AI Error Analysis For PM's & Engineers
AI fails differently than normal software. You need a new kind of practice to monitor errors and fix them. In this workshop, Gua breaks down how the best teams observe Human<>AI interaction, evaluate against expectations, and iterate continuously.
If your team is building AI experience, these questions probably come up often:
How do we evaluate outputs that require expert judgment?
How do we improve the system prompt without creating regressions elsewhere?
Where should engineering effort go first, and what do we not need to test yet?
With no user data, what is the smartest way to start?
How do we prove real progress to stakeholders instead of reporting vanity metrics?
That's exactly what this session will cover.
Speaker: Gua Tabidze, GenAI lecturer at commschool & Product Design Lead from Fortune 5 Tech to YC startups.