Policy-compiler agent
Turns written rules into checks on every action.
Checked against clinicians
Clinician and compliance sign-off; rule tests against labelled cases.
Where a person decides
Policy owner approves every rule version.
Works on
Voice, RCM, payer, EHR agents.
On the roadmap. Tell us if you need it sooner. Reviewed .
The problem it solves
Rules live in PDFs: scope of practice, payer policy, state disclosure laws. Models don't reliably follow them, and nobody checks every action.
How it works
Read the policy
The agent turns written rules into precise, testable checks.
Review with your team
Compliance or clinical leads approve each check.
Run on every action
Checks run automatically on outputs and actions, deterministically.
An example, illustrative
Policy PDF: 'Disclose AI use to the patient.'
Check: every patient message includes the disclosure. 3 failures this week.
The written policies and a named approver.
A check is only as good as its reading of the policy, so a human approves each one.
What it helps measure
Quick answers
What does the policy compiler do?
Compile written rules (scope of practice, payer policy, state disclosure, authority limits) into checkable gates.
How is the policy compiler checked against clinicians?
Clinician and compliance sign-off; rule tests against labelled cases.
Does this give legal advice?
No. It converts rules you provide into checks. Legal interpretation stays with your counsel.
Which rules work best?
Clear yes-or-no rules, such as required disclosures, code validity and scope limits.
What can the policy compiler not do?
Policy owner approves every rule version. No agent makes a clinical decision.