Updated 17 October 2026. People forget the real problem they wanted to solve. Instead of using AI to complement their judgement, they start depending on it to supply the judgement as well. That is the concern I keep returning to.
Keep the problem in the room
Start with what the person needs to accomplish. Describe the difficulty before selecting the model or the agent architecture. If the team cannot explain the benefit without naming the technology, the problem definition probably needs more work.
An illustrative team might want to reduce the time needed to find an approved policy answer. Search improvements, clearer documentation or a simple workflow change may help. An AI assistant is one possible approach, not the definition of the problem itself.
Use generated proposals as material to examine. Ask which assumptions they make, which evidence supports them and what alternatives they omit. A fluent answer can be useful without deserving automatic agreement. It has not earned a promotion to product strategy simply by arriving quickly.
Keep responsibility for the decision visible. The team chooses the objective, the acceptable trade-offs and the release conditions. Delegating a task does not remove the need to understand the consequences of the result.
| Question | Evidence or action |
|---|---|
| Problem | Human outcome and current difficulty |
| Options | AI and non-AI approaches |
| Decision | Evidence, trade-offs and accountable owner |
The practical next step
Return to the original problem after the demo. If the person’s work is not easier, safer or more effective in a way the team can demonstrate, the impressive technology has not yet become a useful product.
Keep the resulting decision, its owner and unresolved questions in a short decision record. For claims about agent behaviour, connect the decision to evaluation evidence rather than relying on a successful demonstration alone.
Further primary reading: NIST AI Risk Management Framework. The review map and product recommendations above are my analysis, not a certification or a legal assessment of a particular deployment.
