Questions

Questions & answers

Answers to the most common questions about AI data extraction, completeness and data sovereignty.

• How can you verify that AI extracted all data from a PDF?

  By not taking the output on trust, but reconciling it against an independent reading of the original: every extracted value needs a location in the PDF, every number in the PDF must appear in the output, and row counts and totals must add up. Anything without a match is flagged as a gap and listed in a verification report.

• Why do AI document conversions silently lose data?

  Because language models produce plausible output, not provably complete output: if a document is too long, the output is cut off; complex tables get shortened or shifted; footnotes and annexes are dropped. The result still looks finished – the error only surfaces when someone needs the missing data.

• Can you use large AI models without sending data to the cloud?

  Yes. Capable open-weight models can run on your own hardware – in your own data centre, at a provider in the country of your choice, or fully offline. The data then never leaves the controlled environment; the price is GPU infrastructure and operational effort.

• How do you convert PDFs to Markdown and JSON for AI agents and RAG?

  In three steps: read the document structurally (text, tables, headings, positions), produce Markdown for running text and JSON for fixed fields, and verify the result against the original. What matters is that every element keeps its source location – otherwise neither a human nor an agent can back up an answer later.

• What does on-premise or fully offline AI document processing require?

  Three things: hardware with enough GPU memory for the chosen models, a software chain of extraction, models and verification that runs without internet access, and an operating process for updates, monitoring and model changes. How large that turns out depends mainly on document volume and the required depth of verification.