Process data for AI training · Made in Munich
European company data, cleared for AI labs.
Earn €20,000+ from process data your team already creates at work. We anonymise it in the EU and license it to AI labs. You stay in control of what is shared.
- Your data stays yours. We license an agreed, anonymised copy. Your originals and ownership stay with you.
About 1 minute · No data upload · No obligation
HubSpot · Anna Huber, Head of Sales MunichPerson A, Sales
Pricing proposal: Pro plan +6%
SAP · Jonas Weber · j.weber@acme.comPerson B
Margin check: at least 8% needed
HubSpot · Pro plan
€49→ €53Jira · Anna HuberPerson A
Ticket “Update price list” done
What we license
Your operations. Never your customers' data.
Under a data processing agreement, your customers' data isn't yours to license, so it never enters a licence. What labs pay for is how your own team works.
Included
- How your team works across tools: chat, email, CRM, ERP, tickets, code
- Decisions, approvals, corrections and what came of them
- Your own AI-agent sessions and the fixes your team made
Always excluded
- Data you process on behalf of customers (anything under a data processing agreement)
- Anything you can't or don't want to share. You decide source by source
- Company names and confidential details. Towards buyers, your company stays confidential
Why labs pay
What is AI data licensing?
AI data licensing means granting an AI developer the right to train or evaluate models on a dataset, with a defined scope and clear restrictions.
The internet is filling up with AI-generated text, so real human data is getting scarce. What labs can't find anywhere else is how people actually get work done: across several tools, with mistakes, corrections and a result at the end.
Your company produces exactly that every day, as a by-product of running the business: process data, the traces of how work moves through your tools. Licensing lets you earn from it without giving up control, and without your raw data ever leaving Europe.
IsarTrust is a Munich-based data licensing company that refines process data from European companies into GDPR-compliant AI training data and licenses it to AI labs worldwide. IsarTrust is a product of Knowlix GmbH.
What counts
Which process data is valuable for AI training?
Records of your own work across the tools your team uses, with a visible outcome. Any industry, from 25 employees and three consecutive years of data.
Sales and pricing
Pricing decisions, approvals and deal updates across CRM, email and chat.
Finance and operations
Purchasing, month-end close and approvals in ERP and spreadsheets.
Engineering and product
Issues, pull requests, reviews and incidents.
Internal IT and helpdesk
Requests from your own team, from diagnosis to fix.
Knowledge and processes
SOPs and wikis, and how they're actually used.
AI-agent traces
Prompts, tool calls, human corrections and results.
How it works
How does licensing your process data work?
Your effort depends on the data sources: for structured data without communications, typically a few hours.
- 01
Assess
Answer a few short questions about your team and tools. No data is shared. You see right away whether licensing is relevant for you.
- 02
Clear
We work through the rights with you: GDPR legal basis, works council and trade secrets, and we separate out anything you process for customers. You decide what's in and what stays out.
- 03
Refine
In the EU, our Trust Engine anonymises your data: names are replaced, details generalised or removed, and the result is checked. We then restructure it into training tasks and complete work records. Raw data never leaves Europe.
- 04
License and earn
We license the refined datasets to AI labs. You decide what is shared. Your past data brings a first payment from €20,000, and you earn again whenever data from your ongoing stream is licensed.
Why IsarTrust
Built for European rules, not around them.
- Where raw data is processed
- Typical brokerOften outside the EUIsarTrustOnly in the EU
- Personal data
- Typical brokerNames removedIsarTrustAnonymised against EDPB guidance
- Employee data
- Typical brokerRarely addressedIsarTrustWorks-council process built in
- Documentation for buyers
- Typical brokerVariesIsarTrustProvenance ready for the EU AI Act
- Your identity
- Typical brokerOften shared with buyersIsarTrustConfidential by default
For AI labs
Sourcing European training data?
Real-world process traces from European companies: rights-cleared, anonymised in the EU, and documented for the EU AI Act.
Who we are
Built in Munich by people who put privacy first.
Founded by Peter Meier and Francesco Wiedemann. Peter co-founded Metaio, the Munich augmented-reality pioneer Apple acquired in 2015. IsarTrust is a product of Knowlix GmbH in Munich.
About IsarTrust →FAQ
Questions companies ask us
Is it legal to sell company data to AI labs in Europe?
Whether and which data can be licensed is something we review together with you in the clearance step, source by source, not across the board. That covers the legal basis for the new purpose, works-council involvement where employee data is concerned, rights to third-party data and the protection of trade secrets.
What company data is valuable for AI training?
Complete records of how your own team works, with a visible outcome. Examples are a pricing decision that moves from a chat thread through a finance check to an updated price list in the CRM; an internal IT request from report to fix; code changes with reviews and incidents; or AI-agent sessions with tool calls and human corrections. Isolated records, data without outcomes and content that is already public are usually worth much less.
How much is my company's data worth?
We only start a licence if it earns your company at least €20,000. Below that, it isn't worth the effort for either side. Above that, the value depends on how rare, complete and recent the data is, how much of it there is, and which rights you grant. Deals range from one-off licences to recurring revenue shares for data you keep producing. The eligibility check takes about a minute and is the first step to a realistic estimate.
What is process data?
Process data are the traces of how work moves through your tools: a pricing decision that goes from the CRM through a finance check in the ERP to a closed ticket. Companies produce it anyway while they work. Licensed in anonymised form, it teaches AI models how real work gets done.
What requirements does our company need to meet?
We currently license data from companies with at least 25 employees and at least three consecutive years of process data, for example from CRM, ERP, ticket systems or code repositories. Industry doesn't matter. The eligibility check shows in about a minute whether you qualify.
How much effort is it for us?
Your effort depends on the data sources. For structured data without communications, it is typically a few hours: an intro call, deciding which tools and time periods are in scope, and approving a sample. Where employee data or communications are involved, steps such as works-council agreements, a balancing of interests and informing employees can be added, and we agree the timeline with you. We do the technical work. You receive a first payment for your past data, and you earn again whenever data from your ongoing stream is licensed.
Can a buyer identify my company?
Buyers receive the dataset under a code name. Company names, people and addresses are replaced consistently across every source, and you review a sample before anything is delivered. Your name is never shared with buyers.
Will my employees' messages end up in an AI model?
Not as they are. Communications are the most sensitive data type, so they get the strictest controls: personal information is removed or transformed until individuals can't reasonably be identified, the works council is involved where one exists, and you can exclude any source completely. Many licences don't include communications at all.
What happens to my customers' data inside our systems?
It stays out. Data you process on behalf of your customers, under a data processing agreement, is excluded from every licence. We only license records of your own operations, anonymised so that neither individual people nor your company can be traced. During the review we separate the two together with you, and nothing is licensed without your explicit approval.
Does IsarTrust use Knowlix customer data?
No. Knowlix processes customer data as a processor and may not use it for its own purposes. IsarTrust sources data exclusively through its own licence agreements with suppliers.
Do you license data to US labs?
Yes. IsarTrust licenses to AI developers worldwide, including US labs. Raw data is processed only in the EU. Only the anonymised dataset leaves Europe, and individual people can no longer be traced from it. Buyers also receive the provenance documentation they need under the EU AI Act.
What is the difference between anonymised and pseudonymised data?
Pseudonymised data has direct identifiers such as names replaced, but it can still be linked back to a person with additional information, so GDPR still applies. Anonymised data can't reasonably be linked to any individual by anyone, so GDPR no longer applies. Removing names alone almost never makes data anonymous.
Who owns the data after licensing?
You do. You grant IsarTrust a licence to refine your data, and IsarTrust licenses the refined dataset for agreed purposes, such as training or evaluating AI models. Ownership of your original data doesn't change, and you decide what is included and what stays out.
How does IsarTrust make money?
IsarTrust earns a share of each licence it sells. The eligibility check is free, and there is no obligation to license anything afterwards.