Healthcare has some of the highest ambition and the tightest funding for AI of any sector. In this episode Lachlan MacBean walks through how his team built a ten person AI capability without a dedicated AI budget, the partnering model that made it possible, and why they made the call to shut down an early project despite strong results.

Key discussion points

• Joining three months before COVID and the operational chaos that followed, and how the crisis proved the value of the analytics function
• The partnering strategy: using clinicians' grant funding and dual academic clinical roles to fund AI work that wouldn't otherwise get a business case
• Structuring the team on a part revenue target model, operating internally like a consulting practice within the hospital
• A handful of models now running in production, built on a mix of structured data and clinical notes
• Why an early AI project got strong results but was shut down: a solution looking for a problem, and a pricing model that made it commercially unsustainable at the time
• The "shark tank" process now used to pressure test new ideas before committing delivery resources
• Keeping teams engaged through all hands meetings, a cross organisation community of practice, and tying work back to patient impact