diff --git a/domains/health/AI scribes reached 92 percent provider adoption in under 3 years because documentation is the rare healthcare workflow where AI value is immediate unambiguous and low-risk.md b/domains/health/AI scribes reached 92 percent provider adoption in under 3 years because documentation is the rare healthcare workflow where AI value is immediate unambiguous and low-risk.md
new file mode 100644
index 0000000..f0ffd33
--- /dev/null
+++ b/domains/health/AI scribes reached 92 percent provider adoption in under 3 years because documentation is the rare healthcare workflow where AI value is immediate unambiguous and low-risk.md	
@@ -0,0 +1,37 @@
+---
+type: claim
+domain: health
+description: "92% of US health systems deploying AI scribes by March 2025 — a 2-3 year adoption curve vs 15 years for EHRs — because documentation is the one clinical workflow where AI improvement is immediately measurable, carries minimal patient risk, and delivers revenue capture gains"
+confidence: proven
+source: "Bessemer Venture Partners, State of Health AI 2026 (bvp.com/atlas/state-of-health-ai-2026)"
+created: 2026-03-07
+---
+
+# AI scribes reached 92 percent provider adoption in under 3 years because documentation is the rare healthcare workflow where AI value is immediate unambiguous and low-risk
+
+By March 2025, 92% of US provider health systems were deploying, implementing, or piloting AI scribes. This technology scaled in 2-3 years — compared to 15 years for EHR adoption. The speed is not an anomaly. It reveals which healthcare workflows AI can actually penetrate and why.
+
+Documentation is structurally different from every other clinical AI application:
+
+**Immediate, measurable value.** Early adopters report 10-15% revenue capture improvements in year one through improved coding and documentation accuracy. Since [[ambient AI documentation reduces physician documentation burden by 73 percent but the relationship between automation and burnout is more complex than time savings alone]], the productivity gain is large enough to justify the investment without complex ROI modeling.
+
+**Minimal patient risk.** A documentation error doesn't directly harm a patient the way a diagnostic error might. The risk profile is administrative, not clinical. This eliminates the regulatory friction and liability concerns that slow clinical AI adoption.
+
+**No workflow disruption.** AI scribes listen to existing physician-patient conversations and generate notes afterward. Unlike clinical decision support tools that require physicians to change how they practice, scribes fit into the existing workflow invisibly.
+
+**Clear competitive market.** Abridge ($300M Series E at $5B valuation), Microsoft DAX Copilot (via $19.7B Nuance acquisition), and Epic's AI Charting are all scaling rapidly. The competition validates the category while driving rapid iteration.
+
+This adoption velocity matters beyond documentation itself. AI scribes are the beachhead — the first AI tool that earns clinician trust through daily use. Since [[the physician role shifts from information processor to relationship manager as AI automates documentation triage and evidence synthesis]], scribes are the entry point for a broader transformation of the physician role. Clinicians who use AI scribes daily (67% use AI tools daily, 90%+ weekly per Bessemer data) develop comfort and trust with AI-assisted workflows that make them receptive to clinical AI applications downstream.
+
+The contrast is instructive: since [[medical LLM benchmark performance does not translate to clinical impact because physicians with and without AI access achieve similar diagnostic accuracy in randomized trials]], clinical AI faces a trust and integration gap that documentation AI has already crossed. The lesson is that healthcare AI adoption follows the path of least institutional resistance, not the path of greatest clinical potential.
+
+---
+
+Relevant Notes:
+- [[ambient AI documentation reduces physician documentation burden by 73 percent but the relationship between automation and burnout is more complex than time savings alone]] — the clinical evidence behind AI scribe value
+- [[the physician role shifts from information processor to relationship manager as AI automates documentation triage and evidence synthesis]] — scribes as beachhead for broader role transformation
+- [[medical LLM benchmark performance does not translate to clinical impact because physicians with and without AI access achieve similar diagnostic accuracy in randomized trials]] — why clinical AI lags documentation AI in adoption
+- [[OpenEvidence became the fastest-adopted clinical technology in history reaching 40 percent of US physicians daily within two years]] — parallel rapid adoption in clinical decision support
+
+Topics:
+- [[_map]]
diff --git a/domains/health/AI-native health companies achieve 3-5x the revenue productivity of traditional health services because AI eliminates the linear scaling constraint between headcount and output.md b/domains/health/AI-native health companies achieve 3-5x the revenue productivity of traditional health services because AI eliminates the linear scaling constraint between headcount and output.md
new file mode 100644
index 0000000..b5a8aea
--- /dev/null
+++ b/domains/health/AI-native health companies achieve 3-5x the revenue productivity of traditional health services because AI eliminates the linear scaling constraint between headcount and output.md	
@@ -0,0 +1,38 @@
+---
+type: claim
+domain: health
+description: "AI-native healthcare companies generate $500K-1M+ ARR per FTE compared to $100-200K for traditional health services, compressing time-to-$100M-ARR from 10+ years to under 5, creating a structural unit economics advantage that incumbents cannot match without rebuilding"
+confidence: likely
+source: "Bessemer Venture Partners, State of Health AI 2026 (bvp.com/atlas/state-of-health-ai-2026)"
+created: 2026-03-07
+---
+
+# AI-native health companies achieve 3-5x the revenue productivity of traditional health services because AI eliminates the linear scaling constraint between headcount and output
+
+Healthcare has historically been a labor-intensive industry where revenue scales linearly with headcount. More patients require more clinicians, more billing staff, more care coordinators. This linear scaling constrains margins and creates the workforce bottlenecks that limit access. AI-native health companies are breaking this constraint.
+
+**The productivity ladder (Bessemer 2026 data):**
+- Traditional healthcare services: $100-200K ARR per FTE
+- Healthcare SaaS (pre-AI): $200-400K ARR per FTE
+- AI-native healthcare: $500K-1M+ ARR per FTE
+
+This 3-5x productivity gap creates two structural advantages. First, **margin structure**: AI-native companies achieve 70-80%+ gross margins at scale, comparable to software companies, while traditional health services operate at 20-40% margins. Second, **time-to-scale**: AI-native healthcare companies reach $100M+ ARR in under 5 years, compared to 10+ years for traditional healthcare software and even longer for services companies.
+
+The evidence is concentrated in a few breakout companies. Hinge Health posted 72% annualized revenue growth with 26% free cash flow margins — a Rule of 40 score of 98%. Tempus grew at 85% with a 9.3x EV/revenue multiple. Function Health reached $100M+ ARR in under two years. These aren't outliers exploiting temporary market conditions — they're demonstrating a structural shift in healthcare economics.
+
+The mechanism: AI replaces the marginal human hours in documentation, triage, coding, claims processing, and care coordination that previously scaled linearly. Each AI-augmented worker handles 3-5x the patient volume. This is why 92% of US provider systems are deploying AI scribes — the ROI is immediate and measurable, with early adopters reporting 10-15% revenue capture improvements in year one through improved coding and documentation.
+
+The implication for the healthcare attractor state: since [[the healthcare attractor state is a prevention-first system where aligned payment continuous monitoring and AI-augmented care delivery create a flywheel that profits from health rather than sickness]], AI-native unit economics make the prevention-first model economically viable in a way that labor-intensive care delivery never could. Prevention requires continuous engagement with healthy populations — economically impossible at $100-200K ARR per FTE, potentially viable at $500K-1M+.
+
+Since [[healthcares defensible layer is where atoms become bits because physical-to-digital conversion generates the data that powers AI care while building patient trust that software alone cannot create]], the most defensible AI-native health companies will be those that control both the data generation (atoms) and the AI processing (bits), not pure-play AI software companies layered onto someone else's clinical data.
+
+---
+
+Relevant Notes:
+- [[the healthcare attractor state is a prevention-first system where aligned payment continuous monitoring and AI-augmented care delivery create a flywheel that profits from health rather than sickness]] — AI-native economics enable the attractor state
+- [[healthcares defensible layer is where atoms become bits because physical-to-digital conversion generates the data that powers AI care while building patient trust that software alone cannot create]] — where the defensible AI-native positions concentrate
+- [[healthcare AI funding follows a winner-take-most pattern with category leaders absorbing capital at unprecedented velocity while 35 percent of deals are flat or down rounds]] — capital flows to the companies achieving these unit economics
+- [[ambient AI documentation reduces physician documentation burden by 73 percent but the relationship between automation and burnout is more complex than time savings alone]] — AI scribes as the beachhead for productivity transformation
+
+Topics:
+- [[_map]]
diff --git a/domains/health/CMS is creating AI-specific reimbursement codes which will formalize a two-speed adoption system where proven AI applications get payment parity while experimental ones remain in cash-pay limbo.md b/domains/health/CMS is creating AI-specific reimbursement codes which will formalize a two-speed adoption system where proven AI applications get payment parity while experimental ones remain in cash-pay limbo.md
new file mode 100644
index 0000000..7ae7f69
--- /dev/null
+++ b/domains/health/CMS is creating AI-specific reimbursement codes which will formalize a two-speed adoption system where proven AI applications get payment parity while experimental ones remain in cash-pay limbo.md	
@@ -0,0 +1,35 @@
+---
+type: claim
+domain: health
+description: "CMS adding category I CPT codes for AI-assisted diagnosis (diabetic retinopathy, coronary plaque) and testing category III codes for AI ECG, echocardiograms, and ultrasound — creating the first formal reimbursement pathway for clinical AI"
+confidence: likely
+source: "Bessemer Venture Partners, State of Health AI 2026 (bvp.com/atlas/state-of-health-ai-2026)"
+created: 2026-03-07
+---
+
+# CMS is creating AI-specific reimbursement codes which will formalize a two-speed adoption system where proven AI applications get payment parity while experimental ones remain in cash-pay limbo
+
+CMS is building the reimbursement infrastructure for clinical AI through a graduated code system. Category I (permanent) CPT codes now exist for AI-assisted diabetic retinopathy autonomous screening, with coronary plaque assessment AI added in 2026. Multiple category III (temporary/experimental) codes are under testing for AI-augmented ECG interpretation, echocardiogram analysis, and breast/thyroid ultrasound.
+
+This creates a formal two-speed adoption system:
+
+**Speed 1: Reimbursed AI (CMS-paced).** Applications that earn category I codes get payment parity with traditional clinical procedures. This unlocks provider adoption at scale because the economic model works within existing revenue cycles. Diabetic retinopathy screening was first because it has the cleanest evidence base — FDA-cleared autonomous AI (IDx-DR/LumineticsCore) with randomized trial data showing non-inferiority to ophthalmologists.
+
+**Speed 2: Cash-pay AI (consumer-paced).** Applications without reimbursement codes depend on consumer willingness to pay or provider willingness to absorb cost. RadNet's AI mammography ($40 consumer co-pay, 36% uptake) and Function Health ($499/year direct-to-consumer) demonstrate this pathway works but creates access inequality.
+
+The two-speed system has a structural feedback loop. Category III codes generate real-world evidence data on AI performance, outcomes, and cost-effectiveness. This evidence supports the transition to category I codes. But the 3-5 year timeline from category III testing to category I permanence means the reimbursement system inherently lags clinical capability by half a decade.
+
+Since [[healthcare AI regulation needs blank-sheet redesign because the FDA drug-and-device model built for static products cannot govern continuously learning software]], the CPT code system faces a similar structural problem: codes are static descriptions of procedures, but AI capabilities evolve continuously. A coronary plaque assessment AI in 2026 will be materially different from the same product in 2028, yet the reimbursement code remains fixed.
+
+The investment implication: companies positioned at the category I boundary — where evidence is sufficient for permanent reimbursement — capture disproportionate value. The transition from category III to category I is the healthcare AI equivalent of the regulatory moat. Since [[value-based care transitions stall at the payment boundary because 60 percent of payments touch value metrics but only 14 percent bear full risk]], AI reimbursement codes could accelerate VBC transition by making AI-assisted prevention and chronic disease management economically viable within fee-for-service billing.
+
+---
+
+Relevant Notes:
+- [[healthcare AI regulation needs blank-sheet redesign because the FDA drug-and-device model built for static products cannot govern continuously learning software]] — the static-code problem applies to CMS as well as FDA
+- [[value-based care transitions stall at the payment boundary because 60 percent of payments touch value metrics but only 14 percent bear full risk]] — AI codes could bridge the payment gap
+- [[AI diagnostic triage achieves 97 percent sensitivity across 14 conditions making AI-first screening viable for all imaging and pathology]] — the clinical capability awaiting reimbursement infrastructure
+- [[the healthcare attractor state is a prevention-first system where aligned payment continuous monitoring and AI-augmented care delivery create a flywheel that profits from health rather than sickness]] — reimbursement codes are a prerequisite for the attractor state within fee-for-service
+
+Topics:
+- [[_map]]
diff --git a/domains/health/FDA is replacing animal testing with AI models and organ-on-chip as the default preclinical pathway which will compress drug development timelines and reduce the 90 percent clinical failure rate.md b/domains/health/FDA is replacing animal testing with AI models and organ-on-chip as the default preclinical pathway which will compress drug development timelines and reduce the 90 percent clinical failure rate.md
new file mode 100644
index 0000000..ae43955
--- /dev/null
+++ b/domains/health/FDA is replacing animal testing with AI models and organ-on-chip as the default preclinical pathway which will compress drug development timelines and reduce the 90 percent clinical failure rate.md	
@@ -0,0 +1,35 @@
+---
+type: claim
+domain: health
+description: "FDA's April 2025 roadmap aims to make animal studies 'the exception rather than the norm' within 3-5 years, endorsing AI-based models and organ-on-chip for preclinical testing — a structural shift that could address the 90% clinical failure rate by improving translatability"
+confidence: experimental
+source: "Bessemer Venture Partners, State of Health AI 2026 (bvp.com/atlas/state-of-health-ai-2026); FDA Strategic Roadmap April 2025"
+created: 2026-03-07
+---
+
+# FDA is replacing animal testing with AI models and organ-on-chip as the default preclinical pathway which will compress drug development timelines and reduce the 90 percent clinical failure rate
+
+In April 2025, the FDA announced a strategic roadmap to fundamentally restructure preclinical drug testing. The goal: make animal studies "the exception rather than the norm" within 3-5 years. The endorsed alternatives are AI-based predictive models, organ-on-chip systems, and in silico toxicity prediction.
+
+This is not an incremental reform. The current preclinical paradigm — animal testing as the required gateway to human trials — has a catastrophic translatability problem: over 90% of drugs that appear safe and effective in animals fail in human efficacy or safety testing. The entire drug development pipeline (10-15 years, $1-2 billion per approved therapy) is built on a preclinical foundation that produces a 90%+ false positive rate.
+
+Since [[AI compresses drug discovery timelines by 30-40 percent but has not yet improved the 90 percent clinical failure rate that determines industry economics]], the current wave of AI drug discovery primarily accelerates the timeline without fixing the underlying translatability problem. AI can identify drug candidates faster, but if those candidates still go through animal models that don't predict human outcomes, the 90% failure rate persists.
+
+The FDA's shift changes this equation. AI-based preclinical models trained on human biology data (not mouse biology extrapolated to humans) could improve translatability at the point where it matters most — before $100M+ is spent on clinical trials. Organ-on-chip systems use human tissue to test drug responses in physiologically relevant microenvironments. In silico toxicity prediction identifies safety signals from molecular structure rather than animal observation.
+
+The economic context makes this urgent. Over 70% of Western preclinical work is currently offshored to China, creating both a strategic vulnerability and a quality concern. AI-native preclinical platforms would reshore this work while simultaneously improving it.
+
+The confidence level is experimental because the roadmap is announced but not yet implemented. The 3-5 year timeline is aspirational. Regulatory inertia, pharma company conservatism, and the validation requirements for new preclinical approaches all create friction. But the direction is clear, and the economic pressure (90% failure rate at $1-2B per drug) creates strong incentive for adoption.
+
+Since [[gene editing is shifting from ex vivo to in vivo delivery via lipid nanoparticles which will reduce curative therapy costs from millions to hundreds of thousands per treatment]], the combination of AI-native preclinical testing and cheaper gene therapy delivery could simultaneously improve success rates and reduce costs — a compounding effect on the drug development pipeline.
+
+---
+
+Relevant Notes:
+- [[AI compresses drug discovery timelines by 30-40 percent but has not yet improved the 90 percent clinical failure rate that determines industry economics]] — the problem this FDA shift addresses
+- [[gene editing is shifting from ex vivo to in vivo delivery via lipid nanoparticles which will reduce curative therapy costs from millions to hundreds of thousands per treatment]] — parallel cost reduction in therapeutics delivery
+- [[the healthcare cost curve bends up through 2035 because new curative and screening capabilities create more treatable conditions faster than prices decline]] — improved preclinical success rates could accelerate this curve
+- [[healthcare AI regulation needs blank-sheet redesign because the FDA drug-and-device model built for static products cannot govern continuously learning software]] — FDA demonstrating willingness for structural regulatory change
+
+Topics:
+- [[_map]]
diff --git a/domains/health/_map.md b/domains/health/_map.md
index 1fe3c10..3c2c476 100644
--- a/domains/health/_map.md
+++ b/domains/health/_map.md
@@ -24,6 +24,9 @@ Vida's domain spans the structural transformation of healthcare from reactive si
 - [[medical LLM benchmark performance does not translate to clinical impact because physicians with and without AI access achieve similar diagnostic accuracy in randomized trials]] — the benchmark-to-clinical gap
 - [[human-in-the-loop clinical AI degrades to worse-than-AI-alone because physicians both de-skill from reliance and introduce errors when overriding correct outputs]] — physician overrides degrade AI from 90% to 68%
 - [[healthcare AI regulation needs blank-sheet redesign because the FDA drug-and-device model built for static products cannot govern continuously learning software]] — Wachter's physician-licensing model for AI regulation
+- [[AI scribes reached 92 percent provider adoption in under 3 years because documentation is the rare healthcare workflow where AI value is immediate unambiguous and low-risk]] — fastest-adopting clinical AI category; beachhead for broader AI trust
+- [[AI-native health companies achieve 3-5x the revenue productivity of traditional health services because AI eliminates the linear scaling constraint between headcount and output]] — structural unit economics shift: $500K-1M+ ARR/FTE vs $100-200K
+- [[consumer willingness to pay out of pocket for AI-enhanced care is outpacing reimbursement creating a cash-pay adoption pathway that bypasses traditional payer gatekeeping]] — RadNet: 36% pay OOP for AI mammography, 43% higher detection
 
 ## Value-Based Care & Devoted Health
 - [[Devoted is the fastest-growing MA plan at 121 percent growth because purpose-built technology outperforms acquisition-based vertical integration during CMS tightening]] — proof of concept for purpose-built payvidor model during CMS tightening
@@ -41,6 +44,7 @@ Vida's domain spans the structural transformation of healthcare from reactive si
 - [[gene editing is shifting from ex vivo to in vivo delivery via lipid nanoparticles which will reduce curative therapy costs from millions to hundreds of thousands per treatment]] — scalability breakthrough for curative medicine
 - [[personalized mRNA cancer vaccines show sustained 49 percent reduction in melanoma recurrence after five years representing a genuinely novel therapeutic paradigm]] — mRNA platform beyond COVID
 - [[the healthcare cost curve bends up through 2035 because new curative and screening capabilities create more treatable conditions faster than prices decline]] — net cost trajectory: inflationary through transition
+- [[FDA is replacing animal testing with AI models and organ-on-chip as the default preclinical pathway which will compress drug development timelines and reduce the 90 percent clinical failure rate]] — FDA April 2025 roadmap: animal studies to become "exception" within 3-5 years
 
 ## Mental Health & Digital Therapeutics
 - [[prescription digital therapeutics failed as a business model because FDA clearance creates regulatory cost without the pricing power that justifies it for near-zero marginal cost software]] — Pear, Akili, Woebot: the DTx autopsy
@@ -54,6 +58,7 @@ Vida's domain spans the structural transformation of healthcare from reactive si
 - [[CMS 2027 chart review exclusion targets vertical integration profit arbitrage by removing upcoded diagnoses from MA risk scoring]] — CMS targeting acquisition-based vertical integration
 - [[anti-payvidor legislation targets all insurer-provider integration without distinguishing acquisition-based arbitrage from purpose-built care delivery]] — structural separation bills threatening payvidor model
 - [[Kaiser Permanentes 80-year tripartite structure is the strongest precedent for purpose-built payvidor exemptions because any structural separation bill that captures Kaiser faces 12.5 million members and Californias entire healthcare infrastructure]] — Kaiser's 80-year precedent for purpose-built integration
+- [[CMS is creating AI-specific reimbursement codes which will formalize a two-speed adoption system where proven AI applications get payment parity while experimental ones remain in cash-pay limbo]] — category I/III CPT codes for AI-assisted diagnosis
 
 ## Epidemiological Transition & Risk Landscape
 - [[the epidemiological transition marks the shift from material scarcity to social disadvantage as the primary driver of health outcomes in developed nations]] — the fundamental discontinuity
diff --git a/domains/health/consumer willingness to pay out of pocket for AI-enhanced care is outpacing reimbursement creating a cash-pay adoption pathway that bypasses traditional payer gatekeeping.md b/domains/health/consumer willingness to pay out of pocket for AI-enhanced care is outpacing reimbursement creating a cash-pay adoption pathway that bypasses traditional payer gatekeeping.md
new file mode 100644
index 0000000..ed5db4c
--- /dev/null
+++ b/domains/health/consumer willingness to pay out of pocket for AI-enhanced care is outpacing reimbursement creating a cash-pay adoption pathway that bypasses traditional payer gatekeeping.md	
@@ -0,0 +1,39 @@
+---
+type: claim
+domain: health
+description: "RadNet's AI mammography study shows 36% of women paying $40 out-of-pocket for AI screening with 43% higher cancer detection, suggesting consumer demand will drive AI adoption faster than CMS reimbursement codes"
+confidence: likely
+source: "Bessemer Venture Partners, State of Health AI 2026 (bvp.com/atlas/state-of-health-ai-2026)"
+created: 2026-03-07
+---
+
+# consumer willingness to pay out of pocket for AI-enhanced care is outpacing reimbursement creating a cash-pay adoption pathway that bypasses traditional payer gatekeeping
+
+The conventional assumption in healthcare AI is that adoption requires reimbursement — if CMS doesn't create a CPT code and payers don't cover it, the technology stalls. RadNet's mammography study demolishes this assumption with the largest real-world evidence dataset to date.
+
+**The RadNet evidence (747,604 women):**
+- 36% of women offered AI-enhanced mammography chose to pay $40 out-of-pocket for it
+- AI-enhanced screening detected cancer 43% more often than standard screening
+- Overall cancer detection rate was 21% higher with AI
+- Positive predictive value for cancer was 15% higher
+
+The significance isn't just the clinical improvement — it's that more than a third of patients voluntarily paid a premium for AI-enhanced care when given the choice. This establishes a cash-pay adoption pathway that bypasses the traditional reimbursement bottleneck entirely.
+
+This pattern is accelerating beyond imaging. Function Health reached $100M+ ARR in under two years selling $499/year comprehensive lab testing directly to consumers — no insurance involvement. ChatGPT Health reports 40 million+ daily users, with 1 in 5 asking health-related questions weekly. The consumer is pulling AI into healthcare faster than the payment system can accommodate it.
+
+The structural implication: healthcare AI adoption will follow a dual-track model. Track 1 (reimbursement-dependent) moves at CMS speed — years of pilot programs, category III CPT codes, payment experiments. Track 2 (consumer cash-pay) moves at consumer technology speed — months to adoption, driven by demonstrated clinical value and willingness to pay. Track 2 will establish the use cases and evidence base that eventually forces Track 1 to follow.
+
+Since [[the FDA now separates wellness devices from medical devices based on claims not sensor technology enabling health insights without full medical device classification]], the regulatory framework already enables consumer-direct health AI without full medical device classification — removing one of the two traditional gatekeepers (FDA and CMS) from the adoption pathway.
+
+The risk: cash-pay adoption creates a two-tier system where AI-enhanced care accrues to those who can afford the premium. This is the equity tension in consumer-led health innovation — early access is wealth-stratified until reimbursement catches up.
+
+---
+
+Relevant Notes:
+- [[the FDA now separates wellness devices from medical devices based on claims not sensor technology enabling health insights without full medical device classification]] — regulatory framework enabling consumer-direct health AI
+- [[Function Health drives down diagnostic conversion costs to 499 per year for 100-plus lab tests making atoms-to-bits health data generation accessible at consumer scale]] — another cash-pay model bypassing traditional reimbursement
+- [[AI diagnostic triage achieves 97 percent sensitivity across 14 conditions making AI-first screening viable for all imaging and pathology]] — the clinical capability that makes consumer willingness-to-pay rational
+- [[healthcare AI creates a Jevons paradox because adding capacity to sick care induces more demand for sick care]] — consumer cash-pay could either accelerate the Jevons paradox (more diagnosis → more treatment) or enable prevention-first models depending on what consumers choose to buy
+
+Topics:
+- [[_map]]
diff --git a/domains/health/healthcare AI funding follows a winner-take-most pattern with category leaders absorbing capital at unprecedented velocity while 35 percent of deals are flat or down rounds.md b/domains/health/healthcare AI funding follows a winner-take-most pattern with category leaders absorbing capital at unprecedented velocity while 35 percent of deals are flat or down rounds.md
index e75e7f9..bb428d9 100644
--- a/domains/health/healthcare AI funding follows a winner-take-most pattern with category leaders absorbing capital at unprecedented velocity while 35 percent of deals are flat or down rounds.md	
+++ b/domains/health/healthcare AI funding follows a winner-take-most pattern with category leaders absorbing capital at unprecedented velocity while 35 percent of deals are flat or down rounds.md	
@@ -3,7 +3,7 @@ description: Global healthcare venture financing reached 60.4 billion in 2025 bu
 type: claim
 domain: health
 created: 2026-02-17
-source: "Health tech VC landscape analysis February 2026; OpenEvidence Abridge Hippocratic AI fundraising disclosures; Agilon Health SEC filings; Rock Health digital health funding reports 2025"
+source: "Health tech VC landscape analysis February 2026; OpenEvidence Abridge Hippocratic AI fundraising disclosures; Agilon Health SEC filings; Rock Health digital health funding reports 2025; Bessemer Venture Partners State of Health AI 2026"
 confidence: likely
 ---
 
@@ -17,6 +17,8 @@ Global healthcare venture financing reached $60.4 billion in 2025, the strongest
 
 The emerging consensus: healthcare AI is a platform shift, not a bubble, but the shift creates winner-take-most dynamics where category leaders absorb capital while everyone else fights for scraps. The IPO window is opening cautiously (Hinge Health at ~60% discount, Insilico Medicine in Hong Kong). 2026 demands fundamentals: clinical-grade evidence, regulatory clarity, proven path to profitability. 15 new unicorns were minted in 2025, predominantly in AI-enabled categories.
 
+**Bessemer corroboration (January 2026):** 527 VC deals in 2025 totaling an estimated $14B deployed. Average deal size increased 42% year-over-year (from $20.7M to $29.3M). Series D+ valuations jumped 63%. AI companies captured 55% of health tech funding (up from 37% in 2024). For every $1 invested in AI broadly, $0.22 goes to healthcare AI — exceeding healthcare's 18% GDP share. The Health Tech 2.0 IPO wave produced 6 companies with $36.6B combined market cap, averaging 67% annualized revenue growth. Health tech M&A hit 400 deals in 2025 (up from 350 in 2024), with strategic acquirers consolidating AI capabilities.
+
 ---
 
 Relevant Notes: