Rowan Tree Health
Consultation Note
Analysis (AI) for Health
Professionals
An AI documentation review layer that gives GPs a second set of eyes before they sign, built on 2,300+ tribunal cases and their own private Azure AI infrastructure.
This SaaS platform is designed to empower Australians to enhance their wealth and achieve their most important life goals. It seamlessly combines the convenience of online tools with access to a licensed wealth coach, providing expert guidance on everything from budgeting and to investing and retirement planning.
Focus Area:
- Medico-Legal Documentation Review
- AI Scribe Verification & Governance
- Regulatory & Case-Law-Informed Risk Scoring
- Clinical Documentation Compliance
- Azure Managed AI Knowledge Base
Our Involvement
- Business Analysis & Solution Design
- Software Architecture & AI Infrastructure
- .NET Microservices Backend Development
- Angular Frontend Development
- Azure AI Integration
- UI/UX Design
- Software Testing & QA
- Azure Cloud Hosting
By the Numbers
- Five-pillar Shield Score assessment model
- TGA Confirmed, outside medical device regulation
- AHPRA-aligned documentation framework
- 17 years of GP clinical experience behind the model
Platform Features
- Works alongside any AI scribe (Heidi, Dragon), no change to existing workflow.
- Five-pillar Shield Score documentation review.NET Microservices Backend Development
- Retrieval-augmented AI review engine
- Structured knowledge base, managed natively in Azure
- Multi-tenant admin portal for account management
- Speech-to-text consultation capture
- Plain-English review prompts, not just a score
- Client-owned AI Azure architecture
- SOAP note generation with change history ledger
- Full audit trail on every review
The Challenge
AI scribes have moved fast into general practice. Tools like Heidi, Dragon and Lyrebird now sit in thousands of consultations, turning conversations into clinical notes in seconds. What they don’t do, because it isn’t the job they were built for, is verify that note before it becomes part of the permanent clinical record. Clinical reasoning, safety-netting conversations and consent discussions can go missing from a transcript, and the GP who signs the note still carries the liability for what’s in it, and what isn’t.
Dr Andrew Wall had seen this problem from the other side of the desk. Across 17 years as a GP in the UK and Australia, including as a partner across a network of 100,000 patients, he handled three to four complaints a week and supported colleagues through tribunal processes. He came to Designpluz with a specific brief: build an AI system that reviews documentation the way an experienced GP partner would, grounded in what actually goes wrong in real complaints, not a generic chatbot bolted onto a scribe.
The Solution,
How we built
Rowan
Rowan needed to do something genuinely difficult for an LLM: reason reliably against a large, evolving body of regulatory guidance and tribunal precedent, and turn that reasoning into specific, human-readable feedback, in the middle of a GP’s working day, not after a formal audit. Here’s how we architected it.
The Knowledge Base: Native to Azure, Not to Us
The regulatory guidance, professional standards, and analysis drawn from 2,300+ Australian and UK tribunal cases that the Shield Score is built on isn’t managed through a tool we built, it’s managed directly inside the Azure Portal, inside Rowan Tree Health’s own tenant. Source documents sit in Azure Blob Storage, are processed through Azure AI Document Intelligence to extract structured text and data, and are indexed by Azure AI Search, all native Microsoft tooling, configured by us but populated and maintained by Rowan Tree Health directly. Their team can add and update the framework as new guidance or case outcomes emerge, with no custom interface and no Designpluz involvement required to do it.
A Retrieval-Augmented Review Engine
That knowledge base is what Azure AI Search retrieves against. When a consultation note is analysed, our .NET microservices pull the specific regulatory guidance and precedent relevant to that note, then pass it, alongside the note itself, to Azure OpenAI Service, the enterprise deployment of the same GPT model family behind ChatGPT, chosen so the note and everything retrieved for it stay inside a controlled Azure environment rather than a consumer API. The model assesses the note against the five-pillar Shield Score and returns specific, plain-English prompts a GP can act on in seconds, not a black-box score, and not a diagnosis.
Consultation Capture, However the Practice Works
While users are free to use their existing transcript/scribe app and paste outputs into our system, we know not all doctors are actually using a separate scribe app, so we built it ourselves – Integrating Azure AI Speech into the platform to convert consultation audio into structured text, Rowan’s review isn’t limited to practices on one particular scribe.
A Multi-Tenant Admin Portal, Built for the Business
Alongside the AI layer, we built Rowan Tree Health, a dedicated admin portal, Angular on the front end, our .NET Web API layer behind it, backed by SQL Server. This is where Rowan Tree Health runs the commercial side of the platform: provisioning new practice tenants, managing account status, paid, free trial, blocked, and tracking platform usage. It’s the operational backbone that lets Rowan Tree Health onboard new GP practices and manage a growing customer base without a Designpluz developer in the loop for day-to-day account changes.
Built for a Regulated Industry: Data Sovereignty by Design
The entire AI stack, Azure OpenAI Service, Azure AI Search, Azure AI Document Intelligence, Azure Cognitive Search and the knowledge base itself, runs inside Rowan Tree Health’s own Azure tenant, configured and maintained natively through the Azure Portal rather than through a system we’ve built and can see into. Designpluz designed, provisioned and maintained the infrastructure around it, the microservices, the admin portal, the DevOps pipeline, but the knowledge base itself is populated and updated by Rowan Tree Health, inside their own environment, with zero visibility from us. We operate the engine and the business layer around it; the client owns everything inside the AI itself. For a platform reviewing clinical documentation, that separation isn’t a nice-to-have, it’s the reason a doctor can trust it with his life’s work.
“We weren’t building a chatbot. We were building a system that could reason against 17 years of tribunal case knowledge and explain itself in plain English to a doctor about to sign his name.”
Project Team
The Result
Rowan is live and in controlled early access with its first cohort of Australian GPs. The platform is built on microservices specifically so Rowan Tree Health can extend into new settings, aged care, allied health, and beyond, as the cohort grows, without re-architecting the system underneath it; because the entire AI layer sits inside Rowan Tree Health’s own Azure tenant, Dr Wall’s team can keep building on their own clinical IP independently of Designpluz, while we continue to support and extend the platform underneath them.
Rowan Tree Health has progressed to the Queensland final of iAwards
in the Business & Industry category!
Technologies Stack
Frontend
Backend
Hosting & Tools
2,300+
Australian and UK
100,000
Network of patients
Security, Data Sovereignty &
Compliance
- Single-tenant architecture inside Rowan Tree Health’s own Azure environment, not Designpluz’s, and not shared
- Designpluz-managed infrastructure with no visibility into the client’s proprietary knowledge base or underlying data store
- Built around the TGA’s confirmation that Rowan sits outside medical device regulation, and an AHPRA-aligned documentation framework
- Every AI-assisted review carries a full audit trail back to the guidance and precedent behind it.
- Rowan is intentionally a documentation and clinical governance tool, not a diagnostic or advisory one, every decision stays with the treating clinician
Key Project
Considerations
The project required careful consideration around IP protection, technical architecture, user experience and scalability.
I have been delighted with the service offered by Designpluz. Over the whole process, from initial design to implementation, delivery and iteration they have been professional, responsive and highly helpful.
Developing a specialised AI product in a highly regulated environment is no small task, and I would thoroughly recommend Designpluz. Huge thanks in particular to Shaun, but also to Jag, Murugesan and the rest of the team. I strongly recommend this team.
