Three questions before you buy the AI
A short, honest filter for executives evaluating their first generative-AI deployment.
- 01If you cannot describe the decision the AI will make, you are buying a demo, not a system.
- 02The model is the cheap part. The integration, evaluation, and change management are the expensive parts.
- 03Pick a use case where the cost of a wrong answer is measurable and bounded.
Most AI initiatives do not fail because the model was wrong. They fail because the question was wrong. Before you sign a contract, sit with three questions long enough that you can answer each of them in a single sentence.
What decision is the AI making?
Not "what is it doing" — what decision. Drafting a reply is not a decision. Routing a ticket is. Approving an invoice is. Flagging a transaction as suspicious is. If you cannot name the decision, you are buying a productivity demo and calling it strategy.
What does a wrong answer cost?
Pick a domain where the answer is auditable and the downside is bounded. AI is at its most valuable when it is doing the eighty percent of work that is high-volume and low-stakes, freeing humans to handle the twenty percent that is rare and consequential.
Who owns it after the pilot?
A pilot with no operator is a science experiment. Before you start, name the human who will live with the system in production — who will tune the prompts, watch the metrics, take the page when it misbehaves. If that person does not exist yet, hire them first.
Drew leads Jonsen LLC — a Denver technology practice guiding law firms and growing businesses through AI, cybersecurity, and systems that compound over time.