Your people are already using AI. Your data is not allowed to go with it.
Both halves are real and both are enforced, and the place to resolve them is the point where the conflict actually happens — the moment somebody presses send. anonymize.solutions sits between the application and the model and takes the personal data out first: on our servers in Germany, or entirely inside your own perimeter where the text never leaves the machine. Use any AI. Give away nothing.
What sits between your text and the model
- 04 · Protection methods5 protection methods, one of them reversible
- 03 · Trained language modelsnames and places in context in 42 of 48 hosted languages
- 02 · Pattern recognisers267 types hosted · 317 own, local
- 01 · Checksum rulesLuhn, MOD-97 · arithmetic, re-runnable
anonym.plus Enterprise: the anonymization server inside your own network
anonym.plus, run as a server in your own network. From 10 client connections upwards, open-ended, scaled by your hardware and set up with you.
For regulated organisations
Hospital groups, banks, insurers and public authorities: the data stays in your house, and the tender clause gets an answer.
The data stays in your house →For software and system houses
REST and MCP inside your customer’s network, and white-label as a custom build under your name.
The engine inside your product →Other tools already claim this problem
Discovery platforms, blocking tools and open detection models are all real products with real customers.
Discovery tools look in the wrong place
Data-discovery platforms find personal data sitting in buckets, warehouses and file shares. They are good at that. They never see the prompt that is being submitted right now, which is where the exposure has moved.
Blocking tools are defeated by their users
An outright ban does not reduce usage. It moves the work to a private device, where there is no log, no policy and no control at all — and the organisation loses even the ability to measure the problem.
A model is not a product
Open detection models are published regularly, and they lower our cost of goods rather than threatening us. What a regulated buyer purchases is the control and evidence layer above the model: reversibility, a processing agreement, liability, support, a deployment they are allowed to run.
The same problem, six different regulators
Anonymization is not a vertical product. The regulation, the failure mode and the deployment that survives procurement are different in every sector — so the packages are, and the brands are not.
Healthcare and life sciences
Discharge letters, findings, coding, cohort studies. The most useful AI applications in a hospital operate on exactly the data with the strictest legal protection there is.
What breaks, and what we deploy → (Healthcare and life sciences)Legal and professional services
Review, drafting, disclosure, due diligence. A law firm has a duty of confidentiality that is older and stricter than data-protection law, and it is owed to the client, not to a regulator.
What breaks, and what we deploy → (Legal and professional services)Financial services and insurance
Know-your-customer files, claims handling, complaints, fraud narratives. Finance is the sector where detection has to be provable rather than merely good, because somebody will audit the log.
What breaks, and what we deploy → (Financial services and insurance)Education and research
Pupil data is the category where consent is rarely the right basis: in an obligatory school relationship it is hardly freely given, and it cannot cover a lesson that already happened. Marking, planning, pastoral notes and SEN documentation all run on exactly that data, and the school is the controller for every use of it.
What breaks, and what we deploy → (Education and research)Software and product development
Coding assistants, agent frameworks, support tooling, test data. Development teams handle other people’s customer data constantly and are the fastest adopters of exactly the tools that forward it.
What breaks, and what we deploy → (Software and product development)Public sector and critical infrastructure
Case files, citizen correspondence, freedom-of-information responses, procurement records. A great deal of text to process, and unusually little freedom about where it may be processed.
What breaks, and what we deploy → (Public sector and critical infrastructure)What differs by sector is the detection preset, the document templates and the evidence pack your auditor asks for, assembled with you. What does not differ is the company you contract with: one sender, one agreement and one place to audit.
Two products and a service, sorted by one question
May the text leave the machine it starts on? If yes, anonym.legal does the work on our servers in Germany. If no, anonym.plus does it on yours — on the machine itself, or with Enterprise on a server inside your own network. Both build on the same Presidio detection core, each with its own recognizer set, both are in production today — and for large estates we do the rollout and the scaling with you.
anonym.legal
Hosted in Germany. The text is sent, encrypted, to our servers.
Free, then €3, €15 or €29 a month
anonym.legal →anonym.plus
Runs on your own machines. The text never leaves them.
Free for one machine, then €149, €399 or €599 once; Enterprise from 10 client connections, through sales
anonym.plus →Implementation and scaling
For large estates, delivered by anonymize.solutions.
By agreement
Implementation and scaling →What each one rules out
Enable, do not block
Every blocking tool produces evasion. We make the use permitted instead of invisible.
Prove, do not claim
Deterministic detection rules and a traceable record. No accuracy percentage is published on this site because no benchmark exists yet.
European, not dependent
Processing in Germany, or on your own machines, and no binding to any one model provider.
Reversible, not destructive
Pseudonymization you can come back from. The authorised reader gets the plaintext answer; pseudonymized data remains personal data under the GDPR.
We publish why anonymization fails, including ours
anonym.community is a research hub with no product in it and nothing to buy. It exists because the honest version of this field is a catalogue of failure modes, and a vendor that cannot name them has not looked.
Two doors, and they lead to different conversations.
If you are evaluating this for an organisation, start with your sector — the regulation and the deployment differ enough that the generic answer is useless. If you are looking at the company, say so and we will send the deck.