# The AI pilot scorecard

Superpowers by Waymaker — practical worksheet

Record actual quality, effort and failures across a few representative cases.

Use permitted sample data. Do not include credentials.

## Pilot identity

Task, owner, model or tool, version if known, settings and date.

Your notes:

---

## Test cases

For each case: input reference, expected result, actual result, accuracy, completeness and corrections. Include an awkward or incomplete input.

Your notes:

---

## Effort

Record baseline minutes, assisted-work minutes, review/rework minutes, usage charges and setup time.

Your notes:

---

## Failure behavior

What happened with missing context, tool errors, repeated events or an instruction hidden in source material?

Your notes:

---

## Decision and limitations

Keep, revise or stop? Which cases were not tested? What will trigger a new review?

Your notes:

---