HOW IT WORKS

Two agents. One knowledge base.
Zero guesses.

Both tools draw from the same set of NDIS policy documents — monitored daily, version-controlled, evidence-gated.

PLATFORM GUARANTEES

Every answer cites a source

No answer is generated without a retrievable policy section to back it.

Policy monitored daily

16 documents checked for changes. Old versions retired before new ones go live.

Refuses when it doesn't know

If evidence confidence falls below threshold, the system says so — it doesn't guess.

Every query is logged

Every question, answer, and refusal — stored in full. Nothing is off the record.

AGENT 01 — POLICY QUERY

Ask a question. Get a cited answer.

Open →
1

You ask

Type any NDIS policy question in plain language. No special syntax needed.

2

We search 16 documents

Matched against the full NDIS corpus using keyword and semantic search simultaneously.

3

Evidence is checked

Confidence must clear a threshold before the LLM is called. If it doesn't, the system asks a clarifying question — then refuses if still insufficient.

4

Cited answer

The response includes the exact document section it came from. No citation means no answer.

AGENT 02 — CLAIM CHECKER

Upload. Cross-check. Submit with confidence.

Open →
1

Upload your documents

Invoice, shift notes, service agreement — any combination.

2

Fields are extracted

Every line item, rate, date, and participant detail pulled from each file.

3

Cross-checked against policy

Line items validated against the Support Catalogue. Rates compared to Pricing Arrangements. Fields cross-referenced across all files.

4

Compliance report

Pass, flag, or fail — with a reason for every finding.

THE KNOWLEDGE BASE

Both agents draw from the same corpus: 16 NDIS policy documents ingested from official sources, split by section, and re-indexed whenever a source changes. The Price Guide and Support Catalogue update quarterly — the corpus reflects that automatically.

16 documentsDaily monitoringVersion-awareAWS Sydney