As healthcare organizations embrace large language models (LLMs), the need for robust governance and systematic evaluation becomes mission critical. This session describes Brook AI’s journey to becoming a world-class AI governance organization – highlighting the processes, architectures, and tools that made it possible.
Brook AI delivers remote patient care, blending clinical teams with AI for continuous, always-on patient support. To evolve from early trials to an enterprise-ready foundation for generative AI deployment, Brook partnered with Pacific AI to implement a comprehensive governance framework aligned with NIST AI RMF, ISO 42001, CHAI, state laws, and federal regulations. This included policy rollouts, stakeholder training, automated governance tooling, and a full architecture and security review of Brook’s multi-agent LLM-based platform.
Brook also embedded LLM-based agent testing directly into its development lifecycle and CI/CD pipelines. Together, the teams created a rigorous benchmark suite covering key aspects of responsible AI in healthcare: fairness, accountability, transparency, explainability, robustness, safety, privacy, and accessibility. The suite was reviewed by healthcare professionals and applied to real-world scenarios mirroring clinician–patient conversations. It also enables comparative evaluations of Brook’s domain-specific agentic AI configurations against general-purpose LLMs such as ChatGPT, and incorporates automated daily evaluations with LLM-as-a-judge plus continuous monitoring for drift and safety incidents in production.
Attendees will leave with a clear blueprint for integrating AI governance into clinical GenAI applications – demonstrating that safety, compliance, and innovation can co-exist at scale.
