Upload datasheet
PDF in, structured live datasheet out
How extraction works
OpenAI GPT reads the uploaded PDF and emits a single JSON object: every field, every table, and every figure with sampled (x, y) points per trace.
Each value is written with page · figure · confidence, so the live datasheet UI can deep-link back to the source.
The reviewer queue lets a human approve, fix axis ranges, or re-run extraction on a single figure.
Set OPENAI_API_KEY in .env to enable OpenAI extraction. Without it, the part name NXH003… loads the bundled hand-digitized fixture; anything else falls back to the synthetic stub.