What you get every issue:


A crystal ball, I'm just kidding. Three curated healthcare AI stories with my take. Companies doing the work. Skills and techniques worth knowing. And opportunities worth pursuing. No hype. Just signal.

Three things worth your time this week:

  1. OPENAI AND ANTHROPIC ARE NOW HEALTHCARE VENDORS
    OpenAI launched OpenAI for Healthcare with HCA, Cedars-Sinai, UCSF, MSK, Stanford Children's, Boston Children's. Anthropic grew its healthcare team 200%+ YoY and shipped Claude for Healthcare with BAA support.
    The headline: "Frontier labs go vertical."
    The reality: The model providers just became the application providers. The defensibility goalposts moved overnight.
    My take: If your healthcare AI startup is a wrapper, you no longer have a moat. Defensibility is now data flywheels, EHR-embedded UX, or owning a regulated counterparty workflow.
    Read it: https://openai.com/index/openai-for-healthcare/
  2. WISeR IS LIVE — AI PRIOR AUTH IN ORIGINAL MEDICARE
    The Wasteful and Inappropriate Service Reduction Model has been operating since Jan 1 in NJ, OH, OK, TX, AZ, WA. AI/ML screens 17 outpatient services for medical necessity; clinicians make final denial calls.
    Why it matters: Original Medicare has never had prior auth before. 6.4M beneficiaries are now in the test pool.
    The pattern: Every regulated counterparty in healthcare is becoming an agent-native handoff. Prior auth, claims, denials, registries, OASIS, MDS — same pattern, different vertical.
    My take: Provider-side counter-tooling is about to surge. The companies that ship the patient-and-provider-facing PA appeals workflow win the next 12 months.
    Full story: https://www.cms.gov/priorities/innovation/innovation-models/wiser
  3. ADOPTION BEATS CERTIFICATION — EVERY TIME
    DeepHealth bought Gleamer for $269M because Gleamer had 700+ hospital contracts. Oxipit had a world-first autonomous AI certification and ~$28K in 2024 revenue. Guess which one set the precedent.
    But here's the nuance: This isn't anti-regulatory. It's anti-novelty-without-distribution.
    Why this matters for builders: A 510(k) is the floor, not the ceiling. The exit math is install base × annual contract value. Build a clearance-and-distribution motion, not a clearance-only motion.
    Source: https://hitconsultant.net/2026/04/20/digital-health-funding-q1-2026-ai-ma-rebound/

COMPANIES DOING THE WORK
Amperos Health — $16M Series A from Bessemer; AI-native denial management for the long tail of small/mid practices. $700M in claims recovered annually.
Latent — $80M Series A; pure-play AI prior auth ahead of CMS-0057-F.
Medisolv — Acquired Health Elements AI; 96% accuracy clinical data abstraction for ACC/STS/AHA registries. Quality measurement is officially being eaten by agents.

AI TOOLS WORTH KNOWING
Amazon Connect Health — Five healthcare agents (identity verification, appointment management, ambient documentation, patient insights, medical coding) — Hyperscaler-as-a-vendor; matters because EHR vendors can embed via SDK without full rebuild.
Abridge Linked Evidence — Maps every AI-generated note element back to source transcript audio — Best in KLAS for ambient AI two years running; the trust traceability pattern every clinical AI tool now copies.
DocToDoor RPM (PointClickCare-integrated) — AI-enabled biometric trend analysis for post-acute settings — Direct adjacency to my CardioWatch work; PointClickCare integration is the right post-acute wedge.

WORK & PARTNERSHIP OPPORTUNITIES
VP/Head of AI Product — OpenAI for Healthcare — Direct hiring need post-launch; HCA / Cedars-Sinai / UCSF as anchor accounts.
Head of Clinical AI / CMO — Anthropic — 200%+ team growth needs a CMO-tier physician leader to anchor vertical sales motion.
VP Product, Prior Auth — Latent — Pure-play AI prior auth post-Series A; FHIR fluency required.
WISeR Medical Director (multiple participating vendors) — Net-new role category from a CMS innovation model.

WHAT I'M WORKING ON:
Building the agent-native handoff thesis into paid AI Audits. The question is: which workflows, which tools, which sequence, and what does safe deployment actually look like in a regulated environment.

Most operators don't have anyone who can answer that question clearly. Not their EHR vendor. Not their IT team. Not the AI startup that keeps emailing them.If you'd like to see the one-pager, reply to this comment.

Also, share this newsletter with a friend or colleague.

— Victor

Reply to this email if you want to discuss any of these. I read every message.

Yuuki Edge • Healthcare AI Intelligence • yuukiedge.com