Invoice Processing Overview
Brikly's invoice processing engine takes the tedium out of cost management. Instead of manually keying line items from every delivery, you upload (or capture) an invoice and let Brikly do the heavy lifting - extracting data, matching ingredients, and updating your costs automatically.
The end-to-end flow
Every invoice follows the same four-stage pipeline:
- Ingest - Drag-and-drop a file into Brikly, capture it from a supplier portal with the Chrome extension, or forward it to your Invoice Inbox email address.
- AI extraction - Brikly's AI reads the document and pulls out the supplier name, invoice number, date, line items, quantities, prices, and VAT.
- Matching - Extracted line items are matched to the ingredients already in your system. The supplier is identified and linked as well.
- Price update - Confirmed prices flow through to your recipes and dishes, so your costings are always current.
Ingest ──▶ AI Extraction ──▶ Matching ──▶ Price Update
(upload, (supplier, (ingredients (recipes &
extension, line items, & supplier) dishes
inbox) VAT, totals) updated)
How it learns over time
Brikly is not a static tool. Every invoice you process teaches the system a little more about your suppliers and their products. Early on you will confirm matches manually; after a handful of invoices from the same supplier, the system handles most items automatically. Read more about this in The Learning System.
Time savings
| Task | Manual approach | With Brikly |
|---|---|---|
| Key in a 30-line invoice | 8–12 minutes | ~30 seconds of review |
| Match items to ingredients | 5–10 minutes | Automatic (once trained) |
| Update recipe costings | 10+ minutes | Instant |
For a busy kitchen receiving dozens of deliveries a week, that adds up to hours saved - and far fewer transcription errors.
Supported formats
Brikly accepts the most common invoice formats:
- PDF - standard supplier invoices, statements, and credit notes.
- PNG / JPEG - photos of printed invoices or screenshots from supplier portals.
For the best extraction accuracy, upload the original PDF rather than a photo whenever possible. PDFs with selectable text are processed faster and with higher confidence than scanned images.
What happens next?
Explore the rest of this section to learn how each stage works in detail: