The AI tools market has never been louder, and most of the noise is indistinguishable. This is the checklist we use with clients to evaluate any AI purchase, from a $30-a-month utility to a six-figure platform. Seven questions. A vendor who cannot answer them crisply is telling you something.
1. Which hours disappear?
Name the tasks, the people, and the hours per month. If the value is described as insights or empowerment rather than work eliminated or revenue gained, the ROI is a hope, not a number.
2. What is the baseline?
You cannot prove improvement without measuring the current state first. Insist on a before number: hours, error rate, cycle time. No baseline, no purchase.
3. How does it get our data?
The demo runs on clean sample data. Your business runs on your data. Ask exactly how the tool connects to your systems, what integration costs, and who maintains it when an API changes.
4. Where do exceptions go?
Every automated process produces exceptions. Serious tools have exception queues and human review flows. Toys pretend exceptions do not exist.
5. What does month 13 cost?
Intro pricing, usage tiers, per-seat creep. Model the realistic second-year bill at your projected volume before signing anything.
6. Can we leave?
Data export, contract terms, switching cost. A tool you cannot exit is a landlord, not a vendor.
7. Who owns it internally?
Tools without an internal owner decay into shelfware within two quarters. Name the person before you buy, not after.
Want a second opinion on a specific tool or stack? Our AI consulting practice does exactly this evaluation, vendor-neutral. Ask us anything.
