Closing panel: what we got wrong about agents
About this session
Four practitioners compare notes on the agent architectures they abandoned, the assumptions that did not survive production, and what they would build differently today.
Speakers
Noah Blackwood
Chief Technology Officer
Fathom Robotics
Noah is CTO at Fathom Robotics, where inference cost is a line item the board reads every month. He previously built pricing systems at a payments company.
Daniel Okonkwo
Director of AI
Meridian Health
Daniel builds clinical decision-support systems at Meridian Health, in a regulatory environment where 'we will fix it in the next deploy' is not an option.
Ava Nakamura
Staff ML Engineer
Lumen AI
Ava leads the evaluation platform at Lumen AI, where she is responsible for making sure model changes ship without quietly regressing customer-facing quality. She spent six years in search relevance before moving to LLM systems.