Upload today's tickets
Add today's tickets — photos or PDFs. You can drag them in or click to browse — JPEG, PNG or PDF, as many as you like. A multi-page PDF is treated as one ticket per page.
Extract the data
Press the button and the tool will read the device info from every ticket. This can take a minute — that's normal.
Reading your tickets…
Please leave this page open. This usually takes under a minute.
Review the spreadsheet
Download the spreadsheet, open it, fix any colored cells, then save it. The colors tell you exactly what needs attention.
The download will appear here right after you finish step 2.
What the colors mean
- White — confident. Nothing to do.
- Amber — a suggestion. Confirm it or fix it; the Notes column says where it came from.
- Red — nothing to suggest. Please fill this in.
- Green — price filled in from the hospital price list (step 5). Nothing to do.
- Rose — an estimated price (step 5). Check it before you rely on it.
When you're done, save the file and move on to step 4.
Send back your corrections
Upload your saved spreadsheet (or several). The tool uses your fixes to get smarter over time. Send back a spreadsheet you ran through step 5 and it also remembers the prices — so next month those cells arrive already filled, instead of blank.
Fill in missing prices
Some components have no price because the ticket quoted one construct total. Upload the spreadsheet and the tool looks each one up in the hospital price list.
Getting better over time
The share of labels the tool reads confidently each day. The more corrections you send, the higher this climbs.
What the reading costs
What the AI reading has cost, and what that works out to per month at the current rate. Both providers have stopped a batch by hitting a spending limit, so this is here to give you warning rather than a failed run. Dollar figures are estimates from published prices.
What you keep having to fix
Every correction you've sent back, grouped by which field it was. The fields at the top are costing you the most review time — and the “wrong without warning” column is the one worth watching: those are cells the tool was confident about, so nothing was coloured and nobody was asked to check them.
What the tool has learned
Everything the tool has picked up from your corrections so far.
Retraining activity
When you sent corrections back, and what the tool learned from them.
Correction uploads
Daily learning impact
Work done
Spreadsheets you've created before. Click to download one again.
Reference data
These are the source/lookup sheets the tool checks each ticket against. They change rarely — usually refreshed monthly — and are kept separate from your daily ticket processing. Upload a new version of any sheet below and it takes effect immediately for new batches.
GTIN Codes
Product master — maps GTIN to SKU.
Part Info (Part Type & Category)
Ref Number to Description, Part Type and Category.
Surgeon Info
Surgeon / hospital / region crosswalk.
Hospital Price List
Per-hospital component prices for step 5. Uploading replaces every tab.
Expiry Log
Lot to expiry validation source.
Daily status email
A day nobody runs the app looks exactly like a day with no surgeries — nothing happens, and nothing says so. This sends a short summary every weekday at 5pm, whether or not anything was run, so a missed day is visible the same afternoon instead of at month end.
The email contains counts only — never a patient, surgeon or hospital name.
Debug Console
Upload one ticket to see every decision the pipeline made: which barcodes decoded, what the AI read per field, which lookup tables matched, how prices were validated, and exactly why each cell ended up white, amber, or red.