Healthcare & Disability /Drafting live field guide · 8 min

Prior-Auth Compiler

Maps the chart against payer criteria, drafts the prior-auth packet with an inline citation on every match, and flags the gaps — clinician approves before submission.

Open the live demo

theater/demos/healthcare_prior-auth-compiler.html · sandbox · read-only

Open

FIG. 1

The live demo, running on fabricated data. Open it to step through the full flow — every output is shown for a person to approve before anything happens.

How it would work

Reads the referral and clinical notes against the insurer's pre-authorisation criteria, drafts the submission with a source citation on every matched line, and lays out the gaps for a clinician or pre-auth officer to approve before anything goes to the insurer.

Input 01

The referral + the chart

The referral letter, imaging report and GP notes for one admission, plus the private health insurer's pre-authorisation criteria for the planned MBS item — diagnosis, conservative-management history, imaging grade, prosthesis.

Agent 02

Extracts, maps, cites

Pulls each fact the criteria demand — membership, diagnosis code, imaging grade, prior treatments, provider and facility — maps each to the criterion it satisfies, and attaches a citation back to the source document.

Output 03

A draft, with its working shown

A complete pre-auth packet with every matched criterion, its citation, a Missing Documentation gap summary and any benefit-gap flag, laid out for a clinician or pre-auth officer to correct and approve before it is submitted.

Where it works well

It does the criterion-by-criterion evidence hunt every time, in full, and shows where each fact came from.

Done by hand it is the better part of an hour per case — and the gap that delays approval is usually a single missing line nobody noticed.
Best for a pre-auth officer or billing team processing high-volume electives: joint replacement, cataract, cholecystectomy, specialty imaging.
At dozens of pre-auths a week, the recaptured hours go back into chasing the genuinely missing records, not re-reading the chart that already has them.

The slow, invisible cost of a pre-auth is the evidence-gathering — tracing the Kellgren-Lawrence grade to the imaging report, the conservative-management history to the GP notes, the prosthesis to the Prostheses List — across three or four documents per case, for every admission.

Where it works badly

It is confidently wrong when the criteria are stale or the chart is ambiguous — and the packet looks more finished than the evidence behind it.

Weak where a criterion needs a clinical judgement the notes only imply — "function-limiting", "conservative management failed" — rather than state. It should flag, not infer.
If your insurer criteria aren't held as current, structured rules, the draft is built on whatever version someone last pasted in — and it can't tell you that.

The honest test

If you cannot say, right now, which version of the insurer's pre-auth criteria this packet was drafted against — this tool makes your wrong answer faster, not safer.

Point it at an insurer's prior policy version, or a note where "trialled physio" has no dates, and it drafts a clean, professional packet that asserts a criterion is met when it isn't. That is the trap.

What it doesn't do — and shouldn't

It drafts. A clinician approves. That boundary is deliberate.

WHAT IT DOES

Surfaces each criterion and the source line it matched it to

Lists everything the criteria require that the chart doesn't supply

Flags a benefit gap where the prosthesis or item sits outside the insurer's no-gap arrangement

WHAT IT WON’T

Submit the packet to the insurer

Assert clinical eligibility or medical necessity

Give the patient informed financial consent

A pre-authorisation submission asserts clinical facts to a private health insurer and underpins the informed financial consent the hospital must obtain in writing before a scheduled admission, under the Private Health Insurance (Health Insurance Business) Rules 2018. A wrong criterion or an unflagged gap means a declined claim, an unexpected patient bill, or a treating practitioner's name attached to a claim that doesn't hold. The accountable clinician stays on the decision because the consequence lands on them — not the tool.

What your data has to look like

The chart as readable documents, and the insurer's pre-auth criteria as current, structured rules.

Typical readiness

across orgs we see, before the first job

Referral, imaging and clinical notes as text, not scans

Usual weak point

Insurer pre-authorisation criteria, structured and versioned

Needs shaping

Conservative-management history with dates

Needs shaping

MBS item and Prostheses List codes for the procedure

Usual weak point

Membership and provider details for the submission block

Usually ready

The real first job

The insurer criteria are usually the weak point — held as a PDF someone downloaded once, or rules in a billing officer's head. Getting each insurer's pre-auth criteria into current, structured, versioned form is usually the real first job — larger and more valuable than the drafting layer on top. Once the criteria and the chart are clean, every pre-auth after that is faster and right by default.

Right fit if…

You process high volumes of elective pre-auths with repeatable insurer criteria

Joint replacement, cataract, cholecystectomy, specialty imaging — structured procedures

You can point to each insurer's current pre-auth criteria, by item, today

Your referrals and reports arrive as text or structured records, not flat scans

Walk away if…

Most of your pre-auths are clinically borderline, judgement-heavy one-offs

Insurer criteria live as last-downloaded PDFs nobody owns or versions

Clinical notes are dictated free text where treatments have no dates

You want a tool that asserts medical necessity so a clinician doesn't have to

Open questions

The worried-buyer questions, answered straight

It can draft a wrong match — which is exactly why nothing is submitted on its say-so. It maps each insurer pre-authorisation criterion to the source line in the chart, shows that citation, and lists under Missing Documentation anything the criteria require that the chart doesn’t supply. A clinician or pre-auth officer checks those before approving. The tool surfaces the evidence; the person stands behind the assertion to the insurer.

It works from readable text — a diagnosis code, an imaging grade, dated treatment history. Where “trialled conservative management” has no dates, or the imaging report is a flattened scan, it flags the gap rather than inventing the detail. Getting referrals and reports into clean, machine-readable form is usually the first piece of work — and the piece that pays off across every pre-auth after.

No. It removes the evidence hunt — the forty minutes of re-reading the chart to find the line that satisfies each criterion — so the pre-auth officer spends their time on the genuinely missing records and the clinician on the call the insurer can’t make: whether the procedure is clinically indicated and the documentation honestly supports it. The submission is still theirs. The recaptured time goes back into patient-facing and chase work.

Current to the policy in force. Private health insurers revise pre-authorisation criteria, MBS item descriptors change, and the Prostheses List that sets prosthesis benefits is reissued — point it at a prior version and it drafts confidently wrong. The honest test: do you know, today, which version of the insurer’s criteria this packet was drafted against?

A patient’s referral, diagnosis and clinical notes are health information — the most sensitive category of personal information under the Privacy Act 1988 and the Australian Privacy Principles, requiring consent to collect and use. Any deployment runs against your own systems and data handling, not a shared pool — we scope where the data sits and who can see it as part of the build. The demo here runs entirely on fabricated data; Margaret Hughes is not a real patient.

It drafts to the criteria it’s given and flags where evidence is missing or a benefit gap exists, but it does not certify compliance. Asserting medical necessity, obtaining the patient’s informed financial consent in writing before a scheduled admission, and standing behind the claim under a practitioner’s registration remain with the people accountable for them. The tool gets the packet most of the way; the accountable person closes the rest.

What it takes to build

3–4 weeks · 4 phases

Reused from template~70%

Bespoke to this skin~30%

stack · Claude · private RAG · review UI

What it would cost

Fixed scope, fixed price, fixed dates.

Bite-sized first piece

One insurer, one procedure, low risk

Pilot build

Most builds land here

Embedded support

Scale across insurers on proof

Considering this for your billing or pre-auth team?

The honest place to start is a bite-sized first piece — one insurer, one procedure, low risk. Tell us where the pre-auth backlog hurts; we'll play it back, scope it, and show you what's possible.

Book a call How we work