Building AI That Saves Lives: Agentic Systems for Cancer Care
Guardrails, ontologies, trusted knowledge, and hybrid AI architecture for high-stakes healthcare applications
Session Overview
Cancer is becoming more prevalent as people live longer and healthier lives, yet treatment information remains complex, fragmented, and rapidly evolving. Patients, caregivers, and clinicians are often forced to navigate a vast body of medical knowledge that is difficult to track in full.
In this technical session, Jack Chua will explain how Valent Health built a trusted knowledge ontology designed to prevent AI systems from making unsupported claims. He will also show how the company developed an agentic pipeline that works like a diligent researcher and caregiver, while preserving traceability, context, and safety.
The session will compare Valent Health's controlled clinical AI system with raw, unconstrained output from cloud-based LLMs. It will also examine how combining cloud and local models can help organizations achieve better cost efficiency, privacy, and inference quality for specialized healthcare use cases.
Key Takeaways
- How to build AI guardrails for high-stakes healthcare use cases.
- Why trusted knowledge ontologies are essential for traceable and evidence-backed AI responses.
- How agentic systems can support patients, caregivers, and clinicians as a diligent research and care assistant.
- The practical differences between a controlled clinical AI system and unconstrained general-purpose LLM output.
- How hybrid cloud and local LLM architecture can improve cost, privacy, latency, and inference quality.
Building a Winning Startup in the Age of AI
Drawing on his experience as an early-stage founder, Jack will share practical observations on three factors that are critical for AI founders seeking to build momentum and raise venture capital within the first six to twelve months of company formation.
- Building investor confidence before a fully developed product exists
- Answering questions from customers and distribution channels
- Developing a deeply personal connection to the problem and the ability to sell the solution
This discussion is designed for aspiring founders and current early-stage startup leaders preparing to take the next step.
Speaker
Jack Chua is the CEO and Head of Science at Valent Health, an AI companion platform for cancer patients, caregivers, and clinicians.
Previously, he served as Director of Data Science at Meta and Senior Director of AI at Expedia Group.
He holds an MS and completed doctoral-level study in Computational Science at the Georgia Institute of Technology, as well as bachelor's degrees in Mathematics, Economics, and Statistics from the University of Chicago.