Case study · Healthcare

A 14-hospital system turned agent activity into evidence its regulator accepts

Prior-authorisation and utilisation-review agents were producing decisions nobody could attribute. Ninety days later every action carried a hashed record naming the clinician whose authority was exercised.

Regional health system · 14 hospitals · United States · published August 19, 2026

Download the PDF Apply for early access

6 days

to assemble a payer audit package

from 11 weeks

100%

of committed agent actions attributable to a named reviewer

from 0% under shared service accounts

3.1%

of proposals refused at verification

record contradicted the proposed action

PHI routes to endpoints without verified training exclusion

Summary

A regional health system running prior-authorisation and utilisation-review agents could not answer the only question its regulator and its own compliance committee cared about: under whose authority did this action commit? Orcher bound each directive to a licensed reviewer, verified the proposal against the EHR before commit, and hashed the result. Evidence assembly for a payer audit fell from eleven weeks to six days.

Orcher components enforced

Immutable Audit Ledger
Hashes the verified action permanently. Evidence, not logs.
Role Identity Fabric
Binds the directive to a human and a role, cryptographically and revocably.
Logic Scrubber
Verifies the proposed action against systems of record before commit.

Solution: Governance & audit evidence

The situation

What was happening before Orcher.

Three agent workflows — prior authorisation, utilisation review and discharge summary drafting — were live across 14 hospitals, writing into the EHR through service accounts shared by an entire department.
Application logs recorded that an action occurred. They did not record which clinician's authority permitted it, and they rotated every 30 days.
A payer audit of 2,400 prior-authorisation decisions consumed eleven weeks of clinical informatics time reconstructing the reasoning after the fact.

What Orcher enforced

The control surface, component by component.

Role Identity Fabric replaced the shared service accounts: each directive class was bound to a licensed reviewer role, with the binding revocable the moment credentialing changed.
Logic Scrubber verified every proposed action against the patient record and the payer's own coverage rules before commit, refusing writes where the record contradicted the proposal.
Immutable Audit Ledger hashed the verified action, the role that authorised it, the model that produced it and the cost, into a permanent record the compliance committee reads directly.
Data Control Gateway enforced the enterprise data boundary so no PHI could reach a provider endpoint without a training-exclusion guarantee verified before routing.

The data

Measured over the engagement window.

Measure	Before Orcher	After 90 days	Change
Audit package assembly	11 weeks	6 days	−92%
Actions with named human authority	0 of 2,400	3,180 of 3,180	complete
Agent actions reversed post-hoc	47 per quarter	6 per quarter	−87%
Average review cost per directive	$1.84	$0.61	−67%
Mean time to answer 'who authorised this?'	4.5 days	under 1 minute	−99%

Ninety-day measurement window, three agent workflows, 3,180 committed actions.

Outcomes

What the organisation did next.

The compliance committee moved from quarterly sampling to continuous review, because the evidence is queryable rather than reconstructed.
Two agent workflows that had been paused pending governance sign-off went back into production inside the window.
Clinical informatics recovered roughly nine weeks of effort per audit cycle.
The system extended the same pattern to a fourth workflow — coding validation — without a new governance review, because the control surface was already accepted.

“We were not asking for better logs. We were asking for something a regulator would accept as proof of authority, and that is a different artefact entirely.”

— Chief Compliance Officer, regional health system (partner declined attribution by name)

Read further

The research behind this engagement.

Disclosure. Design-partner engagement, anonymised at the partner's request. Figures are measured by Orcher's own observability and ledger instrumentation over the stated period and have not been independently audited. Market figures carry their own source line.

More engagements

Other organisations that had to prove the same thing.

Insurance

Run this pattern on your own workflow

Design partners deploy Orcher on one directive class, instrument it the way this study was instrumented, and keep the evidence whatever they decide afterwards.

Apply for early access All case studies