Feature · AI account coding
AI invoice capture: from document to reviewable entry.
accuno combines document content, business partners and client-specific history into a reviewable account coding suggestion. Check accounts, tax codes and amounts, correct where needed and approve the posting.
An everyday example
Review a supplier invoice without retyping it
A supplier sends you an invoice as a PDF. Instead of entering its number, dates and amounts in several forms, upload the document to accuno. The original and the extracted fields appear side by side for review.
Read the document
accuno extracts the invoice data and prepares proposals for the business partner, account and tax code. The proposal gives you a starting point; you decide whether it reflects the actual transaction.
Correct against the original
Compare the dates, amounts and tax details with the PDF. Correct any misread field directly. Even a proposal whose figures add up needs your review, for example for a credit note or an overseas supplier.
Continue with a draft
Transferring the result saves a draft entry. Add missing information and check the account coding before posting. The original document stays linked to the transaction.
The practical benefit is reviewing and correcting existing invoice data instead of entering everything again. Extraction, drafting and posting remain distinct, traceable steps.
What it does
Text PDF, scanned PDF, images, and e-invoices
Searchable PDFs and non-searchable scanned PDFs, JPG/JPEG, PNG/APNG, GIF, and WebP plus XRechnung in UBL or CII syntax and ZUGFeRD/Factur-X. Files are checked for malware and existing content duplicates before storage.
Posting details in context
Business partner, document date, service date, amounts and tax details appear alongside the original document for your review.
Account-coding suggestion right away
A traceable suggestion based on receipt type, partner data, and client-specific history, ready for your review.
Corrections in the learning history, no model training by accuno
Your corrections remain as client-specific, traceable evidence and can support the suggestion logic. accuno does not use them to train or fine-tune a foundation model; you retain control over the final account suggestion.
Batch processing
Start up to 500 receipts in one run, process them in the background, and follow progress per receipt.
Small tools that make everyday work easier
For the steps you repeat throughout the working day.
- Review documents with the keyboard
- Use the arrow keys to move between documents. Press Enter to mark a document as reviewed when you are not editing an input field.
- Remove unsuitable learned rules
- Disable or reset tenant-specific rules when they no longer fit. Suggestions still need your professional review.
- Keep AI usage in view
- Set a monthly extraction budget. New extractions stop when it is reached. Requests already running are still completed.
Typical scenarios
Freelancers
Read receipts as a batch, review suggestions clearly, and work through the run step by step.
Business keeping its current accounting system
Review account coding suggestions and document data in accuno, export the results as CSV and continue in your existing system. The required import mapping depends on the target software.
Tax advisors
Clients submit documents. Review suggested accounts, tax codes and amounts, resolve exceptions and approve postings.
Practical guide
3 reasons why AI will not fully replace accounting
AI will automate accounting deeply, but it will not replace it completely: agents still require a reliable accounting system, automated workflows need traceable controls and accountable ownership, and compute-intensive models create the most value when used selectively for ambiguous work.
Read the guide →Works with
Bank import
Extracted and posted invoices create open items that banking ranks as payment suggestions.
More →
DATEV export
Extracted and posted receipts as a DATEV (Germany's accounting-interchange standard) posting batch for your tax advisor.
More →
Compliance Center
Posted receipts remain traceable; accuno does not use the documented corrections to train or fine-tune a foundation model.
More →
Frequently asked questions
Which receipt formats are supported?
Searchable PDFs and non-searchable scanned PDFs, JPG/JPEG, PNG/APNG, GIF, and WebP plus incoming XRechnung as UBL/CII XML and ZUGFeRD/Factur-X with embedded invoice XML. Multi-page PDFs remain one connected receipt and are processed together.
How do I review an account coding suggestion?
Compare business partner, accounts, tax codes and amounts against the original document and the actual transaction. Correct the suggestion as needed and decide whether to post it.
What happens if the AI reads a field incorrectly?
Overwrite the value directly in the extraction workspace. The correction is retained as client-specific evidence, and you confirm the final suggestion yourself.
Are my receipts used to train the AI?
accuno does not use receipts or corrections to train or fine-tune its own or a third-party foundation model, and it does not send the learning history as a training dataset. The required receipt content is still processed by an approved inference provider for the extraction you request. The privacy notice and DPA describe that provider's contractually approved purposes, retention, processing locations, and safeguards.