 42 CFR 422.566(d) requires that an adverse organisation determination '...must be reviewed by a physician or other appropriate health care professional with expertise in the field of medicine or health care that is appropriate for the services at issue, including knowledge of Medicare coverage criteria, before the MA organization issues the organization determination decision' (canon REG-PA-004, eCFR title 42 issue date 2026-07-30). Checked across all 250 rows: every denial-facing product is provider-side revenue recovery — Anomaly decoding payer behaviour, Adonis on denials and underpayments, Waystar's 'silent denials' — contesting on the provider's behalf for the provider's money, not on the member's behalf for the member's care. Honest limit, and the likely reason it is empty: the member has no budget and a successful appeal saves the plan nothing, so this may be an absent business model rather than an unoccupied one. + Gap The evidence trail behind an AI-issued determination Every row sells the decision or the edit; none sells the auditable record that the decision met the standard — the artefact a state insurance regulator, a CMS auditor or plaintiff's counsel would actually test. The demand drivers are on the record: the clinician-review floor at 42 CFR 422.566(d) (REG-PA-004), prior-auth decision clocks of 7 calendar days standard and 72 hours expedited running since 1 Jan 2026 (REG-PA-001), and live enforcement — canon CM-PAY-01 records that Cigna's R49 auto-downcoding policy (one-level reduction of 99204-05 / 99214-15 / 99244-45) was paused after AMA and specialty-society advocacy, that the Maryland Insurance Administration fined Cigna $80,000 on 13 Mar 2026 and ordered it stopped, and that it then extended the prohibition to every insurer in the state via Bulletin 26-9 (7 Apr 2026), while the underlying audit right stayed untouched. Checked across all 250 rows: the only 'audit' products are chart audit for provider revenue (SmarterDx) and workplace/medspa compliance — nothing audits a coverage determination. Honest limit: this is the gap most likely to be a feature rather than a company. Anterior already markets a KLAS-verified 99.24% clinical-accuracy figure, which is the same instinct sold as a spec sheet, and horizontal AI-governance vendors would never appear in a health-specific corpus. + Gap The buyer below the top-10 plan — weakest of the four, name it as such Every distribution claim in this list is large-plan: Lyric at 9 of the top-10 US payers, HealthEdge at 115+ plans and 110M lives, Anterior at 50M lives, Cohere Health at 12M+ PA requests a year. Yet CMS-0057-F's 1 Jan 2027 API obligations bind state Medicaid and CHIP FFS agencies and Medicaid/CHIP managed care as well as MA organisations and FFE QHP issuers (REG-PA-001), and no row here — or anywhere in the 250 — describes a product for a state Medicaid agency, a small regional MCO or a third-party administrator. Honest limit that nearly eats the claim: Medicaid fiscal-agent and MMIS work is held by large systems integrators (Gainwell, Deloitte, Accenture, Cognizant — none of them corpus rows), so absence here reports the sampling frame at least as much as it reports the market. S10 Ambient scribes saturated 13 5 gaps identified — open for the gaps, incumbents and entry risks. Abridge US · Enterprise ambient scribe + 'care intelligence' (coding, CDI), deep Epic integration, health-system sales Microsoft Dragon Copilot (Nuance DAX) US · Unified voice dictation + ambient AI assistant inside EHR workflows; the enterprise incumbent Ambience Healthcare US · Ambient documentation + point-of-care coding/CDI platform for health systems OpenEvidence US · AI clinical search/evidence copilot ('ChatGPT for doctors'), free to verified clinicians, ad-monetized Nabla US · Ambient scribe expanding to agentic clinical workflows (EHR actions, coding, dictation) Suki US · AI voice assistant/scribe for clinicians; platform API licensed into EHRs and partners Heidi Health AU · Freemium AI scribe → 'AI care partner'; clinician-led bottom-up adoption, 110 languages Lyrebird Health AU · AI scribe for GPs/specialists; AU data sovereignty focus, expanding UK + Middle East Corti EU · Healthcare-specialized foundation models + API infrastructure; EU sovereign cloud; powers other vendors Tortus UK · Regulated clinical AI assistant built with/for the NHS; ambient notes, letters, codes Freed US · Self-serve AI scribe for independent clinicians; $39–119/month, consumer-SaaS growth motion Splose AU · AI-powered practice management + scribe for allied health (physio, OT, speech, NDIS) Doximity US · US physician network + clinical AI, documentation and telehealth tools; monetized by pharma/health-system ads + Gap Vendor-neutral measurement of an ambient deployment All thirteen rows sell the tool; none is described as instrumenting whether it works, and the new row does not change that — Doximity sells documentation and telehealth tools to the same clinicians, funded by pharma and health-system ads. The buyer-side need is documented — PHTI (25 Mar 2025) recorded that 'health system leaders indicate that there are gaps in evidence regarding the impact of ambient scribes on productivity and financial performance', vendor marketing runs ~6x ahead of the controlled measurement (Dragon Copilot's '13 additional appointment slots per provider per month' against a measured 2.1), and only about 32% of adopters in the JAMA multisite study used their scribe on >=50% of visits. Honest caveat: this may be consulting work rather than a venture-scale product, and absence from a 250-company snapshot is not absence from the market. + Gap Provider-side documentation-integrity and audit defence for AI-generated notes — narrowed this cycle Commure is now a corpus row (revenue-cycle segment) running health-system back-office agents across 'intake, coding, notes, claims, denials, appeals, collections' at 130+ large health systems and 500+ orgs, 85%+ of the work autonomous, so the claims-level dispute with a payer is no longer unowned. What is still unclaimed is defence of the note itself: Abridge and Ambience both sell coding/CDI in the revenue-capture direction, SmarterDx audits the chart pre-bill to find missed diagnoses and charges, and Anterior and Machinify sit on the payer's side of the table. No row in this corpus is described as defending a provider's AI-written note against a retrospective auditor, and the exposure is real — PHTI's April 2026 assessment reports a health system seeing a 5% increase in Level 5 encounters after deploying AI scribes, and while Cigna's R49 auto-downcoding policy was paused (Maryland Insurance Administration fined Cigna $80,000 on 13 Mar 2026, extended to all state insurers by Bulletin 26-9, 7 Apr 2026), the underlying audit right is untouched. + Gap A value case that survives capitation — narrowed because one of the two exits from it is now occupied Doximity removes the clean version of this claim: its documentation and clinical AI tools are funded by pharma and health-system advertising, so their economics are indifferent to whether the clinician is paid fee-for-service, and it is the second ad-funded row here after OpenEvidence — the first with public accounts behind it. But that route answers the question by making the clinician not the payer, not by proving value to a capitated provider organisation. Still unclaimed: every row that does charge the provider rests on FFS coding uplift — 'care intelligence (coding, CDI)' at Abridge, 'point-of-care coding/CDI' at Ambience — and the E/M channel is worth ~$0 wherever the clinician is not paid FFS: Kaiser Permanente (24,600 physicians), the VA, staff-model plans, full-risk MA/ACO. No company here is built around time, quality-measure closure or handoff quality as the primary paid sale. Caveat: these vendors do sell into those settings (Kaiser's own published ambient return is 15,791 documentation hours saved across 2,576,627 encounters, expressed in hours and satisfaction, never in dollars) — the gap is the value story, not the logo. + Gap Inpatient bedside and perioperative capture The thirteen rows are ambulatory- and consult-shaped: GP and specialist consults (Heidi, Lyrebird), health-system ambulatory (Abridge, Ambience), allied health (Splose), NHS consultations and A&E (Tortus), and now an outpatient physician network monetised by ads (Doximity). Nursing flowsheet, shift-handover and OR/anaesthesia documentation — the highest-volume documentation burden in a hospital — is described by no row in this segment. Weakest of the four claims and no stronger than last cycle: Microsoft's own row states it is 'expanding to nurses and EU markets 2025-26', so the incumbent has announced the move even if the corpus does not yet show it shipped. Absence from a 250-company snapshot is not absence from the market. + Gap Documentation for the capitated and public payers as a first-class product rather than a downmarket tier Tortus is the only row whose position is built on a public system's own evidence (17,000 patients, 9 sites, £500k NHS England backing), and it is UK-only; Corti supplies EU-sovereign infrastructure to other vendors rather than selling the workflow. No row is described as purpose-built for a US public or staff-model buyer — the VA, Medicaid managed care, FQHCs — where procurement, data residency and the absence of an E/M uplift all cut against the enterprise pitch at once. Honest caveat: this is the thinnest of the five claims and the likeliest to be an artefact of what a 250-company snapshot indexes, since public-sector contracts generate less funding press than a Series C, and the enterprise vendors here plainly do sell into these settings without describing them as the product. S11 Owning insurer or pharmacy saturated 13 5 gaps identified — open for the gaps, incumbents and entry risks. Oscar Health US · ACA-marketplace tech-native insurer; AI-heavy admin and member experience Clover Health US · Medicare Advantage insurer; Clover/Counterpart Assistant AI clinical platform, also sold as SaaS Wefox EU · Insurtech that owned a carrier; restructured to asset-light MGA/distribution Sidecar Health US · Employer major-medical insurance with upfront transparent pricing and member payment card Southern Cross Health Society NZ · Not-for-profit friendly society; NZ's dominant private health insurer nib NZ NZ · NZ #2 private health insurer, Australian-owned Amazon Pharmacy US · Amazon's US online pharmacy; price transparency, Prime delivery, RxPass Mark Cuban Cost Plus Drugs US · Cash-pay transparent pharmacy: manufacturer cost +15% margin +$5 fill +$5 shipping; PBM bypass Fuze Health (Alto + Truepill + LetsGetChecked) US · Rolled-up digital pharmacy + diagnostics + virtual care platform Medly Pharmacy US · Digital pharmacy; cautionary full-stack failure — overgrowth before unit economics Zoom Pharmacy / Zoom Health NZ · NZ online pharmacy; sachet adherence packing, scheduled delivery, pharmacist support Chemist Warehouse / Sigma Healthcare AU · ASX-listed pharmacy retail franchisor + wholesaler; AU category dominator, also operates in NZ Bargain Chemist NZ · Kiwi-owned discount pharmacy chain + online store + Gap Nobody in this list owns the pharmacy benefit itself Cost Plus routes around the PBM and Amazon Pharmacy dispenses into it; the nearest thing in the corpus is Prescryptive Health in the adjacent 'Pharmacy software + AI-native pharmacy SaaS' segment, which sells pharmacy-benefits software rather than acting as principal. Honest caveat: the integrated PBMs appear nowhere in the 250-row corpus at all, so this is an absence from an AI-native snapshot, not an unoccupied market. + Gap No entity pairs insurance risk with dispensing The insurers here (Oscar, Clover, Sidecar Health — plus Alan in the adjacent payer-SaaS segment) do not dispense, and the pharmacy owners carry no underwriting risk. Amazon is the closest integrator in the whole corpus and stops short of the risk layer: Amazon Pharmacy plus One Medical (adjacent care-delivery segment) is care + dispensing, not insurance. Same caveat — the vertically integrated payer-pharmacy incumbents are simply not in this corpus. + Gap No AI-native Medicaid or dual-eligible plan Coverage here is ACA marketplace (Oscar), Medicare Advantage (Clover), employer major-medical (Sidecar Health) and NZ private (Southern Cross, nib NZ). 'Medicaid' appears once across all 250 rows — Akido Labs, a clinic operator serving Medicaid/Medicare/commercial patients, not a plan. Caveat: state procurement and rate-setting are plausible reasons no venture-backed entrant appears, so read this as an unattempted job, not an easy one. + Gap No specialty or limited-distribution pharmacy Zero of the 250 rows mention specialty pharmacy or infusion; Cost Plus is 2,000+ generics with biosimilars planned, Fuze Health is retail delivery plus diagnostics. This is the highest-dollar, most prior-auth-intensive dispensing channel, and canon shows federal plumbing moving toward it — CMS-0062-P (91 FR 19890, published 14 Apr 2026, comments closed 15 Jun 2026) proposes extending electronic prior authorisation to drugs, though that is payer-side plumbing and a proposed rule, not a market. Caveat: a snapshot skewed to ambient/RCM/ANZ may simply not have sampled this channel. + Gap No administrator of self-funded employer plans 'TPA', 'self-insured', 'stop-loss' and 'captive' return zero rows across all 250 companies; Sidecar Health sells insured employer major-medical rather than administering someone else's risk. This is the one shape in the segment that needs neither reserves nor a carrier licence — which makes its absence from an AI-native snapshot notable rather than self-explanatory. S12 Beauty & wellness saturated 13 4 gaps identified — open for the gaps, incumbents and entry risks. Fresha UK · Subscription-free booking/POS marketplace for beauty+wellness; monetizes payments and marketplace fees; strong in Australasia Zenoti Other · Enterprise all-in-one platform for beauty/wellness/medspa chains; heavy AI push; expanding into fitness/gyms Boulevard US · Premium scheduling/payments/AI-marketing platform for salons, medspas, self-care businesses Moxie US · Medspa business-in-a-box: launch support, EMR, compliance (Compliance Defender), medical-director matching, supplies GPO; revenue-share model Mangomint US · Automation-first salon/spa management for independents; marketing automation flows GlossGenius US · Mobile-first booking/payments app for solo beauty professionals and small teams Phorest EU · Salon + medspa software with marketing/retention focus; strong UK/IE/EU presence Timely NZ · ANZ-born salon/clinic booking software; the NZ incumbent, now inside EverCommerce portfolio RepeatMD Other · Membership/rewards/e-commerce + patient-financing layer ("Medcommerce") on top of aesthetics practices' existing stacks PatientNow US · Aesthetic EMR/practice management for plastic surgery, dermatology, medspas; patient acquisition + RxPhoto imaging Aesthetic Record Other · EMR + practice management purpose-built for medspas/aesthetics; charting, before/after photos, POS Haut.AI EU · B2B AI skin-analysis/diagnostics SaaS licensed to beauty brands and clinics Perfect Corp Asia · Public AI/AR beauty-tech: virtual try-on + AI skin analysis for brands; consumer subscription apps + Gap A clinical-safety layer for injectables — structured adverse-event capture, a complication registry, and product lot/batch traceability tied to the treated patient. Every imaging surface in this list is marketing or charting (PatientNow's RxPhoto, Aesthetic Record's before/after photos) or retail try-on (Perfect Corp, Haut.AI); none of the thirteen records harms or outcomes against the product administered. Caveat: absence from a 250-company August-2026 snapshot is not absence from the market, and a pharmacovigilance vendor outside the beauty-software category would not have been captured here. + Gap A GLP-1 clinical-program layer sold INTO the aesthetics practice — prescribing protocol, titration, side-effect monitoring, adherence and supply. The GLP-1 surge is named as the demand driver in Boulevard's own funding coverage, but the only GLP-1 capability anywhere in this corpus sits with DTC operators in the adjacent care-delivery segment (Eucalyptus/Juniper, Hims & Hers) and consumer metabolic apps (Signos, Levels) — all of whom own the patient rather than equip the clinic. This is a gap and a threat at the same time. + Gap Unbundled multi-state compliance and medical-director infrastructure Moxie is the only company here that touches it (Compliance Defender, medical-director matching), and only inside its own revenue-share operating model for the ~250 medspas it launched — nobody sells supervision, delegation and scope-of-practice infrastructure as an independent layer to the tens of thousands of practices already running on Zenoti, Boulevard, PatientNow or Aesthetic Record. That Moxie built it as a product at all is the corpus's own evidence that the burden is real. + Gap Record portability and a consumer-side longitudinal loop Each of these platforms owns its customers' chart, photo history and calendar with no interoperability between them, and the corpus's six interoperability rows (Metriport, Particle Health, Zus Health, Fasten Health, Epic Systems, Innovaccer) are all built for the clinical-records network, not for aesthetics. Nothing joins a consumer's at-home skin or appearance data to the practice chart: Haut.AI and Perfect Corp sit on the brand/retail side of that line, Aesthetic Record and PatientNow sit inside the visit. S13 Workplace safety software saturated 12 4 gaps identified — open for the gaps, incumbents and entry risks. SafetyCulture AU · Frontline operations/inspections platform (iAuditor); freemium PLG; now pushing agentic AI Intenseye US · Computer-vision AI on existing cameras detecting unsafe acts/conditions for enterprise EHS Protex AI EU · Privacy-preserving CV safety platform plugging into CCTV; Fortune 500 manufacturing/logistics Voxel US · Real-time CV safety on existing security cameras; warehousing, manufacturing, insurance partners EcoOnline UK · EHS + chemical management + ESG SaaS suite; serial acquirer (12 acquisitions in 4 years) HSI / Donesafe AU · EHS training + no-code EHS software platform; PE roll-up Ideagen UK · GRC + EHS software (Ideagen EHS, ex-ProcessMAP); 10,000+ org customers Cority Other · Enterprise EHS + occupational health platform; strong in occ-health/industrial hygiene VelocityEHS US · Mid-market EHS SaaS (Accelerate platform); SDS/chemical heritage Avetta Other · Contractor prequalification / supply-chain risk network (peer of ISN, Veriforce) Safe365 NZ · H&S maturity assessment / capability platform; the NZ incumbent Evotix (SAI360) UK · Mid-market EHS software (ex-SHE Software); merged into SAI360 GRC group + Gap Sensing beyond the camera All three AI entrants here (Intenseye, Protex AI, Voxel) read visible unsafe acts off existing CCTV; nothing in this list captures the non-visual occupational exposure stream — noise, dust/silica, heat, gas, vibration — as sensed data an AI interprets. Cority is described as strong in industrial hygiene, but as a platform over data someone else captured. Honest caveat: exposure-monitoring instrumentation is a real established (largely non-AI) industry, so this is likely as much a corpus-scope effect as a market gap. + Gap Nobody spans the software and the clinical delivery Cority is the only occupational-health platform in this list, while the actual occ-health delivery sits entirely in the adjacent 'H&S consultancy & occupational-health services' rows — Sonic HealthPlus (pre-employment medicals, drug/alcohol testing, injury management), Habit Health (29 occ-health clinics), APM/Konekt (Australia's largest private workplace rehab) — none of which owns software. No company in either list runs incident → clinical care → return-to-work → claim cost end to end. + Gap The AI-native workplace-injury risk carrier Voxel has insurance partners and Tokio Marine as an investor, and EvolutionIQ (adjacent payer segment, sold to CCC for $730M) does AI claims guidance for workers-comp insurers — but no company in this list underwrites the risk it observes. The adjacent 'AI-native ownership of the insurer' rows (Oscar, Clover, Wefox, Sidecar) prove the model exists in health cover; nothing in the 250-company snapshot applies it to workplace injury. + Gap An independent measurement layer for safety-AI claims The strongest outcome numbers here are the vendors' own — Protex AI's 'claims 64% risk reduction in 3 months', Voxel's '147% YoY growth claimed'. The canon table that supplies trial-grade measurement for the clinical segments (JAMA/NEJM ambient-scribe records) carries zero records on EHS, so no third party in this segment produces the evidence an insurer or regulator would price against. S14 Longevity & health checks saturated 11 4 gaps identified — open for the gaps, incumbents and entry risks. Neko Health EU · Daniel Ek's non-invasive full-body scan clinics (sensors + blood), AI-assessed, doctor-reviewed Function Health US · 160+ biomarker lab membership + AI 'medical intelligence'; added MRI/CT via Ezra Superpower US · $199–499/yr 'health super-app': biannual 100+ biomarker panels + AI care plans Prenuvo Other · Whole-body MRI screening clinics with AI-assisted reads; expanding to blood/body-composition Everlab AU · Blood-first preventive/longevity membership: 100+ biomarkers, 1,000+ disease screens, telehealth doctors, AU$499–2,999 tiers Sprinter Health US · At-home phlebotomy + preventive care visits sold to US health plans (B2B2C) Getlabs US · At-home lab collection network for Labcorp/Quest; API for diagnostics companies Fountain Life US · High-end longevity centers (Robbins/Diamandis): full-body MRI, 100+ biomarkers, concierge membership Mito Health US · YC-backed AI biomarker interpretation + at-cost lab testing marketplace Q Bio US · Whole-body 'digital twin' scanner (Mark I + Gemini platform) Anderson Longevity Clinic US · Longevity/hormone-optimization clinic launching nationwide turnkey franchise model + Gap The evidence a payer would need Not one traction field in these eleven rows cites an outcome — no stage-shift, mortality, or cost-offset study; the metrics are funding, members, scans and waitlist length. That absence is load-bearing rather than incidental: canon's REG-CODE-001 records 26 CPT codes describing clinical AI as of Jan 2026, of which 3 are Category I and 23 Category III which 'may lack an associated payment rate and do not guarantee reimbursement' (Bipartisan Policy Center, secondary), and REG-CODE-002 puts FDA AI authorisations against Category I codes at 508:1. Whoever generates the actuarial case converts this from a consumer category into a covered one — but note honestly the gap may be empty because the study is slow and expensive, not because it was overlooked. + Gap The molecular substrate Every company in this list resells commodity reference-lab chemistry and third-party imaging — Getlabs is explicitly a collection network for Labcorp/Quest, Mito Health is an 'at-cost lab testing marketplace', and Function, Superpower and Everlab package panels rather than develop assays. Across all 250 corpus rows there is no sequencing, proteomics, epigenetic-clock, or multi-cancer-early-detection company at all. Stated honestly: this is a gap in the SNAPSHOT, not in the market — molecular diagnostics incumbents plainly exist outside it (canon prices Tempus AI as a listed company at PB-027), so the defensible claim is only that this segment's differentiation sits in packaging and interpretation, never in the underlying test. + Gap The abnormal finding These rows sell the scan, the panel and the care plan, and stop. Neko is 'doctor-reviewed', Everlab has telehealth doctors, Function sells 'medical intelligence' — but no row describes owning the diagnostic workup, the specialist referral, or the economics of what happens after a whole-body MRI surfaces something. Checked against the adjacent care-delivery segment before claiming it: One Medical, Akido Labs, Carbon Health and Crossover Health run primary and urgent care, none of them described as taking screening incidentalomas. This is also the segment's unanswered clinical objection — over-diagnosis — and nothing in the list resolves it. + Gap An employer or benefit channel for the screen itself Sprinter Health is the only row with a non-consumer payer, and it sells the phlebotomy visit rather than the longevity product. Crossover Health, in the adjacent care-delivery segment, runs employer clinics for Apple, Amazon and Microsoft and merged into a ~900-centre operator with Premise in March 2026 — so the employer channel exists in this corpus, just not carrying an AI longevity panel through it. Partially covered, therefore claimed narrowly. S15 Occupational health services saturated 8 4 gaps identified — open for the gaps, incumbents and entry risks. IMPAC NZ · NZ's largest H&S firm: consulting, training, software, PreQual prequalification, recruitment, PPE Habit Health NZ · NZ's largest allied-health/occupational-health provider: workplace medicals, physio, EAP, injury rehab Citation Group UK · Subscription H&S + HR/employment-law + ISO compliance for SMEs — the roll-up playbook Peninsula Group (Employsure in ANZ) AU · Subscription HR + WHS advice, documentation, representation for SMEs in AU/NZ Alcumus UK · Contractor prequalification (SafeContractor), safety compliance services and certifications APM (incl. Konekt) AU · Human services giant; Konekt is Australia's largest private workplace rehab/injury-management provider Sonic HealthPlus AU · Australia's leading occupational-healthcare provider: pre-employment medicals, drug/alcohol testing, injury management, national clinics Safewise NZ · Typical NZ owner-operator SME consultancy: systems, audits, training, ICAM investigations + Gap The advisory hour itself, delivered by AI Citation and Peninsula between them sell human advice lines, document libraries and representation to roughly a quarter of a million SMEs, and neither row — nor any of the other six — mentions AI. The nearest adjacent products sell tooling to the employer (SafetyCulture's inspections platform, Safe365's maturity assessment), not the regulated advice; no company in this corpus sells the consultant's judgement in AI-delivered form. + Gap AI inside the statutory occupational-medicine encounter — fitness-for-work certification, health surveillance, drug/alcohol testing workflow, injury-management case handling. Sonic HealthPlus and Habit Health run these at national scale with no AI in their rows, and the corpus's 13-company ambient-scribe cohort plus the AI-native clinic operators (Akido's ScopeAI, where the model runs the visit and the doctor approves) are all pointed at general and specialty care. Deliberately narrowed: employer-paid *primary* care is NOT a gap — Crossover Health and Premise Health cover it as a ~900-centre operator across 47 states. + Gap An AI operating layer sold TO the fragmented consultancy long tail rather than acquiring it Safewise is flagged in its own row as the exemplar of that tail (owner-operator, Waikato-focused, renting Site App Pro), and the only consolidation mechanism this corpus shows is purchase — IMPAC bought Safeworx, HSI/Donesafe did 8+ follow-on acquisitions, EcoOnline did 12 in four years. Nobody here arms the independent consultant instead of buying them. + Gap A worker-held, portable health-surveillance and fitness-for-work record Pre-employment medicals, exposure monitoring and drug-testing results are generated per employer by providers like Sonic HealthPlus and stay there; the corpus's interoperability rows (Metriport, Particle Health, Zus Health, Fasten, Epic Systems, Innovaccer) are clinical-record plumbing for digital health, none of it occupational. Weakest of the four — the absence from this list is real, but no row evidences demand for it, and the employer who paid for the medical has an obvious interest in the record staying theirs. S16 AI roll-ups contested 7 4 gaps identified — open for the gaps, incumbents and entry risks. HATCo (General Catalyst) US · VC-owned hospital operator; buys health systems as AI-transformation proving grounds Thrive Holdings US · Thrive Capital permanent-capital holdco buying accounting/IT firms, rewiring with AI; OpenAI holds a stake Long Lake US · GC-incubated HOA-management roll-up run on AI agents Metropolis US · Computer-vision checkout-free parking; bought largest US parking operator to deploy its AI Dwelly UK · AI roll-up of UK lettings agencies; closest small-scale analog to a clinic roll-up Radiology Partners US · Largest US radiology roll-up (PE, pre-AI playbook); the leverage cautionary case Tāmaki Health (TPG) NZ · Largest NZ private GP/urgent-care group, PE-owned + Gap No AI-attributable operating result on an acquired healthcare asset Every healthcare row here reports capital or scale, never post-close margin from AI: HATCo's traction is committed capital ($350M over 5 yrs + $200M over 7) on top of the $515M Summa close; Tāmaki Health's is price and clinic count; Radiology Partners' is deleveraging. The one P&L proof point in the segment — Long Lake at $100M EBITDA in under 2 years — is HOA management, not care delivery. Nobody in this list has published what the AI dividend compounds to on a bought clinical book. + Gap The small-cheque, single-specialty clinic roll-up Every healthcare target here is large-ticket ($450m-$515M+, or multi-billion for Radiology Partners). The corpus flags Dwelly (£69M / $93M, 10 agencies, 10k+ properties) as the closest small-scale analog — and it is UK lettings. Sub-$100M AI-native consolidation of independent practices is absent from this list; the practice consolidators in adjacent corpus segments (Heartland Dental 1,900+ offices, Abano 180+ practices, ForHealth 95+ centres, Habit Health 130+ clinics) run pre-AI DSO/PE playbooks. + Gap Purpose-built credit for AI-enabled services roll-ups Every capital row in and around this segment is equity — the Investors segment is 15 equity funds and accelerators, and General Catalyst’s AI-rollup pool is an equity allocation. Only one row in this list combines equity with debt (Dwelly: £32M GC-led equity + £37M Trinity debt), and the one levered healthcare roll-up is the segment's own cautionary case. No specialised lender or credit platform serving this strategy appears anywhere in the 250-company corpus. + Gap Non-US sponsors The roll-up capital here is US-domiciled with two exceptions, and neither is a non-US health sponsor: Dwelly (London) is lettings, and Tāmaki Health (Auckland) was bought by US firm TPG. No EU, Canadian, Japanese or LatAm sponsor is running a healthcare AI roll-up in this list, even though the corpus carries EU health operators (Neko, DentalMonitoring) and EU funds (Heal Capital, Sofinnova) elsewhere. This is the weakest of the four gaps — a 250-company snapshot skewed US/ANZ is exactly where absence is least informative. S17 Interoperability & records saturated 6 4 gaps identified — open for the gaps, incumbents and entry risks. Metriport US · Open-source universal API for patient medical records (Carequality/CommonWell access) Particle Health US · FHIR API to query nationwide patient records for digital health companies Zus Health Other · Shared patient-data platform (Zus Aggregated Profile) for digital health builders Fasten Health US · Open-source, self-hosted personal/family health record manager Epic Systems US · Enterprise EHR for US health systems; MyChart patient portal and Cosmos de-identified research dataset Innovaccer US · Unifies hospital EHR + payer claims data into one record layer, feeding AI agents and care coordination + Gap Payer-side compliance APIs against the CMS-0057-F clocks - narrowed, no longer empty The prior read could say every row retrieved PROVIDER records only; Innovaccer's row now describes unifying hospital EHR with payer claims into one record layer at 80M patient records, so the payer side of the join exists in-segment. What no row's stated product claims is the compliance article itself: the Patient Access / Provider Access / Payer-to-Payer / Prior Authorization FHIR APIs that MA organisations, state Medicaid and CHIP FFS, Medicaid/CHIP managed care and FFE QHP issuers must have live by 1 Jan 2027 (CMS-0057-F, 89 FR 8758), nor the drug-ePA extension proposed in CMS-0062-P (91 FR 19890, Apr 2026). Checked adjacent and unchanged by this refresh - none of the 14 additions landed in the payer segment, whose rows remain claims editing and payment integrity (Machinify, Lyric, Rialtic), PA phone-call agents (SuperDial) or core insurance platforms (EIS, Sapiens). Caveat the caveat: a one-line row is a description, not a product catalogue, so read this as unclaimed in this list - and note that an entrant would now be building beside a $200M-ARR incumbent that already holds both sides of the data it would need. + Gap A patient-held record that is not tethered to one provider's EHR The prior read claimed consumer PHR as near-unfunded white space on the strength of Fasten Health's ~$20K raised; Epic's row falsifies that as written, because MyChart is a patient portal backed by the vendor sitting in 43.7% of US acute-care hospitals and 56.9% of beds. What survives is narrower and still real: MyChart is the patient surface of the institution's own EHR, so its reach is Epic's footprint and it follows the site of care, not the person. Nothing in this roster sells a patient-controlled record that spans vendors, and the only row aimed at that job has ~$20K behind it. Checked adjacent: the consumer hardware rows (Ultrahuman, Dexcom Stelo) hold device-generated signal, not the clinical record. Honest caveat - the obvious real occupants of a person-held record are the phone platforms, which have no row anywhere in this 250-company snapshot, and absence from a 250-company snapshot is not absence from the market. + Gap Non-US standalone record-retrieval rails All six rows are US-HQ'd - San Francisco x3 (Metriport, Fasten, Innovaccer), New York (Particle), Watertown MA (Zus), Verona (Epic) - and Carequality / CommonWell / TEFCA are US-only, as are the KLAS share figures by which the segment's largest name is measured. The corpus does contain non-US health-data plumbing, but every instance is bundled inside another product: Fred IT / eRx (Australia's sole national e-prescription exchange), Medtech Global's ALEX ecosystem in NZ, Corti's EU sovereign-cloud API, Lyrebird Health's AU data sovereignty, Praktika's AU onshore residency. A standalone cross-vendor retrieval API outside the US is still absent from this list, and neither addition touches it - both describe US-only footprints. + Gap Patient identity resolution and consent as the product rather than as an internal step Five of the six rows now do matching inside something else: Zus ships an 'Aggregated Profile', Metriport and Particle must match to return a chart, Innovaccer holds 80M patient records across EHR and claims, and Epic's Cosmos is a de-identified dataset assembled from its own installed base. None sells identity or consent management standalone. The two additions make this claim weaker as an opportunity rather than stronger - the buyers who most need matching are precisely the ones who have now internalised it at national scale, which removes them from the addressable market instead of adding them to it. Weakest of the claims here: the real-world occupants of this job are well known and simply absent from the corpus, so treat it as unrepresented here, not unbuilt. S18 Veterinary AI contested 4 4 gaps identified — open for the gaps, incumbents and entry risks. Scribenote Other · AI scribe for veterinarians; auto-generates SOAP notes from consult audio Digitail US · Cloud all-in-one veterinary PMS with embedded AI (Tails AI) VetRec Other · AI veterinary scribe; SOAP notes plus enterprise tier for vet networks ScribbleVet (Instinct Science) Other · Vet AI scribe with Plumb's drug integration; university partnerships (UC Davis, U Florida) + Gap Turning the note into money Nothing in this list audits whether the invoice matches what was actually done in the room — the veterinary analogue of the coding/CDI attach Abridge and Ambience bundle into their scribes, and of the standalone missed-charge and revenue-integrity products the corpus lists on the human side (SmarterDx, CodaMetrix, Fathom, Nym). Digitail invoices as part of being a PMS; none of the four turns the consult record into recovered charge. Absent from this snapshot is not absent from the market. + Gap Anything pointed at the pet owner All four rows are clinician-facing note or record tooling. No triage, adherence, follow-up-compliance or post-visit explanation layer for the paying client appears anywhere in the 250 rows — the 16-company consumer AI health segment is entirely human-directed. The client is the one who pays the bill in this market and no company here talks to them. + Gap Veterinary imaging or cytology AI The corpus carries 22 human clinical-AI diagnostics companies and three cleared dental radiograph readers (Overjet, Pearl, VideaHealth) and not one veterinary equivalent. This is the weakest of the four gaps and should be read as such: the corpus does not enumerate veterinary diagnostics, analyser or imaging vendors at all, so the hole is at least as likely to be in the snapshot's coverage as in the market. + Gap Anywhere outside North America All four are US or Canadian (Waterloo, Austin, US, US). The corpus finds non-US clinical-AI players easily elsewhere — Heidi in Melbourne across 116 countries, Lyrebird, Nabla in Paris, Tortus in the NHS, Corti in Copenhagen, plus a dense bench of ANZ/UK practice-management incumbents — so the absence of any European, ANZ or LatAm veterinary AI vendor is at least suggestive rather than pure collection bias. S19 Clinical trials open 2 4 gaps identified — open for the gaps, incumbents and entry risks. Lindus Health UK · 'Anti-CRO' — AI/tech-enabled end-to-end clinical trial delivery (Citrus platform) Unlearn.AI US · AI digital twins of trial participants to shrink control arms (TwinRCTs) + Gap Sponsor-side trial operations sold as software rather than as a service — data capture, randomisation, outcome collection, site payments. No row in the 250 sells any of it. This is the weakest of the four gaps and should be read as a sampling artifact: it is precisely the layer where the largest trial-software incumbents live, and this corpus does not sample them. + Gap Regulatory-submission and post-market safety work Nothing in the 250 rows touches safety reporting, dossier assembly, or pharmacovigilance for a sponsor. The nearest adjacent capability is Bunkerhill Health's registry management inside 15 health systems, which is provider-side surveillance, not sponsor-side submission. + Gap Site-side trial enablement Both segment rows sell to, or replace, the sponsor's delivery organisation. The corpus is full of the provider-side buyers a trial needs — 19 care-delivery rows, 13 ambient-documentation rows, health systems already running deployed clinical AI — and not one of them connects that installed base to trial conduct. + Gap Record-graph patient matching for recruitment The plumbing exists in the corpus and the application does not: Metriport, Particle Health and Zus Health all sell nationwide record retrieval, but to digital-health builders, not to sponsors screening for eligibility. An entrant here would rent that graph, not build it. S20 Senior care open 2 4 gaps identified — open for the gaps, incumbents and entry risks. Sensi.AI Asia · Audio-based 24/7 in-home senior monitoring sold to home-care agencies Honor (Home Instead) US · Home-care tech platform that acquired the Home Instead franchise network + Gap Facility-side senior living and skilled nursing operations No assisted-living or SNF record, staffing, or assessment-coding vendor appears in any of the 250 rows; every care-delivery row is clinic, telehealth or hospital-side (Akido Labs, One Medical, Carbon Health, Crossover Health). The corpus names August Health, a US senior-living EHR, only in a disambiguation note attached to an unrelated Indian company — it is never listed as a row. This is the clearest gap, and also the clearest evidence the snapshot under-samples the category rather than the category being empty. + Gap The adult child as the paying buyer Sensi.AI sells to agencies and Honor sells care hours; nothing in the corpus sells eldercare monitoring to the family paying out of pocket. The consumer-hardware segment (Oura, Whoop, Withings, Garmin, Samsung Galaxy Ring, Eight Sleep, Dexcom Stelo, Abbott Lingo) is self-quantification for the wearer — no row is built around a second person's decline — and the corpus's consumer-side record products are either provider-tethered or tiny — Epic Systems' MyChart rides the health system's own record, Fasten Health is a self-hosted PHR at ~$20K raised — neither being monitoring. + Gap Passive sensing that is neither audio nor worn Sensi.AI is audio-only. No row in the corpus does contactless radar, computer-vision, or bed/floor sensing for falls and gait, and every consumer device listed requires the elder to wear and charge it — the population least likely to comply. Caveat: this is a gap in the sampled list, and contactless home sensing is an active commercial category outside it. + Gap Dementia-specific longitudinal cognitive and behavioural management as its own product Sensi.AI lists cognitive change as one of 100+ insight types, not as the product; the entire Mental-health + behavioural AI segment holds one company (Slingshot AI's Ash, a consumer therapy app). Nothing in the corpus is organised around decline trajectory, caregiver burden, or behavioural-symptom management. Weakest of the four gaps — Sensi already detects the signal, so the gap is packaging and workflow, not detection. S21 Mental health saturated 10 5 gaps identified — open for the gaps, incumbents and entry risks. Slingshot AI (Ash) US · Foundation model for psychology; Ash consumer AI-therapy app Lyra Health US · Employer-contracted mental-health benefit — therapy, coaching and medication management through its own provider network, plus Lyra AI, a conversational AI guide that supplements human care between sessions (Oct 2025 pilot, moving to US general availability during 2026) Spring Health US · Employer and health-plan mental-health benefit; ML triage routes members to coaching, therapy or medication, with Guide, an AI-led support experience, in limited rollout to selected customers since Apr 2026 Sonder AU · 24/7 employee wellbeing and safety subscription sold to employers — nurse-led triage, psychology and in-person safety response delivered by human clinicians; AI features were stated as roadmap at its Oct 2025 raise, not a shipped model Modern Health US · Employer-contracted mental-health benefit (coaching, therapy, psychiatry, self-guided tools); AI does provider matching and content recommendation today, while the member-facing Skye assistant and AI care plans are still written in future tense on its own site Unmind UK · Workplace mental-health platform sold to employers; ships Nova, a member-facing generative-AI wellbeing agent that gives conversational support and routes users to human care TELUS Health (employer solutions, ex-LifeWorks) Other · Incumbent employer-paid EAP/wellbeing programme — counselling, financial and legal advice, telehealth via the TELUS Health One app — delivered by human counsellors, with AI appearing as member-facing chatbots and internal agent assist rather than a clinical model Converge International AU · Employer-paid EAP across ANZ — human counselling, critical-incident response, manager support and psychosocial risk assessment — plus a wearables-fed wellbeing app acquired with HeadUp Labs; no clinical AI model of its own is disclosed on its current product pages Headspace (incl. Ginger / Headspace Care) US · Employer- and health-plan-contracted mental health benefit stacking meditation content, coaching, therapy and psychiatry (the former Ginger network, merged 2021 and rebranded Headspace Care in Oct 2023) with Ebb, a shipped in-app generative-AI companion trained on motivational interviewing that reflects and routes to human care rather than treating Calm Health US · Employer-, health-plan- and provider-contracted behavioural health app: validated GAD-7/PHQ-9 screeners plus self-reported health inputs route members into clinician-built condition programmes or out to an EAP or in-person care; personalisation is algorithmic and no generative-AI clinical tool has been announced for Calm Health + Gap Vendor-neutral measurement of an employer's behavioural benefit All ten rows sell the care and grade their own homework. Spring Health's Guide launch rests on a Spring Health study (up to 25% greater symptom improvement for those needing care most; 60% of members staying consistently engaged in therapy); Lyra labels its own guide 'clinical-grade' and 'clinically vetted' in its own announcements. No row here is the party an employer hires to check whether the benefit worked. This is the same gap the ambient-scribe segment carries and it is sharper in this one, because the outcome measure is a self-reported symptom scale administered by the vendor being assessed. + Gap Substance-use disorder as a product A regex sweep of all rows' name/what/traction for addiction, substance, SUD and opioid returns one hit in the whole corpus — Sonic HealthPlus, filed under occupational health, for drug-and-alcohol testing, which is employer screening rather than treatment. Requalify the old absolute claim, though: TELUS Health and Converge International sell general EAP counselling whose conventional scope includes substance use, so the need is now adjacently reachable through a human counsellor. What has no occupant is a company whose product is addiction care. + Gap Serious mental illness and inpatient psychiatry — the surviving half of an older, broader claim The prior read asserted that every behaviourally-adjacent row in the corpus was consumer, self-directed and subclinical; that is now false. Sonder sells 24/7 nurse-led triage with critical-incident and trauma response delivered by human clinicians, and Lyra's AI guide carries a risk-flagging path that escalates to a live 24/7 care team. Acute escalation has occupants. What is still empty is the population: an employer-paid benefit is structurally aimed at the employed and the mild-to-moderate. Modern Health and Headspace list psychiatry, but that is outpatient prescribing inside a benefit — no row here runs a psychiatric bed or manages schizophrenia or bipolar I, and SMI is where behavioural spend concentrates. + Gap A cleared generative-AI mental-health device The FDA has cleared none. Its Digital Health Advisory Committee first took the class up on 6 Nov 2025, working a hypothetical prescription LLM therapy device for adults with major depressive disorder and landing on tailored premarket evidence, explicit human-escalation pathways and risk-stratified postmarket surveillance, with bias, hallucination and sycophancy named as the novel risks (Orrick client alert, Nov 2025). Every AI in this segment is deliberately specified to sit outside that path — supplementing, triaging, routing, reflecting, never treating. Whoever goes through the device door first owns something none of these ten has, and pays for it in evidence rather than in distribution. + Gap Clinician-side behavioural documentation and medical-necessity review — kept from the prior read, but downgraded from empty to contested. Spring Health's Alma acquisition brought national and regional payer relationships plus provider infrastructure for independent in-network clinicians, which is the exact substrate a therapy-note or utilisation-review product needs. Same caveat as before: this is a product-specification gap in the list, not proof the ambient-scribe incumbents cannot extend into the therapy session — extending remains the obvious move. Point-in-time snapshots, not market censuses. Absence from a radar is not evidence a company does not exist, and funding and traction age fast — re-verify a row against its source before it informs a decision. Not investment, commercial or legal advice.
