01 / CONTEXT
The problem to solve
A collector wants to recognize physical cards and understand their market context without manually juggling catalogues, variants and price sources.
02 / ENGINEERING
From need to system
- Use image-assisted recognition with a documented OCR fallback.
- Resolve card identity across edition, language, condition and variation.
- Integrate market data from multiple sources and explain pricing differences.
- Support mobile-first scanning, collection management and historical snapshots.
03 / DESIGN DECISION
A meaningful choice
Treat recognition uncertainty and entity resolution as explicit product interactions rather than silently assigning an identity.
04 / EVIDENCE AND LIMITATIONS
What is actually documented
The private project documentation describes these workflows. Recognition accuracy and price reliability still require independent evaluation.
05 / NEXT VALIDATIONS
How I would strengthen the proof
- Benchmark recognition on a varied photo sample.
- Measure catalogue and price coverage.
- Improve correction UX and error analysis.
These are planned validation steps, not already measured results.