Free
€0for one user
- No credit card
- Hosted in Germany
- For testing, without a DPA
GDPR-compliant AI for company knowledge.
souveraen.ai answers questions from your documents and names the source. Documents and the search index stay in Germany, your content is never used for training, and from Pro a data processing agreement applies.
Free for testing with non-sensitive documents, without a DPA. From Pro with a DPA under Art. 28 GDPR.
When is an AI GDPR-compliant?
No AI is GDPR-compliant on its own. What matters is how your company uses it: which data goes in, where it is processed, who can access it and whether a data processing agreement is in place. Your data protection officer checks these points for your use case.
The provider can make that check easy or hard. It is easy when storage location, sub-processors, deletion and access rights are clearly described and a DPA exists. It is hard when inputs may be used for training or it stays unclear in which country data is processed.
That is why souveraen.ai describes exactly what happens to your data and makes no blanket promise of compliance. From Pro we sign the DPA with you, and the security concept is ready for your review.
Data protection by planOn Demand
- Location
- Germany
- Training with your content
- never
- Data processing agreement
- not included
- Meant for
- testing with non-critical documents
Data protection the platform enforces.
- Data in Germany
Documents, search index and accounts are stored in data centres in Germany. Every customer has their own tenant.
- No training on your data
The models run as fixed versions and are not trained further on your content.
- DPA from Pro
From Pro we sign a data processing agreement under Art. 28 GDPR. Free is meant for testing with non-sensitive documents.
- Rights checked on every request
The platform checks which workspaces the person may see. Restricted content appears neither in the answer nor as a source.
- A source for every answer
Every answer shows document, version and page. If sources contradict each other, the answer says so.
- Changes only after approval
Agents read without asking. Anything that changes something in your systems waits for your approval and is logged.
What happens to your data.
The questions data protection and IT ask before you start, answered side by side.
| Feature | Public AI chat on a personal account | souveraen.ai | Self-built with open models |
|---|---|---|---|
| Data processing agreement (DPA) | – | from Pro | not applicable |
| Guaranteed hosting in Germany | – | own hardware | |
| No training on your data | depends on settings | ||
| Own tenant per customer | – | not applicable | |
| Access check on every request | – | in-house development | |
| Log of retrievals and actions | – | in-house development | |
| Security concept for your review | – | write it yourself |
The comparison shows the typical features of public AI chats on personal accounts. Business plans from some providers may differ. As of October 2026.
ChatGPT, Copilot and data protection.
Public AI chats on personal accounts are made for private questions. For customer data, personnel files or client records they usually lack a DPA, guaranteed storage locations and rights per workspace. Business plans from some providers handle parts of this differently, so check the contract in each case.
The same questions help with ChatGPT, Microsoft Copilot and any other tool. Where is the data stored? Are inputs used for training? Is there a DPA? Does each person see only what they are allowed to see? And can you trace where an answer came from?
souveraen.ai answers these questions the same way on every plan: data in Germany, no training on your content, an access check on every request and a source reference for every answer. The DPA is available from Pro.
Questions for any AI toolAnswers for souveraen.ai
Documents, search index and accounts are stored in data centres in Germany, with a separate tenant for each customer.
No. The models run as fixed versions and do not learn from your content.
Yes, from Pro, under Art. 28 GDPR. Free is meant for testing with non-critical documents.
Yes. Every request checks which workspaces the person may see.
Every answer names document, version and page. If sources contradict each other, the answer says so.
1 of 5 checked
What happens with external models.
Alongside open models hosted in Germany, you can use larger models from external providers. These always run with Zero Data Retention: the provider does not store requests and answers and does not use them for training.
Chat and search run in the EU. Reranking and image generation may also be processed outside the EU, again without storage. You decide in the model catalogue which models your team may use.
Where a request is processedby model type
- Documents and search indexon the appliance
- Modelon the appliance
- Storage at the providernone
With Private Spark, no token costs
When data must not leave the building.
For records that have to stay in-house there is Private Spark: models, search index and documents sit on an NVIDIA DGX Spark in your network, with a management connection to souveraen.ai. Local requests cost nothing extra.
Air-Gap Enterprise works with no outside connection at all. We design it with you based on a quote. Both run the same software as On Demand.
Data centres in Germany. Open models in hosting, external models without storage. Free, Pro and Max.
Models, search index and documents on an NVIDIA DGX Spark in your network, with a management connection to souveraen.ai.
No connection to the outside at all. We size hardware and models to your workload.
What it costs.
All prices net, excluding VAT. Compare all plans
Pro
€24.90per user per month
- As many users as you need
- With a DPA under Art. 28 GDPR
- Hosted in Germany
Private Spark
€2,790per month, hardware included
- The AI runs on your own premises
- Local models with no token costs
- Can be ordered online
Questions before you start.
Rather talk in person?
We will show you souveraen.ai in 30 minutes, using your own example.
+49 3744 365 2202Is souveraen.ai GDPR-compliant?
Whether a deployment is GDPR-compliant is for your data protection officer to assess in your case. We provide the basis: data in Germany, no training on your data, rights per workspace, a data processing agreement from Pro and the security concept for your review.
Where is our data processed?
On the Free, Pro and Max plans, documents, search index and accounts are stored in data centres in Germany. External models always run with Zero Data Retention. Chat and search run in the EU, reranking and image generation may also run outside the EU. With Private Spark everything sits on an appliance in your building.
Do you sign a data processing agreement?
Yes, from Pro. The Free plan does not include a DPA. Use it for testing with non-sensitive documents.
Is our data used for training?
No. The models run as fixed versions and are not trained further on your content.
Is ChatGPT GDPR-compliant?
That depends on the plan, the contract and how you use it. Check with the provider whether a DPA is offered, where data is processed and whether inputs are used for training. A personal account is usually unsuitable for company data.
Who sees which content?
Every request checks which workspaces the person may see. Content they are not authorised for appears neither in the answer nor as a source.
Is souveraen.ai suitable for professionals bound by secrecy?
Yes. For law firms, tax advisers and other professionals bound by secrecy, we sign the confidentiality undertaking under § 203 StGB (German professional secrecy law) online.
Are you certified?
We do not advertise certificates. You receive the security concept and, from Pro, the DPA, and you decide yourself whether our operation meets your requirements.
Introduce AI with data protection.
Test for free with non-sensitive documents. For real data, we sign the DPA from Pro.
No credit card needed. Rather call: +49 3744 365 2202