A useful report connects a learning goal to an observed task and the assistance required. “Completed an advanced chatbot” is less decision-useful than a description of what the learner can explain and change.
Our proposed one-page template:
| Field | Illustrative entry—not an actual learner record |
|---|---|
| Goal | Explain why unsupported questions should not receive invented answers |
| Work attempted | Tested a fictional-club FAQ prototype |
| Evidence | Added one unsupported question and identified the incorrect response |
| Assistance | Needed a conceptual hint to locate the relevant setting |
| Contributions | Mentor supplied interface scaffolding; AI drafted a function; learner revised tests |
| Limitation | Cannot yet explain the full model connection |
| Next independent step | Add and justify two new test cases |
| Learner voice | Wants to improve the interface before extending the topic |
Request one small artifact or an in-session demonstration, not a growing archive of recordings.
Commercial reporting promises should be checked against this standard. CodaKid, for example, advertises post-session reports; that establishes the promised frequency, not the depth or usefulness of a particular report. (codakid.com)
Ask the mentor to review the report with the learner before sending it. Let the child correct the contribution description or explain where they still feel uncertain.