Making company knowledge searchable. Every answer backed by a source.

What does “searchable” mean?

Searchable means: you ask a question in plain language and get an answer from your own documents, with the source and page number.

No flipping through binders, no Ctrl+F across ten PDFs, no hunting your inbox for the right keyword. You ask the way you'd ask a colleague, and you get an answer backed by evidence straight from the original document.

This pays off especially where the same question keeps coming up but the answer sits somewhere different each time: contract terms, internal rules, a customer's history.

How is the answer generated?

A search first finds the matching passages in your documents, then a language model turns them into an answer and names the source.

1. Search

   Finds the matching passages in your documents.

2. Answer

   A language model reads the passages the search found and summarises them for you. It does not invent anything freely.

3. Source

   The exact spot in the original document comes with it. Every answer can be double-checked in seconds. No answer without that source.

This principle is called Retrieval-Augmented Generation, or RAG: search first, answer second, always with a source. That is what sets it apart from a chatbot that answers from its training and cannot name a source.

Which documents work?

Contracts, files, e-mails, manuals and scanned paper all work, as long as the text is prepared in a readable form.

Paper files

The "Paper made searchable" package scans a box of files in a single day, names and sorts every sheet, and then makes it searchable.

E-mail threads

Instead of an inbox full of subject lines, the system searches the content of the threads and finds the spot where a commitment was actually made.

Handbooks and know-how

What used to sit as a note on someone's desk or in a colleague's memory becomes something the whole team can ask.

Which formats work in your specific case is something we check against your actual documents in the intro call.

Where does the data live?

On servers in Germany, or entirely on your own hardware on-premises if you prefer.

• Language models with a German provider in Berlin, under a data processing agreement
• No training on your data
• No data shared with third parties and no analysis by outside providers, regardless of how sensitive the documents are

For the details, including the legal basis and the contract behind it, see GDPR-compliant AI.

How long does setup take?

One box of paper in a day, a digital collection in two to four weeks. Which of the two applies to you is what we work out in the intro call.

Sort first, scan second. Before anything is scanned, we sort your collection into three groups right at the shelf:

• What comes in from now on
• What your team still handles regularly
• What only sits in the cabinet because of retention rules. This group stays paper.

In most businesses that halves the volume before the first binder is opened.

Setup in three phases.

1. Review and prepare documents

   Paper: one box of files is scanned and searchable in a day. Ten boxes are one working week on site.

2. Set up the system and test it on a sample

   Digital files (PDF, Word, e-mails): setup and a test run on a sample take one to two weeks. You already see at this point whether the system works for your documents.

3. Extend to the full collection

   The rollout then runs in the background. After four weeks at the latest, your team works with the complete collection.

For larger projects, I measure on a real batch from your collection how many sheets pass through per hour before I quote. That half-day pilot run is part of the offer and replaces any guesswork.

What does it cost?

The entry point is one of the three fixed-price packages from €990, larger projects are billed by effort.

All three packages know their price upfront, not only at the end. All prices and what each package includes are on the pricing page.

What does it look like?

The easiest way to see it is the interactive demo on the homepage: pick an industry, ask a question, watch the answer arrive with its source.

The demo runs on sample documents from different industries, so you can see the principle on a case similar to your own before committing your real documents.