HomeIntegrations › Fortnox
Accounting · Sweden

What SayFocus adds on top of Fortnox's invoice reading

Fortnox already reads invoices on its own. A pipeline on top handles the coding, matching and exception review that turn a read invoice into a posted one.

Native invoice readingFortnox already extracts invoice data through Fakturatolkning; the gap is account coding, order matching and instant updates.
Region SwedenNative capture Fakturatolkning (built-in OCR)API updates Poll-based, no pushE-invoicing Via operators & PEPPOL
Captureinvoice arrives by email or upload
Native OCRFortnox reads the invoice on its own
Codelines mapped to chart of accounts
Matchlines checked against purchase orders
Poll & postpipeline polls Fortnox, posts exceptions

What lands in Fortnox

Header fields

Supplier, invoice number, dates and totals are set on the purchase invoice record.

Line items coded

Each line carries an account code from the client's own chart of accounts.

Order reference

A matching purchase order is linked where one exists.

Source PDF

The original invoice file is attached to the record.

Exception flag

Anything uncertain is marked for review instead of posted.

Review task

A task is created in the client's existing review channel for anything flagged.

Where it gets awkward

Every system has edges. These are the ones that shape the build — better known before the pilot than during it.

No push notifications

Fortnox's API does not notify an integration when a record changes, so the pipeline polls on a schedule for new and changed invoices rather than reacting instantly.

OCR reads, it doesn't code

Fakturatolkning extracts the invoice, but mapping lines to the client's own chart of accounts and matching them to orders still needs a layer on top.

Is a custom pipeline worth it here?

Worth building
  • Suppliers send documents in different formats and through different channels.
  • Invoice lines need coding to a detailed chart of accounts, not a supplier name.
  • Purchase orders or goods receipts need matching before anything is posted.
Native tools are enough
  • One supplier, low volume, and checking Fakturatolkning's output by hand is fine.
  • No order matching or account-level coding automation is needed.

The task behind this

Solution

Supplier documents reconciled before they hit the books

Match supplier invoices to purchase orders and delivery notes, check the arithmetic and VAT, flag duplicates, and post only clean documents.

How the stream is built

Processed in the EU. The original files and extracted data stay in your own systems; a data-processing agreement is available on request — see security and data handling.

From audit to autopilot

Step 1

Audit

We map one document or enquiry stream end-to-end, count volumes and name the systems it touches. You get a written scope and a fixed pilot price.

Free · 3 working days
Step 2

Pilot

One stream goes live on your real documents. Acceptance is measured, not felt: straight-through rate, minutes per document, exceptions.

About two weeks · fixed price
Step 3

Rollout

The remaining streams follow the proven pattern. Monthly report, support, and changes as your systems change.

Stream by stream

Frequently asked questions

Does Fortnox already read my invoices?

Yes. Fortnox has native invoice interpretation, called Fakturatolkning, that extracts data from incoming invoices. It does not code each line to your chart of accounts or match it against a purchase order, which is where a pipeline adds value.

Why would I need anything else if Fortnox has OCR?

Fakturatolkning gets the header and line data off the page. Turning that into a correctly coded, matched, ledger-ready entry across many suppliers and formats is the part that still takes manual work, and that is what a pipeline does.

How does the pipeline know about new invoices if Fortnox doesn't send webhooks?

Fortnox's API has no push notifications. The pipeline polls the API on a schedule, checking for new and changed purchase invoice records and picking up anything it has not seen before.

What happens if the pipeline is not confident about a line?

It is not posted. The document goes to a review queue in a tool you already use, so a person makes the final call before it reaches the ledger.

Where is the data processed?

Processed in the EU. The original files and extracted data stay in your own systems; a data-processing agreement is available on request.

Using Fortnox? Start with the audit

Three working days, one document stream mapped end to end, a fixed pilot price on paper. No commitment.

Get a free automation audit