The AI harness for physician-governed healthcare

See every clinical AI output. Control who signs it. Own the audit trail.

Less than one FTE to operate a complete AI-native care stack — from patient intake through RTM billing, with a physician reviewing and signing every output before it touches a patient or a claim. The harness enforces that review at the schema level, writes every attestation to a 7-year WORM-locked audit chain, and produces FHIR R4 output automatically. You don't build physician governance; you inherit it.

See the stack Talk to us
3 Revenue brands — one shared harness
<1 FTE To operate the full stack
FHIR R4 Native output — every attestation
7-yr WORM-locked audit chain
Three products

Families. Surgeons. Attestation.

Each brand solves a distinct problem. All three run on the same AI harness with the same physician governance model.

For families

co-op.care

Worker-owned home care for aging families. AI generates care plans, schedules caregivers, and drafts letters of medical necessity for HSA/FSA reimbursement. A physician reviews and signs every LMN.

co-op.care →
For surgeons

SurgeonValue

Nine AI agents for orthopedic practice management: missed billing detection, remote therapeutic monitoring, prior authorization, patient outcomes tracking, and referral capture. Connects directly to Epic, Cerner, and athena.

surgeonvalue.com →
For attestation

ClinicalSwipe

A physician review marketplace. AI companies submit clinical outputs. Physicians review and attest via a structured swipe interface. Every accept, modify, or reject generates FHIR-structured data and a signed audit record.

clinicalswipe.com →
Numbers that matter

The economics behind the infrastructure.

These are the unit economics SolvingHealth runs on — the same numbers a health system adopting the harness inherits.

$64–147K
Per surgeon / year in missed billing codes
AAPC research finds 37% of encounter notes are under-coded — documentation already supports a higher code, but no system flags it before the claim leaves. SurgeonValue Wonder Bill targets this gap. Average first panel scan finds $3–8K in a single month of already-documented work.
$199
Per LMN at 10–20 letters / hour physician review
AI drafts the letter of medical necessity. Physician reviews and signs in 3–5 minutes. Immediate revenue — no caregivers, no license, no operations required.
$936
Per family / year in HSA savings unlocked by LMN
co-op.care families use IRS 213(d) LMNs to pay for companion care with pre-tax HSA/FSA dollars. The LMN pays for itself in the first month.
2,500
Hospitals in scope for CMS CJR-X mandatory episodes
When CMS finalizes CJR-X, every participating hospital needs a 90-day episode capture layer. SurgeonValue CJR-X Tracker is that layer — active July 2026. Surgeons who know their episode cost before reconciliation hold the negotiating position.
13 hrs
Physician + staff time lost to prior auth per week
The AMA's 2024 prior authorization survey (n=1,000 physicians) found practices complete 39 prior authorization requests per physician per week, consuming 13 hours of physician and staff time. SurgeonValue Prior Auth Agent targets this time directly — letter in under 60 seconds, per-payer formatting included.
Infrastructure thesis

We are in the Attestation Era.

AI generates the draft. A physician reviews, edits, and signs. The signature is the product — not the generation.

The SolvingHealth harness is an AI orchestration layer that wraps any clinical workflow. A single JavaScript embed routes requests to the appropriate AI model, enforces physician review requirements, and writes every attestation to a WORM-anchored audit chain. You don't need to build physician governance — you inherit it.

The shift is structural, not aspirational. Healthcare AI companies that generate outputs without a documented physician review trail face a simple problem: when a claim is challenged, a compliance officer asks for evidence, or a patient outcome is questioned, a policy document that says "physicians review AI outputs" is not a defense. A cryptographically verifiable ledger entry — physician NPI, timestamp, review duration, scroll depth, output hash — is. The harness generates that record automatically on every transaction.

We are an operator, not a vendor. SolvingHealth built this infrastructure to run its own care company, not to sell software to health systems. The proof of value is in our own P&L — the same P&L we open-book to any partner. Build trust through operations, then license the stack.

Hard Intercept

Physician sign-off is enforced by the database schema, not by policy. Review time (minimum 30 seconds wall clock), viewport activity (minimum 20 seconds foreground), and scroll depth (minimum 80% of the output) are all validated by CHECK constraints before a signature record is written. The AI output cannot reach a patient or a billing claim without a schema-enforced review event. Policy documents are not an OIG defense; database constraints are.

FHIR-Native

Every attestation produces structured FHIR R4 output — not a PDF, not a free-text summary. Home care Omaha System nursing assessment data maps to hospital-grade clinical records automatically, generating the kind of longitudinal data corpus that health systems and payers license. The data corpus is the long-term moat; it grows with every review and becomes more valuable as the network scales.

Audit Chain

Daily SHA-256 digests of all decision ledger entries are written to Cloudflare R2 object storage with a 7-year WORM compliance lock. The next day's cron job verifies by reading back the digest. Every attestation — accept, modify, or reject — is indexed in the ledger with physician NPI, timestamp, review metrics, and the signed output hash. The ledger is OIG audit-ready from transaction one, not retrofit-ready at audit notice.

AKS/Stark Built In

Compliance scoring is baked into the decision ledger — not a post-hoc legal check. Flat per-encounter fees (not revenue-share percentages), not referral-volume-tied. The billing architecture follows OIG Advisory Opinion 25-03 as a structural template for MSO-physician practice arrangements. Every transaction is structurally defensible before a compliance officer looks at it.

What physician attestation looks like — end to end

The thesis is simple: AI generates the clinical draft; a licensed physician reviews, edits, and signs before it touches a patient or a claim. Every step from generation to signature to audit record happens inside the harness — no stitching required. Here is what one letter of medical necessity looks like in practice.

Step 1 · Generation
AI drafts the output

A family member submits a care assessment through co-op.care. The harness routes the request to the appropriate model (Claude for complex clinical reasoning). An LMN draft is produced in under 30 seconds — citing IRS 213(d) eligibility criteria and the patient's specific documented conditions.

Step 2 · Review gate
The Hard Intercept fires

The draft is routed to a licensed physician in the ClinicalSwipe queue. The physician sees the full draft, the patient's relevant history, and the clinical justification. The system tracks wall clock time, foreground-tab time, and scroll depth. A signature is not accepted until all three thresholds are met — enforced by database constraints, not policy.

Step 3 · Attestation
The physician signs

The physician accepts, modifies, or rejects the draft. Each decision — including every edit — is written to the decision ledger with the physician's NPI, timestamp, the review metrics, and a SHA-256 hash of the signed output. The signed LMN is delivered to the family as a FHIR R4 document and a PDF. Average review time: 3–5 minutes per letter.

Step 4 · Audit anchor
WORM-locked for 7 years

At 23:59 MT daily, a cron job computes a SHA-256 digest of all that day's decision ledger entries and writes it to Cloudflare R2 with a 7-year WORM compliance lock. The ledger anchor is verified by read-back the following day. An OIG auditor asking for the attestation history of any signed output gets a cryptographically verifiable chain from generation to signature.

Entry A patient For surgeons SurgeonValue PROMs · missed billing · prior auth Attestation ClinicalSwipe Physician reviews + signs SolvingHealth The harness — routing, attestation, FHIR, audit chain For families co-op.care Home care · LMN · HSA unlock Output Attested data
The thesis
One harness underneath three products.
SolvingHealth is the routing layer between every brand. Press Follow a patient to watch one person move from a surgeon's office, through physician attestation, into worker-owned home care — with their data becoming a FHIR-native, audit-locked record at every hop. Or tap any node to see what connects to it.

Tap any box to isolate its connections. The harness in the center touches all of them.

For physician vibe coders

A physician who wants a personal AI assistant uses Gemini Spark. A physician who wants to build HIPAA-compliant clinical AI tools uses JetBridge + Claude.

The SolvingHealth harness is the professional-grade stack for clinicians building with AI. Physician governance baked in. FHIR-native output. No compliance consulting required.

You are an MD who has realized that the fastest way to change clinical practice is to ship software, not papers. You can write a system prompt. You can call an API. You need the infrastructure that makes what you build defensible.

JetBridge Harness is the production layer: physician attestation schema, OIG audit chain, FHIR R4 output, and AKS-clean billing hooks — pre-built so you can focus on the clinical problem you actually want to solve.

"We built this to run our own care company. The same rails that generate $520K/year in LMN revenue are the same rails you license. Operational proof, not a sales deck."

See the JetBridge Harness Talk to a builder
SolvingHealth Harness — embed in 1 line
// Add to any page — physician governance included <script src="https://harnesshealth.ai/harness.js" data-site="your-site-id" defer></script> // What you inherit: // ✓ Physician review enforcement (Hard Intercept) // ✓ SHA-256 audit chain → Cloudflare R2 (7yr) // ✓ FHIR R4 structured output // ✓ AKS/Stark flat-fee billing hooks // ✓ OIG audit-ready from transaction #1
For partners

Three lanes into the network.

Organizations that want to plug in — whether you're a health system, a payer, or a developer — have a named lane with named economics.

Lane 1 — Health systems & employers
Bulk membership
$3–7 / seat / month
Covers access to the full family-care surface network — co-op.care, ComfortCard, CareGoals, and the LMN pipeline. Embedded benefit, no EHR integration required.
Lane 2 — Payers & device companies
Attested data + cohort access
PMPM / custom
Physician-attested outcomes data with full audit chain. FHIR R4 export. Payers, device manufacturers, pharma co-marketing, and government programs. The data corpus grows with every review.
Lane 3 — Developers & AI companies
Software licensing
$49–499 / month SDK
FHIR connectors are MIT-free. The intelligence layer — encounter linking, compliance scoring, physician network — is the subscription. One embed, full governance.
Partnership inquiries

Talk to us.

We work with health systems, payers, device companies, and AI developers who want physician-governed AI infrastructure they don't have to build from scratch.

We respond within one business day.

Your inquiry was received. We'll be in touch within one business day.