HealthAI In development

In development — not yet available

The chart, drafted from the conversation. Signed only by you.

HealthAI is an AI-native, FHIR-native electronic health record for solo and very small cash-pay practices. With the patient's consent it listens to the visit, drafts a structured chart entry and a clinical note, and hands both to you to edit and sign.

Request early access

Please don't include patient or health information in email.

A visual sequence built from synthetic sample data: a spoken conversation becomes transcript text, which becomes draft chart entries with clinical codes, awaiting clinician review.

The problem

The documentation follows you home.

When you practice alone, there is no billing department and no one to hand the notes to. The writing waits until the last patient has left, and the structured part of the chart — coded problems, the medication list, results — waits longer still. Most software built to fix that was designed for large organizations and then sold down to solo practices.

How it works

Consent, conversation, draft, review, signature.

The software prepares the work. Nothing reaches the record without your approval.

  1. The software prepares.

    Consent

    Recording starts only after the patient has explicitly consented.

  2. The conversation

    You talk to your patient. The audio is held in memory, transcribed, and zeroed immediately afterward. It is never written to disk and never sent to a third-party transcription vendor — only the transcript text moves on.

  3. A draft chart and a draft note

    The transcript becomes clinically-coded findings, a structured chart entry, and a clinical note. Codes are selected from a bounded list of real ICD-10-CM, RxNorm and LOINC codes; the model cannot write one of its own.

  4. The clinician decides.

    Your review

    Everything arrives as a draft, item by item. Advisory notes — a possible interaction, a possible discrepancy between the conversation and the chart — sit alongside it, for you to accept or dismiss. You decide what goes in the chart.

  5. Your signature

    Only a licensed clinician can sign. That rule is enforced in three independent places, and no automated process anywhere in the system can produce a signed note.

Interface preview

What the review screen looks like.

An interface preview, built with synthetic records. Every name, code and value below is fabricated test data.

What makes it different

Specifics, not adjectives.

Seven properties of how the system is built. The specificity is the point: this is a system of record, not a documentation layer on top of someone else's EHR.

  • Audio is never stored.

    The recording is held in memory only, zeroed immediately after transcription, never written to disk, and never sent to a third-party transcription vendor. Only the transcript text moves onward.

  • The model cannot invent a code.

    A separate terminology step builds a bounded candidate list of real ICD-10-CM, RxNorm and LOINC codes, and the language model may only select from it. A validation gate then independently re-checks the finished record and rejects any code that wasn't in the candidate set. The rule is enforced by code, not by asking the model nicely.

  • Every record is validated before it is written.

    Each record passes HL7's own official FHIR validator against the US Core 6.1.0 implementation guide before anything is stored. If the validator can't run, that counts as a failure — "couldn't check" is never treated as "passed."

  • Only a licensed clinician can sign.

    Enforced three independent times: in the application code, in the server's access-control policy, and in a runtime guard on every background process. No automated process anywhere in the system can produce a signed note.

  • Everything starts as a draft.

    Anything the AI produces stays a draft until a clinician approves it, item by item.

  • Standards-native means portable.

    The chart is standard FHIR R4 rather than a proprietary schema, so the data isn't locked in.

  • The audit log cannot be rewritten.

    Every write is audit-logged, and the log is append-only. It cannot be edited or deleted — including by the system's own internal processes.

Built on standards

A chart you could take with you.

The record underneath the note is US Core FHIR R4 — the US profile of the FHIR standard for health data exchange — not a schema invented for this product. Problems, medications, allergies, observations and orders are stored as standard resources, and the terminologies are ICD-10-CM, RxNorm and LOINC.

Practically, that means the chart can be read by other software without a custom export, and the vocabulary in it is the vocabulary the rest of US healthcare already uses.

Single-tenant, self-hosted.

The clinic runs its own instance. The data stays in that instance.

Who it's for

Built for one clinician at a time.

HealthAI is being built for solo and very small practices that take payment directly from patients — behavioral health therapists first, direct primary care physicians in parallel. It is not built for hospitals or large groups, and it is not a scaled-down version of something that was.

Early access

It is early, and I am one person.

HealthAI is in development. You cannot buy it or sign up for it today. There are no customers yet, and there are no testimonials on this page because there is no one to quote.

If you want to be involved early, write to me. Early means hands-on: you would be working directly with the person building the software, on a system that is still changing. If that sounds like more work than you want right now, later is a perfectly good answer.

Request early access

Please don't include patient or health information in email.