# The Tool Evaluation Filter

Five questions that decide whether a new AI tool deserves 30 minutes of your attention. Most tools fail by question two.

**Tool:** [TOOL NAME]  **Where you heard about it:** [SOURCE]  **Date:** [DATE]

Answer the questions in order. Stop at the first No.

## Question 1: Does it touch a task on my leverage map?

Open your Stage 1 leverage map. Find the task this tool serves.

- Task on the map: [TASK or NONE]
- **Answer: [Y/N]**

If the task isn't on your map, the tool is not for you, no matter how good the demo looks. No mapped task, no test. This question alone eliminates most of what crosses your feed.

## Question 2: Is it clearly better than what I use now?

"Clearly better" means half the time or twice the quality on your work — or a capability you don't have at all. A 10 percent improvement never pays back the cost of switching, re-learning, and re-documenting a workflow.

- What I use for this task today: [CURRENT TOOL OR WORKFLOW]
- The claimed improvement, specifically: [CLAIM]
- **Answer: [Y/N]**

## Question 3: Can I test it on my real work in 30 minutes?

No sales call. No onboarding session. No "book a demo." If you can't sign up and run one of your own real tasks through it today, the answer is no — for now. Re-check next quarter.

- **Answer: [Y/N]**

## Question 4: Can I live with where my data goes?

Find the data policy. Where do your inputs go, who can see them, are they used for training? Attorneys: client confidentiality applies. Accountants: client financials apply. Anyone: if you can't find a clear answer in five minutes, that is your answer.

- Data policy says: [SUMMARY]
- **Answer: [Y/N]**

## Question 5: What does it cost at my real usage in month three?

Not the free tier. Free tiers shrink, hit limits, or disappear. Price the plan you'd need at your normal volume, and ask whether the time saved covers it.

- Real monthly cost: [$N]  Hours/month it would save: [N]
- **Answer: [Y/N]**

## Scoring

- **Five Yes answers:** run the 30-minute test below this month.
- **Four:** note the failing question and re-check next quarter.
- **Three or fewer:** ignore the tool. Close the tab.

The default answer is no. Your current setup works. A new tool has to earn its way in.

## The 30-minute test

1. Pick one real task from your week. Not a sample — real work.
2. Run it twice: once with your current workflow, once with the new tool.
3. Compare output quality and total time, including setup.
4. Decision: [ADOPT / DROP]. If adopt, document it as a workflow and add it to your library. If drop, write one line about why, so you don't re-test it in three months.

Limit: test at most one tool per month. Two tests a quarter is a normal pace. Zero is fine.

Part of the ADAPT framework — vetslearn.com/framework