STARTER CHECKLIST / 01

Your first AI project
Start small. Build for real.

Choose one small task you repeat, then use this checklist to plan an experiment. Print this page or choose “Save as PDF” in your browser’s print menu.

1. Choose one specific task

   For example, turn sample meeting notes into an action list, rather than starting with “use AI across the whole company.”

2. Record your baseline and success criteria

   How long does the task take today? What must the result include? What errors are acceptable? Measure both time and quality.

3. Prepare appropriate examples

   Start with synthetic data or data you have permission to use. Remove personal and confidential information before testing external services.

4. Build the simplest workflow

   Input → instruction → output → human review. Start with one manual run at a time before adding automation.

5. Test several cases and record results

   Test typical examples, incomplete inputs, and ambiguous tasks. Record the tool, model version, date, prompt, and results so you can repeat the experiment.

6. Count the full cost

   Include tool costs, data preparation, review and correction time, and ongoing maintenance. Compare with the original process using the same task.

7. Decide based on what you find

   Continue, adjust the task, or stop: each can be a useful outcome. Before using it in production, assign an owner, define review points, and plan how to return to the original process.