Security
Your company data stays in the EU.
Verselo processes chats, documents and company knowledge in EU data centres and trains no AI on them. This page shows what runs where, who has access and the one point where data can leave the EU.
What runs where?
Every part of the platform has a fixed place. One row is the exception, and it is listed too.
| What | Where | Operator |
|---|---|---|
| AI models from Anthropic (Claude) and OpenAI | EU data centres: Frankfurt with EU-only profiles and Azure EU regions | AWS Bedrock, Microsoft Azure |
| Database and files: chats, documents, company knowledge | Frankfurt, EU | Supabase |
| Application servers | Germany | Hetzner |
| Web search, only when a chat uses it | Only the search terms go to Google. Files, chats and company knowledge stay where they are. Admins can switch web search off. |
Who can reach your data?
No training on your data
Verselo trains no models on customer data. The model providers are bound to zero data retention: your requests are neither stored nor used for training.
Web search: the one exception
When a chat searches the web, Verselo sends only the search terms to Google. If you don't want that, switch web search off for the whole workspace. One switch in the admin area.
Access and roles
Sign-in with an email code. Admins invite people, assign the roles Admin, Editor and Member, set allowed models and budgets and revoke access in one click. SSO is coming soon.
Support only with your knowledge
Our support only sees your workspace through a support login: 30 minutes, you get notified, every access is logged.
Audit log and retention
The audit log is kept for 12 months. Your admins decide how long chats are stored: unlimited, 30, 90 or 365 days. Backups are encrypted.
Export under GDPR Art. 20
Admins export the whole workspace as a ZIP, each person their own chats. You take your data with you if you leave, and you can answer access and portability requests.
DPA and subprocessors
You sign the data processing agreement with Verselo at launch. Until then the subprocessor list is available on request, and at launch it will be on this page.
Help with AI Act Art. 4
Under AI Act Art. 4, companies must take measures to build their staff's AI literacy. The workshop with an attendance record and the overview of who uses AI in the workspace help you meet that duty. See the workshop →
How Verselo handles personal data on this website is set out in the privacy policy (Spanish).
Security questions
Where does Verselo process our data?
In the EU. Verselo runs the models from Anthropic (Claude) and OpenAI in EU data centres of AWS Bedrock in Frankfurt and Microsoft Azure in EU regions. Database and files sit with Supabase in Frankfurt, the servers with Hetzner in Germany. The exception is web search: when a chat uses it, only the search terms go to Google. Admins can switch web search off.
Does Verselo train AI on our company data?
No. Verselo trains no models on customer data, and the model providers are bound to zero data retention, so your requests are neither stored nor used for training there. Your documents only answer your own team's questions, with the source. Your admins decide how long chats stay in the workspace: unlimited, 30, 90 or 365 days.
Does Verselo offer a data processing agreement?
Yes, at launch. Verselo signs a data processing agreement with every customer, available at launch. Until then Verselo sends the subprocessor list on request. If your data protection officer needs the agreement earlier, mention it in the demo call.
What privacy features does Verselo offer as a ChatGPT alternative?
Verselo processes in EU data centres, does not train on your data, offers a DPA at launch and gives admins control over roles, allowed models, budgets, retention and the audit log. That does not make your use GDPR compliant by default. How your team uses AI stays your responsibility. Verselo gives you the tools for it.
Questions from your data protection officer?
Bring them to the demo call. We go through the table above row by row.