When it has to be
exactly right.
We build desktop tools for the kind of work where a mistake gets expensive, starting with getting personal data out of a document before an AI ever sees it. They run on your own machine.
One tool, one principle.
The Pseudonymisierer is the one that travels: the GDPR applies across the EU, and getting real names out of a document before an AI sees it is the same problem everywhere. Two further programs, travel expense reports and receipt capture, are built around German tax law and are described on the German site.
Use AI without giving up control.
Language models save real hours in an office. At the same time, professional confidentiality and the GDPR do not allow real names, addresses and IBANs to be put into a cloud AI. Our answer to that is not a ban, but a switch that sits with you.
The Pseudonymisierer replaces personal data with placeholders before anything is sent, and puts it back into the answer. That way you work with the best model available without handing over client data.
The Pseudonymisierer can switch on a local language model: it finds occurrences that rigid rules miss, and it runs on your machine rather than in the cloud. It is off by default. You decide whether you need it.
Where a program can call a service on the internet, that function is off out of the box and runs through an account you hold with the provider yourself. We do not sit in between and receive no data.
If it doesn't exist, we build it.
Is there a routine in your office that eats time every week, with no program to take it over? That is exactly what we build: small, robust desktop tools made to order, built the same way as our own products.
Informally by email: what is done by hand today, and how often? We come back with questions.
Scope, price and timeframe, in writing and in plain language. Only then do you decide.
Installed on your machines, with no forced cloud, adapted further on request.