Prompt used
The brief, constraints, model/tool, and important follow-up prompts.
Back to LibraryView case
Baseline ReportPublished draft
Chess Scout V1 baseline
AI helped with structure and UI direction, but product judgement needed human correction.
Linked case
Chess Scout
A data-heavy chess scouting product used to test whether AI models can improve UX, insight quality, report clarity, and code reliability.
Content structure
Full write-ups use this structure as tests are completed, with prompts, screenshots, model outputs, code notes, caveats, and final scorecards where relevant.
Starting point
The baseline product state, known issues, screenshots, and repo constraints.
Before / after
Screenshots, code diffs, UI notes, and behavioural changes from the model output.
What worked
The parts worth reusing: structure, implementation choices, UX improvements, or analysis.
What failed
Bugs, weak assumptions, overreach, regressions, or outputs that needed correction.
Human rescue needed
The manual judgement, debugging, edits, and product decisions needed to make it usable.
Verdict
A practical call on whether this model/tool should be used again for similar work.