Five questions that keep the conversation focused on what you are actually buying.
Over the past few weeks, we have looked at vendor stability, model fit, and the true cost of AI. Those ideas become more useful when you turn them into the same set of questions for every vendor.
A consistent checklist makes a polished demo easier to evaluate—and competing proposals easier to compare.
- What job are we asking the AI to do?
Define the task, volume, input format, and measurable result before discussing technology.
- Where is this running in production today?
Ask for comparable customers at a similar scale, then speak with at least two of them.
- Which model powers it—and what happens when that changes?
Cover routing, version notifications, upgrade testing, and whether a version can be pinned.
- What will this cost at our real and peak volume?
Request the 12- and 36-month cost, the exact billable unit, and everything excluded from the quote.
- If we leave, what can we take with us?
Confirm export rights for your data, configurations, and workflows before signing.
Listen to how the vendor answers. You are not looking for perfection. You are looking for clear definitions, real examples, and a willingness to put important promises in writing.
One useful step this week: Send these five questions before the next demo and ask for written responses. A side-by-side comparison becomes much easier when every vendor is answering the same things.
Bonus tip: After each meeting, score every answer from 1 to 5 while the conversation is fresh. “We liked the team” is hard to compare later; a written score and one sentence of evidence are not.
Building an AI vendor shortlist?
Ian would be glad to compare notes on the questions that separate a promising demonstration from a workable long-term partnership.
Talk with Ian: Schedule a Conversation with Ian
