One patient record across every line your practice runs
MedOp is the AI-native system of record for behavioral health, rehab therapy (PT/OT/SLP) and medical aesthetics — built vertical-deep, priced per provider with everything included, and run by a governed workforce of 27 AI agents. This paper sets out the platform, the market, the economics and the evidence, and says plainly where a capability is planned rather than shipped.
20 pages · every platform figure anchored to a tagged production build · every market figure cited · no form to fill in.
The paper makes five claims
The problem is labour, not software features.
The three verticals are large, growing and structurally under-served — and integrated practices that run more than one line have no vendor at all.
MedOp is a system of record, not an AI overlay. The overlay is a commodity now, and it is terminable by the EHR it sits on.
The differentiator is the machinery, not the agent count: minute bands from the session clock, the 8-minute rule as a tested function, an enforced Good Faith Exam, Part 2 accounting, a CoCM registry — on a governed agent substrate.
The commercial model restores expansion revenue and removes the reasons practices distrust their vendor: published per-provider prices, a metered AI workforce with a hard cap, a 0% platform fee on payments, and a Fair Terms Charter in the contract.
Ten sections, two appendices, every figure sourced
Section 01
Executive summary
Five claims, each with evidence: the problem is labour; the three verticals are large and under-served; system of record, not overlay; mechanics over agent count; a commercial model built on expansion and fair terms.
Section 02
The problem
13 hours a week on prior authorization (AMA, 2026). 11.8% of claims denied on first submission (Kodiak, 2025). 86% of rehab therapists burned out, 59% naming documentation (Ensora, 2026). Software that meters every add-on and charges to export your data.
Section 03
The market
61.5M US adults with a mental illness and half untreated (SAMHSA). 457,000 people employed in PT/OT/SLP offices (BLS). 10,488 med spas averaging $1.40M (AmSpa). And the segment nobody serves: integrated practices running more than one line.
Section 04
The platform
Scheduling, intake, ambient documentation, grounded coding, claims and RCM, e-prescribing with EPCS, portal and secure messaging, payments, engagement, telehealth, reporting and a public API — one system, one record, composable verticals with cross-line containment.
Section 05
The AI workforce
Four pods, 27 agents, and the substrate underneath: memory and retrieval, reflection, calibrated confidence, one approval inbox, kill switches, cost attribution in CI, closed-loop improvement. A denied claim, step by step.
Section 06
What sets MedOp apart
Six checkable differentiators — from one record across every line to database-enforced tenancy and fair terms as contract language — and the competitive landscape in one paragraph.
Section 07
Benefits
For clinicians, the front office and the owner, plus an illustrative value model: roughly 36 hours a week reclaimed for a three-provider practice, against a bill smaller than the software it replaces.
Section 08
Revenue model and economics
Published per-provider pricing, the AI Workforce meter, Revenue Recovery at 2.5%, worked examples, why per-provider matters, unit economics, and the seven-clause Fair Terms Charter.
Section 09
Security, compliance and trust
Engineered controls — RLS on 71 tables, a tamper-evident PHI audit chain, MFA, Part 2 gating, AI governance — and an honest statement of attestation status and the trust roadmap.
Section 10
Engineering and roadmap
1,138 commits, 546 tagged releases, 45 ADRs, 21 CI gates, 112 models, ~186k lines of TypeScript. The Proof Scope, vertical depth in order, and the trust surface through Q2 2027.
Appendix A lists all 27 operator-facing agents with pod, model tier and job. Appendix B lists every third-party source. Capabilities described as planned, scheduled or in the Proof Scope are not yet generally available; illustrative value figures are models built from cited public benchmarks, not measured customer results.
Read it, then check it
Every platform claim in the paper can be verified in a demo or a sandbox. Every market claim can be verified at its source.