Use AI as a thinking partner.
AI can help you make a problem easier to reason about. Give it a role in examining the question, the assumptions, and the evidence before asking it to recommend an answer.
Turn a vague decision into a useful model, competing explanations, and a concrete next test.
Find the question underneath the request.
“Which CRM should we buy?” might really mean “Why do leads fall through the cracks?” A purchase decision can be premature if the actual failure is unclear ownership. Describe what is happening and ask the model to propose alternative problem statements.
You remain the judge of which framing fits reality. Confirm it against the people and records involved.
Separate facts, assumptions, and preferences.
Write down what you know, what you are guessing, and what you value. These categories do different jobs. A fact can constrain a decision; an assumption needs testing; a preference can guide a tradeoff.
Ask for the strongest competing explanation. Several models agreeing is not independent proof, especially when they received the same incomplete story. Look for evidence from the operating environment.
Make the alternatives concrete.
Have AI describe the simplest plausible options, their prerequisites, and their likely failure modes. Ask what each option would require from your team. This keeps a convincing recommendation connected to actual work.
Avoid repeatedly asking “Is this a good idea?” That invites endorsement. Ask what would have to be true for the idea to work and what observation would show it is failing.
Choose the next question that reduces uncertainty.
The useful output may be an experiment, a short interview, or a check of existing records. Prefer a next step that could change your decision over another round of broad brainstorming.
For a lead follow-up problem, inspect recent missed inquiries. Did the messages arrive late, lack an owner, or go unanswered after assignment? The answer determines whether your next move is a process change, better visibility, or automation.
Example: investigate missed follow-ups
This prompt produces a decision brief and a test plan. It does not substitute model confidence for business evidence.
Replace the bracketed fields before using this with AI.
Try it in your business.
- Name a decision you are currently leaning toward.
- List the assumption that would most weaken it if false.
- Use the decision-partner template, then check one claim outside the conversation.