PE-backed healthcare group. 50 practices. $111M revenue. One AI partner.
Four interconnected AI systems deployed across operations, compliance, finance, and reporting — eliminating manual work and giving leadership instant access to their own data.
The organisation and the problem
A private equity-backed healthcare service organisation operating 50 practices across Australia and New Zealand — spanning WA, VIC, NSW, QLD, SA, NT and NZ.
Under PE ownership, the business was acquiring multiple practices per quarter. Each acquisition added a new legal entity, a new Xero account, a new practice management system instance, and new staff — all needing to report into head office. The result was a growing organisation with no unified view of itself.
The group had invested years building a data warehouse. Their finance team had spent significant effort pulling data from Xero, their procurement platform, and their practice management software into one place. But having the data wasn't the problem. Getting answers from it was.
The entire analytics team was dedicated to extracting reports. Every question — how is Practice A tracking on conversion? What's procurement spend vs production at Practice B? — required multiple full-time resources to pull data, build a spreadsheet, and deliver a report. At 50 practices, over 300,000 patient records, $111M in production and 130 KPIs to track, manual reporting was no longer viable.
What Stack10 built
Four interconnected solutions — spanning Stack Discover and Stack Build — that fundamentally changed how the organisation operates.
The Intelligence Layer
MCP Agent · Plain-English Data Queries
The problem: Leadership couldn't get answers without waiting for someone to build a report.
What we built: An MCP (Model Context Protocol) server that sits directly on top of their data warehouse. For each database view, we created tools with pre-configured SQL queries that allow an AI agent to query the database in real time. Any leader can now ask a question in plain English and get an answer in seconds.
Use cases it unlocked
Procurement vs production reporting
Previously required a dedicated analyst to compile across multiple systems. Now: ask the agent, get an instant report — with drill-down into the exact line items driving any variance.
Doctor transition management
Finding every affected patient, reading notes, and preparing a rebook list would traditionally take a week. The agent returned the full list in 2 minutes.
New patient conversion tracking
Which practices were converting consultations into treatment plans — and which weren't. Used to take days of cross-referencing. Now instant, by practice, by region, or across the group.
The Compliance Agent
WHS Automation · Autonomous Document Management
The problem: Every chemical agent in every practice needs a current Safety Data Sheet. Compiling the chemical register quarterly was manual, error-prone, and scaled linearly with every acquisition.
What we built: An autonomous AI compliance agent that integrates with procurement order history, identifies hazardous materials, and autonomously searches, validates, and centralises safety documentation from manufacturer websites. When it can't find a document with high certainty, it flags it rather than guessing — the WHS team shifted from data entry to exception management.
- Manual compilation every quarter
- Scaled linearly with every acquisition
- Error-prone, always behind
- WHS team = full-time data entry
- Autonomous document retrieval
- Scales instantly to new acquisitions
- 94% success rate, always current
- WHS team = exception management only
The Reconciliation Agent
Xero Integration · AI Financial Classification
The problem: Across 16+ separate Xero entities, high-value consumable spend was buried in a generic account code. Group leadership couldn't answer basic questions: How much are we spending per supplier per practice? Are practices over-ordering?
What we built: A full-stack reconciliation application that connects to every Xero entity, pulls invoice line items, uses AI classification to identify high-value consumable purchases, and writes reclassification journal entries back to Xero — with human sign-off at every step. For the first time, the group has visibility into their largest consumable cost category, automatically, across all entities.
Group-wide visibility into their largest consumable cost category for the first time — creating direct leverage for supplier renegotiation across all 50 entities.
FastAPI backend · Next.js + Shadcn/ui frontend · Multi-step AI pipeline (Haiku + Sonnet) · Supplier learning database · Full audit trail
Procurement Compliance Workflow
Automated Reporting · Regional Manager Workflow
The problem: Practice managers needed to act on their numbers, but Regional Managers were buried managing multiple practices — too many numbers to stay on top of and communicate downward.
What we built: An automated procurement reporting workflow. The system generates a spend report and sends it to the Regional Manager. The RM reviews in their email, provides feedback via text, and approves. The report — with the RM's recommendations — is then sent directly to the practice manager with actionable items attached. Instead of practices pulling their own reports, the data comes to them with specific actions. This workflow has become one of the primary tools driving procurement compliance across the group.
Procurement report compilation eliminated entirely. Regional Managers spend minutes reviewing, not hours compiling — and practice managers receive context and actions, not raw numbers.
Results across all four systems
→ 2 min Doctor transition workflows
What’s next
Based on the success of these projects, Stack10 is now working with the group’s in-house development team to:
Ready to see what this looks like for your organisation?
Book a 30-minute Discovery Call. We’ll tell you whether Stack10 is the right fit — and if not, we’ll tell you honestly who is.