Faster referral reviews. Clinicians stay in control.
MedAI extracts referral details and suggests a priority using your rules. A clinician checks the source evidence and decides.
The problem: referral review still relies on copying, retyping and chasing.
MedAI is designed for practices where referral PDFs arrive by email and staff must manually find details, assess urgency, prepare a note and enter information into practice software.
We begin with one defined workflow, not a practice-wide AI rollout.
Referral volume creates a regular clinical or administrative backlog.
Reception retypes information already present in the PDF.
Clinicians repeat the same review steps for each letter.
The practice wants automation without replacing existing systems.
How MedAI helps: one queue, a structured draft and clinician sign-off.
MedAI extracts the details, drafts a summary and suggests a priority using rules agreed with your practice.
- Patient and referrer details
- A structured clinical summary
- A suggested priority based on your rules
- The source wording behind the suggestion
- A copy-ready note and actions for reception
The clinician can inspect the source evidence and makes the decision.
Automatic entry into practice software is only used where an official connection exists or the vendor gives written permission. Otherwise, staff receive a checked, copy-ready result.
From referral inbox to clinician-reviewed result.
The practice remains the system of record. One AI processing step occurs through a Sydney-hosted service.
The referral arrives
The PDF moves from the practice inbox to the dedicated MedAI computer. The email itself is not copied.
The PDF is processed in Sydney
The referral is sent once to a Sydney-hosted AI service. MedAI extracts the information and applies your agreed priority rules.
A clinician reviews the evidence
The clinician sees the draft, suggested priority, reasoning and supporting words. The clinician confirms or changes the result.
Reception receives the result
Reception receives the confirmed details, note and next steps. Automatic entry requires an official connection or written vendor permission.
See what leaves the practice, what stays and who decides.
The referral PDF is processed once in Sydney. The source documents, results and activity records are stored on the dedicated computer at your practice.
One Sydney processing step
The referral PDF is sent to a Sydney-hosted service for the AI read. The data flow is documented before go-live.
Records stay at the practice
Referral documents, structured results and activity records are stored on the dedicated MedAI computer.
No consumer chatbot accounts
Patient information is not pasted into consumer ChatGPT, Claude, Gemini or similar accounts.
The clinician decides
MedAI can draft and suggest. It does not diagnose, make the final priority decision or act without review.
The source remains visible
The clinician can inspect the wording used to support each suggestion before confirming the result.
Connections require permission
MedAI uses an official connection where one exists. Otherwise, automatic entry requires written vendor permission.
Before a pilot begins, the practice receives a plain-English record of what is processed, where it is stored, who can access it, which services it depends on and what happens if a service is unavailable.
A working referral system, with time savings still to be measured.
The first MedAI referral workflow is operating in a specialist practice. The pilot will measure review time, corrections and workflow exceptions.
Pilot planning estimate, not a measured result
Estimated clinician review time for 30 referrals
Potential time saved: about 2½ hours
Planning estimate for the first pilot. Actual time will depend on referral length, complexity, practice rules and workflow.
“I wanted the referral inbox handled properly before anything else. Every recommendation shows me its reasoning and the exact lines in the letter it came from, and nothing goes overseas. That is the only way I would put AI near patient data.
The current data flow
The referral PDF is processed once by a Sydney-hosted AI service. The source document, result and activity record are stored on the dedicated computer at the practice.
This is an early pilot at Progenics, where Dr Verma is both the pilot clinician and a MedAI clinical co-founder.
Start with a demo, a focused audit or a scoped build.
You do not need to commit to a practice-wide project. Start with the referral workflow or one defined administrative process.
See the product
Referral workflow demo
See the queue, source evidence, clinician review and reception output using synthetic patient information. We also explain the Sydney processing step and dedicated practice computer.
Request a referral demoOne-day review
Practice workflow and AI safety audit
We review repetitive work, identify where automation may save time and document the clinical, privacy and vendor constraints.
- A map of the current workflow
- Suitable tasks and tasks to avoid
- Data-handling and safety constraints
- A recommended first step
Fixed scope, usually 2 to 8 weeks
Build and installation
We agree the controls, connections and acceptance checks, then install the workflow on a dedicated computer and train the team.
You receive a fixed scope and price first. External service charges and support are shown separately.
After go-live
Support and maintenance
We provide updates and respond to changes in connected software. Remote access is restricted and recorded.
External service and software dependencies are documented before go-live.
Questions a practice should ask before using AI with referrals.
Does patient information leave the practice?
Yes, for one defined step. The referral PDF is processed by a Sydney-hosted AI service. The document, result and activity record are stored on the dedicated computer at the practice.
Does MedAI make clinical decisions?
No. MedAI prepares a draft and may suggest a priority using practice rules. A clinician reviews the source evidence and makes the decision.
Will it enter information into our practice software?
Only where an official connection exists or the software vendor provides written permission. Without approval, MedAI provides a checked, copy-ready result.
Do we need to replace our current software?
No. MedAI sits alongside the systems the practice already uses. The workflow and any approved connection are agreed before implementation.
What will it cost?
After the initial discussion, MedAI provides the relevant audit fee or a fixed build scope and price. Support and third-party service charges are listed separately before you agree.
How disruptive is installation?
The system runs on a dedicated computer at the practice. The implementation plan covers staff training, testing, go-live and what to do if the system is unavailable.
Are we dependent on MedAI or another vendor?
The workflow depends on the Sydney-hosted AI service and may depend on approved practice software connections. These dependencies, support arrangements and available outputs are documented before go-live.
What privacy information will we receive?
Before a pilot, the practice receives a written summary covering processing location, storage, authorised access, external services and activity records. Retention, deletion, model training and incident handling are included once contractually confirmed.
Clinical oversight and experienced software delivery.
MedAI is at the pilot stage. Time-saving figures on this page are planning estimates until they have been measured in practice.
MedAI brings together a specialist doctor and an experienced software engineer. Matt Platts has fifteen years of experience building and operating software. Dr Kunal Verma is a clinical geneticist, cardiologist and founder of Progenics, the pilot practice for the first MedAI referral workflow.
The aim is practical: reduce repetitive administrative work while keeping patient data handling explicit and clinical decisions with clinicians.
See the referral workflow before discussing a build.
We will show you the system using synthetic patient information, explain how the referral moves and discuss whether it fits your practice and software.
A demo request does not commit you to an audit or implementation. contact@med-ai.com.au. Based in Melbourne, working across Australia. We reply within two business days.