brianletort.ai
Qwen3.8 vs GPT-5.2

Practical meeting-notes extraction browser tool

A practical local browser tool comparison: paste meeting notes, extract decisions, owners, and risks. Qwen is preserved from the breakout gallery; GPT-5.2 is the matched single-call counterpart and remains visibly truncated.

Showcase, asymmetric evidenceeditorial showcase with known asymmetry

Publication-safe copies generated from preserved showcase artifacts. Source research paths, local absolute paths, endpoints, and secrets are intentionally omitted.

Prompt and source disclosure
Build a single self-contained HTML page.
Requirements:
- Textarea for messy meeting notes
- Button "Extract"
- On click (inline JS, no CDN): parse notes into three sections: DECISIONS, OWNERS, RISKS
- Heuristic only (keyword lines / bullets) — no network
- Dark clean UI, no purple gradients
- Include a "Load sample" button with embedded sample notes about an AI intake forum
Output ONLY the HTML document starting with <!DOCTYPE html>.

Prompt SHA-256

2e18dedbe148054a6adb42b29a65efb4c2e2e17a745177ea6a612e55cfa11140

Claims

  • Qwen artifact is preserved from the first-blog breakout gallery and copied into the public demo set.
  • Qwen artifact has exact prompt, budget, temperature, and run record provenance.
  • GPT-5.2 receives the recovered user prompt, temperature, and completion budget once; output is retained without repair.