Skip to content
PMCert

Tools

Build an AI feature

Frame the problem, check whether AI is the right path, test cheaply and do the cost maths before scaling.

The flow

  1. Frame with how might weFrom the problem to the questions that open ideation, without jumping to the solution too early.How might we generator
  2. Assess feasibilityAnswer eight questions and see whether the idea is ready to go, needs a prototype first or should not move ahead yet.AI feature feasibility checklist
  3. Plan the prototype testDecide before the test what to learn, with whom, how to measure and what to do with each result.Prototype test plan
  4. Work out inference costThe volume, tokens and prices you enter become cost per request, per month and per active user.AI inference cost calculator
  5. Organise opportunities and solutionsConnect the outcome the team is pursuing to customer needs, to solutions and to the tests that will decide between them.Opportunity solution tree