Demo · /upload

Drop one supplier feed, normalize it inline, diff it on the next run.

Single CSV or PDF per supplier. Every row gets a confidence score, and the per-row price-change badge lights up as soon as a second upload arrives for that supplier — so re-uploading the same supplier shows the diff end-to-end.

The diff story

Where it shows up
How a per-row badge lights up once a second upload arrives for the same supplier.
01

Drop one supplier feed

Upload a single PDF or CSV; the page parses it inline. Each row carries its own confidence score.

02

Upload again to compare

When the same supplier is uploaded a second time, the prior run becomes the baseline and per-row diffs fill in.

03

Diff badges attach to SKUs

Up / down / new / retired badges line up next to each row — driven by the heuristic + LLM result, not a re-typed catalog.

Recent uploads

Latest runs
Last few supplier feeds, newest first.

Upload

Drop one supplier feed, parse it inline
One PDF or CSV per supplier per upload. Larger files route through the LLM when the heuristic looks weak.

Use the name your suppliers write on their filename header. Re-uploading under the same name lets the diff surface.

Catalog

Rows will appear here
Drop a PDF or CSV, hit Parse & diff, and the parsed rows land inline with a per-row confidence score and the diff badges above (once a prior run exists).