18 agents. One job each.
They measure, find causes and block. A person approves every change.
Checked against clinicians
Each agent that judges anything is calibrated on clinician labels and shows its error rates.
A person stays in charge
No agent makes a clinical decision, contacts a patient or changes production alone.
The dots show where each agent sits in the five-step loop. "In audits" means it's delivered inside a Quality Audit today. "Building" and "Roadmap" are in development.
Specify
Measure
Clinical reviewer
Grades answers against rubrics your clinicians wrote.
Omission auditor
Finds what the note left out.
Hallucination auditor
Finds claims the source never supports.
Dose and negation checker
Catches wrong doses, flipped negations and left-versus-right slips.
Severity grader
Ranks every error by how much harm it could do.
Judge calibrator
Shows how often the AI grader itself is wrong.
Drift monitor
Re-tests on a schedule and flags real changes.
Equity auditor
Compares error rates across patient groups and languages.
Replay and simulate
Diagnose
Repair and prove
Escalation gate
Forces a handoff to a human when red flags appear.
Policy compiler
Turns written rules into checks on every action.
Repair agent
Proposes fixes, accepted only if they hold on new cases.
Evidence-pack agent
Assembles results into the documents reviewers ask for.
Want an agent
on your product?
Tell us which one you need first. Most start with the omission auditor or the replay diagnostician.
hello@precision.example