Free account
Read this concept with a free account
Read this concept at no cost by signing in with Google.
no card needed
Core
Kill Criteria: Time-Boxed AI Bets With Pre-Registered Thresholds
AI projects carry feasibility uncertainty that traditional projects do not. Kill criteria are thresholds and dates, written before work begins, that decide whether a bet continues, changes or stops. This guide covers writing them, choosing thresholds, honouring them, communicating stops, and making stopping a normal part of the portfolio.
CONCEPT / LEADING AI TEAMS
RELATED CONCEPTS
PRACTISE THIS CONCEPT
Behavioral & LeadershipTell me about a time you decided to stop a project or feature. How did you make and communicate the call?→Behavioral & LeadershipAI projects are full of uncertainty. How do you handle ambiguity as a PM?→Behavioral & LeadershipHow do you build and communicate a product roadmap when you don't know what models will be able to do in six months?→Live Prototyping & AI-Native WorkflowYour team is arguing about which prompt and which model to use for a feature. How would you settle it in an afternoon?→AI Product SenseYou're writing the PRD for an AI feature that drafts replies to customer reviews. What goes in it that a normal PRD would not have?→Behavioral & LeadershipTell me about working with research scientists whose goals or metrics differed from the product team's. How did you align?→