Use cases
Using AI with e-invoices at work
Structured invoice data can be combined with supporting documents. AI can help with search and interpretation, while people remain responsible for review.

Understand formats and obligations
An e-invoice contains structured data that can be processed electronically. A plain PDF is not automatically an e-invoice. XRechnung and ZUGFeRD are examples of formats; seek professional advice on the German obligations and exceptions applicable to your business.
What does – and does not – need AI?
Structured XML can be parsed using deterministic rules; a language model is not automatically needed. AI can help interpret unstructured supporting documents, answer questions or prepare comparisons. Results and sources still require review.
An example with human review
Receive an invoice → parse structured data → check rules and completeness → search or summarize when appropriate → have a responsible person approve the result. Ambiguous or inconsistent entries go to manual review.
Prerequisites and pilot
A bounded pilot defines document types, data access, integrations, permissions and review questions. Accounting, retention and tax responsibilities remain with the responsible people. AI alone does not guarantee legal or accounting compliance.
German Federal Ministry of Finance: e-invoice information
Discuss your e-invoicing use case