Industries · Healthcare operations

Demonstrated · design partner wanted

Let AI act on the record — with a clinician in command.

A human approval and an AI action are the same kind of governed event in CHP — which is exactly what a clinical sign-off needs. This page shows how it would work with today’s primitives; the access and retention layer an audit requires is what we build with a design partner.

Read the essay: “Who authorized the AI step?” →

The pain

AI scribes and agents draft notes, summarize charts, and prepare orders. A clinician has to stay in command and sign off — and when something is questioned, you need to show what the AI did, who approved it, and that the record is intact, under HIPAA scrutiny.

The trigger

A safety event, a payer audit, or a malpractice inquiry asks: “What did the automated step do, and who authorized it?” Reconstructing that from application logs is slow and not defensible.

Who owns it

The CMIO and clinical operations, with compliance — accountable for patient safety, the integrity of the record, and what automation is allowed to touch.

What CHP would do

Each AI action crosses a capability boundary that records a structured event with the acting subject and correlation; the clinician sign-off is a governed approval in the same trace. Payloads are redacted by default, so the evidence captures who-did-what without storing the PHI body.

The demonstrated record

The AI action — attributed, before sign-off.

This is a CHP invocation: the capability, the acting subject, and the correlation that ties it to the encounter. The clinician approval is recorded as the next governed event in the same trace, so the record shows both what was proposed and who authorized it.

Demonstrated, not yet shipped: invocation, approval, and redacted evidence exist today. Role-based access and compliant retention are what we build with a design partner.

invocation — the attributed AI action
json
{
  "invocation_id": "inv_session_abc_001",
  "capability_id": "schedule_technician",
  "version": "1.0.0",
  "mode": "sync",
  "correlation": { "correlation_id": "session-abc" },
  "subject": { "id": "agent://planning-assistant" },
  "payload": {
    "job_id": "job_456",
    "window": "tomorrow"
  },
  "requested_at": "2026-06-16T15:14:20.000Z"
}

Putting AI into clinical workflows?

Bring a real sign-off flow. We’ll map the AI action and clinician approval onto the protocol and build the access and retention your compliance team requires, together.

Build it with us