Healthcare 8 weeks Perth, WA

AI Medicare Rules Engine for Healthcare Admin

AI assistant trained on Medicare rules to answer billing and compliance queries in real time. Reduced billing errors and eliminated hours of manual MBS code research.

AI DevelopmentCustom Software Development
85% Fewer billing errors
< 3s Answer time
4 hrs/week Admin time saved
Perth Based. Australia Wide.
18+ Years in Custom Software
Fixed-Price Delivery
Full Code Ownership
Client Context

Multi-location GP network — billing and administration

A GP network with 5 clinics and 18 practitioners. Their billing team handled Medicare claims across all locations — processing claims, resolving rejections, and answering practitioner questions about MBS item numbers, bulk billing rules, and patient eligibility.

Medicare billing rules are complex and change regularly. The team relied on the MBS online search tool, printed quick-reference guides (often outdated), and institutional knowledge held by the two most experienced billing staff. When those two were away, errors spiked.

The Challenge

What needed to change

Medicare rules are complex and constantly changing. Over 5,800 MBS item numbers, each with specific rules about who can bill them, when, in what combination, and at what rate. Changes happen quarterly and sometimes more frequently.

Knowledge was concentrated in two people. The senior billing staff knew the rules from years of experience. When they were on leave or busy, the rest of the team made errors — wrong item numbers, missed co-claiming restrictions, rejected claims.

Searching for answers was painfully slow. The MBS online tool is functional but not intuitive. Finding the answer to a specific billing question often required checking multiple item descriptions, cross-referencing explanatory notes, and reading through guidelines. Even experienced staff could spend 15–20 minutes per complex query.

The Solution

What we built

An AI assistant trained on the full MBS schedule, explanatory notes and Medicare guidelines. Staff ask billing questions in plain English and get accurate, sourced answers instantly.

Knowledge Base

Full MBS schedule, explanatory notes, Medicare guidelines and DVA billing rules ingested and indexed. Updated automatically when Medicare publishes changes.

Natural Language Search

Staff type questions in plain English — "Can I co-bill 23 and 36 for the same patient on the same day?" — and get a clear yes/no answer with the specific rule cited.

Source Citations

Every answer links to the specific MBS item, explanatory note or guideline section. Staff can verify and learn the underlying rules, not just get answers.

Auto-Updates

When Medicare publishes MBS updates, the knowledge base refreshes automatically. Staff are notified of changes relevant to their most commonly billed items.

Built with:
PythonOpenAI GPT-4LangChainPineconeFastAPIReactAWS
In Practice

How it works

1

Billing staff types a question

Natural language query in the search bar — no need to know MBS item numbers upfront. Questions can be about item eligibility, co-claiming rules, patient criteria or bulk billing entitlements.

2

AI retrieves relevant rules

The RAG engine searches the indexed MBS schedule and guidelines for the most relevant sections. Retrieves specific item descriptions, notes and cross-references.

3

Clear answer generated

GPT-4 generates a concise answer based on the retrieved Medicare rules. Constrained to source material — no hallucination.

4

Citations link to source documents

Answer includes links to the specific MBS items and guideline sections. Staff can click through to read the full rule text.

5

Feedback improves accuracy

Staff can flag incorrect or incomplete answers. Corrections inform retrieval tuning and prompt refinement.

Results

Measurable outcomes

85% Reduction in billing errors
< 3s Average query response time
4 hrs/week Admin time saved per clinic
95% Query accuracy rate
70% Fewer claim rejections
5 Clinics using the system

My team used to come to me with billing questions 30 times a day. Now they ask the AI first and it is right almost every time. When I am on leave, errors do not spike anymore. That is the real win.

Senior Billing Manager GP Network
Delivery

How we delivered it

1

Data Ingestion

2 weeks

Ingested the full MBS schedule (5,800+ items), explanatory notes, Medicare guidelines and DVA billing rules. Processed, chunked and embedded for semantic search.

2

RAG Engine Build

3 weeks

Built the retrieval and generation pipeline. Optimised chunking and retrieval for billing-specific query patterns. Implemented citation linking and confidence scoring.

3

Interface & Integration

2 weeks

Built the search interface — clean, fast, designed for busy billing staff. Added browser extension for quick access during claims processing.

4

Testing & Refinement

2 weeks

Tested with real billing scenarios across all 5 clinics. Validated accuracy against senior billing staff knowledge. Refined for edge cases and complex co-claiming queries.

Similar Project?

Want something similar for your business?

Tell us about your industry, your workflows, and what you want to achieve. We will scope it, quote it fixed-price, and build it.

Tell Us About Your Project

What industry are you in? What systems do you use? What is the biggest operational problem you want solved? We will come back with a plan and fixed-price quote.

Prefer a quick chat? Call 0425 531 127 – we're Perth-based and we answer the phone.