Airbnb CTO Ahmad Al-Dahle tells Sam Charrington how Project Everest captures institutional knowledge from employee workflows to help AI agents work across unfamiliar parts of the company.
Al-Dahle describes evaluating agent performanceEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria. and warns that stale or incorrect context can mislead models. He discusses gradual disclosure of relevant informationContext engineering designs how relevant instructions, evidence, state and tool information are assembled for a model's task. as one possible approach rather than a solved architecture.
They explore measuring engineering productivity, the limits of AI support in disputes between guests and hosts, and tradeoffs in search experiences and specialized models. Claims about Airbnb’s internal systems are attributed to Al-Dahle.
Watch on YouTube
