Priors, rethought

Things teams believe about AI, crossed out and rewritten starting from the outcome. Each one has a short answer you can read in about a minute.

• We need an AI strategy firstDo we need an AI strategy?
• Pick the AI tool, then find the use caseShould we pick an AI tool and then find use cases for it?
• Every company needs an AI chatbotDoes every company need an AI chatbot?
• More AI agents means more outputDo more AI agents mean more output?
• AI work should be billed by the hourShould AI implementation work be billed by the hour?
• High AI adoption proves ROIDo AI adoption and usage numbers prove ROI?
• Prompt engineering is the key AI skillIs prompt engineering the most important AI skill?
• MCP is complicated plumbingWhat is MCP, and is it complicated?
• A great AI demo means it's readyIf an AI demo works, is it ready for production?
• AI should summarize every meetingShould we use AI to summarize every meeting?
• AI benchmarks tell you which model is bestDo AI benchmarks tell you which model is best for your work?
• AI agents will replace junior staffWill AI agents replace junior staff?
• A human should approve every AI stepShould a human approve every step an AI agent takes?