To avoid stitching together multiple vendors, Fetch built its entire sales, service, and data ecosystem on a single Google Cloud stack. This architecture allows the team to deploy custom logic that directly serves the customer, rather than managing a complex web of third-party tools.

At the heart of the infrastructure is Gemini in Vertex AI, which acts as the engine for analyzing messy clinical histories. Instead of a human agent reading 100 pages of notes, the AI ingests unstructured data—including handwritten vet records and inconsistent terminology—and organizes it chronologically. It identifies potential pre-existing conditions and presents a clear summary to a vet-trained agent in a custom-built portal. This human-in-the-loop workflow allows agents to validate assessments in minutes rather than hours, ensuring high accuracy without burnout.

Delivering total transparency upfront was once impossible due to the sheer volume of data, but the single-stack approach has made it viable. Fetch now achieves more than 95% accuracy in coding veterinary conditions, a level of precision that allows the company to give owners certainty without the substantial overhead. This means capital isn't wasted on maintaining servers—it is reinvested directly into keeping premiums low and coverage clear.

To better support Australian pet parents, Fetch is now pivoting to become an all-in-one health partner. By analyzing clinical insights collected during the upfront check in BigQuery, Fetch aims to deliver proactive advice that helps pets live longer. For example, by tracking when vaccinations are due, the app could proactively prompt an owner to schedule a clinic visit and even facilitate the booking directly with their vet. This is part of a larger plan to shift the focus from transactional insurance coverage to a long-term health journey.