Dijit.app Dijit.app Documentation

What Dijit.app is

Dijit.app is a document management platform built on generative AI and OCR for purchase-to-pay and accounting. It reads invoices, delivery notes, receipts and business paperwork; it extracts and checks the data on them, from the supplier and the line items to the taxes, the totals and the nominal account; and it hands the result to your ERP or accounting software. It also covers the work that comes next: matching delivery notes against invoices, auditing prices, raising purchase orders, costing recipes, flagging payments that are about to fall due, running approvals and keeping a searchable archive with a full audit trail.

What the product covers

Dijit.app works across the purchase-to-pay cycle and input VAT. Its job is to take away the typing, and the checking that always comes with it: confirming that what you were invoiced for is what you actually received, that the prices are the ones you agreed, and that every document ends up against the right nominal account.

What happens to a document

Every document goes through the same sequence, whichever way it arrived:

  1. It arrives. Uploaded through the web application on a desktop or a phone, sent in by email, dropped into a connected Dropbox folder, or pushed in through the API.
  2. It is read. OCR picks up the content and the generative AI models work out what each value means. This takes under three seconds a page.
  3. The data is pulled out. Header data first, meaning the issuer, the recipient, the dates, the numbers, the net amounts, the taxes and the totals, then the line items with item, quantity, price and discounts.
  4. It is checked. The system tests the document against itself: that the lines add up to the totals, that the taxes are consistent, and that this is not a document you have already had.
  5. It is coded. Nominal accounts are filled in from your own history and from the rules you have set.
  6. It gets a status. The document appears in its list with a status that either says the checks passed or points at the one thing that needs a look.
  7. You approve it, on its own or in a batch.
  8. It goes out. The data is posted to your ERP through a connector, or written to a file in the format your accounting software accepts.

You are not expected to check everything. Reading and coding are 99 % accurate on average on an original PDF; on a photograph it depends on the quality of the image. Every document is checked, but only the ones where a check came back inconclusive are flagged with a status. You look at the exceptions, not at the pile.

What it does

Purchasing

Accounting

Internal paperwork

Set up around how you work

The system is fitted to each customer's way of working with no development needed at your end:

Documents it reads

Purchase invoices, simplified invoices, import documents, till receipts, goods-received and sales delivery notes, transport invoices, purchase orders, production sheets, job sheets, payslips, payment documents and transport manifests.

Which of these you have depends on how your account is set up.

Ways documents come in

RouteHow it worksSet-up needed
Web application (desktop)Upload PDFs, images or ZIP files by hand, choosing which pages to process.None.
Web application (phone)On screens of 600 pixels or less you get Quick Access instead of the dashboard. The upload buttons for each document type lead to the same upload screen as on a desktop, with the same options: page selection, auto-merge and grouping photographs.None.
EmailA mailbox is synchronised and its attachments are processed on their own.Connect the mailbox, per document type.
DropboxA folder is watched and anything dropped into it is processed.Connect the account and pick the folder.
APIDocuments pushed in from your own systems.Switched on with the technical team.

The routes work side by side: one account can take most of its invoices by email while several people upload by hand at the same time. Where each document came from is recorded and shown in the list.

Where the data goes

There are two ways out:

On top of that there is Excel export with a layout of your own and a REST API for anything else. Connectors that are not on the list are built on request.

Performance, security and compliance

Reading and coding accuracy99 % on average on an original PDF. On photographed documents it follows the quality of the image.
Processing timeUnder 3 seconds a page.
VolumeBatch processing, from 100 to over a million documents a month. There is no design ceiling; capacity is sized per customer.
Unit of usageA processed page.
HostingMicrosoft Azure servers in the European Union.
ComplianceGDPR.
EncryptionTLS 1.3 in transit, and encrypted at rest.
Sign-inMulti-factor, by role.
AI modelsThey run inside Dijit.app's own Azure estate. Customer documents are never used to train models.
Interface languagesSpanish, English, French and Portuguese.

The security architecture and the compliance detail are set out in Security and compliance.

What it deliberately does not do

Documentation index

1. Getting started

Setting up an account, the free trial, signing in, the interface and a glossary of terms.

6 pages published

2. Account and plan

Profile, users and roles, plan and usage, group companies and preferences.

15 pages published

3. Uploading documents

Manual upload, ZIP files, phones, merging, email, Dropbox and where each document came from.

13 pages published

4. Working with documents

Lists, filters, statuses, approval, deletion, transfers and downloads.

10 pages published

5. Document types

Purchase, expense, sales and transport invoices; delivery notes, customs documents and bespoke formats.

14 pages published

6. Consolidation

Matching invoices against their delivery notes and dealing with the differences.

6 pages published

7. Checking the document header

Header data: supplier, customer, dates, numbers, amounts and taxes.

14 pages published

8. Checking the line items

Line items: quantities, prices, discounts, units of measure and ERP codes.

12 pages published

9. Master records and catalogues

Products, suppliers, clients and recipe costings. How a master record differs from a catalogue.

16 pages published

10. Accounting

Nominal accounts, how documents are coded and practice mode.

8 pages published

11. Purchase orders and projects

Raising and sending purchase orders, and tracking spend by project.

10 pages published

12. ERP integrations

SAP, Business Central, Sage X3, Ágora, ICG, Holded, Contasol, a3 and Cegid.

17 pages published

13. Dashboard and reports

Indicators, usage, summaries and data exports.

7 pages published

14. Help inside the app

Questions, issues and follow-up from inside the application itself.

3 pages published

15. Troubleshooting

What to check before raising an issue, and which support channel fits each case.

2 pages published

16. Reading correction rules

Rules that automatically correct the reading errors repeated across one supplier's documents.

13 pages published

↑ Back to top

Last reviewed: 11 September 2026 · The Dijit.app team