Most people pick an AI tool the same way: they search, open five tabs, sign up for three free trials and give up on two of them before the week ends. Here is the checklist we use when we review tools, written so you can apply it in about fifteen minutes per tool.

1. Write the task in one sentence first

"I need to turn a 40 minute meeting recording into notes with action items" is a task. "I need an AI for meetings" is a category. The sentence tells you what to test: input format, length limits, output format and who will read the result.

If you cannot write the sentence, you are not ready to compare tools yet.

2. Read the pricing page before the homepage

Homepages describe the best case. Pricing pages describe the limits. Look for:

  • What the free tier actually includes. Free sign-up credits that run out are not the same as a permanent free plan.
  • How usage is counted. Credits, minutes, words, messages or seats. A cheap plan with a tight monthly cap can cost more than a pricier plan for your real volume.
  • Whether the price you need is hidden. "Contact sales" for the plan with the feature you want is useful information.

We sort every tool into one of four labels, and the definitions are worth borrowing:

Label Meaning
Free Usable with no payment and no expiry
Freemium A permanent free tier or free credits, with paid plans for more
Free Trial Full or partial access for a limited time, then payment
Paid No way to use it without paying

3. Check where your data goes

For anything with documents, recordings or customer data, find three answers in the privacy policy or help center:

  • Is your input used to train models, and can you opt out?
  • Where is data stored, and for how long?
  • Can you delete your workspace and files yourself?

If you cannot find the answers in ten minutes, write to support. How fast and how clearly they answer is part of the review.

4. Test with your own input, not the demo

Demos use inputs the tool handles well. Use a real file from your work, including the messy one: the recording with crosstalk, the PDF with tables, the spreadsheet with merged cells. Run the same input through two or three candidates and compare the output side by side.

Keep a short note for each run: time taken, what was wrong, what you had to fix by hand.

5. Look for the exit before you commit

Before you move a team or a workflow onto a tool, check:

  • Can you export your data in a standard format (CSV, Markdown, DOCX, JSON)?
  • Does it connect to what you already use (Zapier, Make, an API, native integrations)?
  • When was the last real product update? A changelog or release notes page that has been quiet for a year is a signal.

6. Keep a shortlist, not a favorite

AI tools change pricing and features often. A tool that was the right pick six months ago may now have a stricter free tier or a better competitor. Keep two alternatives noted for every tool you depend on, and check pricing again when your plan renews.

A shortcut for the research part

Steps 2 and 5 are the slow part. We built Artificial Intelligence Tools List (AITL) to do that research once: each listing states the pricing label, key features, limits, integrations and main alternatives, and you can compare up to three tools side by side. Prices and major updates are checked weekly. It is free to use.

The checklist above still applies. A directory can tell you what a tool claims and costs; only your own input tells you whether it works for your task.

What is the one check you always do before paying for an AI tool? I would like to add it to the list.