AI is only valuable if it is safe.

Processing in EU data centres or on-premises in your environment. The rms AI Suite is designed for the European legal area – transparent, verifiable and controllable.

What that means in concrete terms:

✅ EU processing (hosting of your choice)
✅ AV contract (order processing) possible
✅ Privacy by design: data minimisation & deletion concepts
✅ No services outside the EU without your decision
✅ Optional: completely isolated operating models (depending on security requirements)

An AI solution that does not promise data protection, but anchors it architecturally.

Your content stays where it belongs: in your systems and storage locations – without any ‘side storage’.

✅ No use of customer data for model training
✅ No unwanted storage with third-party providers
✅ External model providers only after your approval setup (policy-driven)
✅ Vector storage & data management according to your architecture (e.g. customer-side / dedicated)

Security is not a function for us, but a principle.

Your content stays where it belongs: in your systems and storage locations - without "secondary storage".

✅ No use of customer data for model training
✅ No unwanted storage with third-party providers
✅ External model providers only after your approval setup (policy-controlled)
✅ Vector storage & data storage according to your architecture (e.g. customer-side / dedicated)

Security modules (depending on setup):
TLS-encrypted communication - Encryption of data at rest - Network segmentation/firewall rules - Dedicated environments

Only authorized persons are granted access - role-based and down to document level if required.

✅ API key/token auth (depending on integration)
✅ Roles & rights concepts for teams
✅ Optional: SSO/OAuth2/Azure AD (enterprise scenarios)
✅ Control of admin/editor/viewer permissions

Full transparency and control over who can see, use and manage what.

Every relevant action is traceable - as a basis for auditing, security and compliance.

✅ Logging of system and integration events (e.g. API calls)
✅ Audit trails for security-relevant processes
✅ Monitoring/health signals for operation & support
✅ Error analysis without unnecessary access to personal content

Data security over the entire life cycle - controlled, transparent and traceable.

✅ Deletion of chats, knowledge objects and logs after defined intervals
✅ Immediate manual deletion possible at any time
✅ Automated re-indexing when data is updated
✅ No shadow copying and no hidden storage
✅ TR-03108-compliant security procedures possible

Your data is only stored for as long as you wish - and not a moment longer

You decide which models are used - and under which security conditions.

✅ Use of leading models such as OpenAI, Gemini, Mistral, Llama, Phi-3
✅ Alternatively: use of your own model (e.g. Llama 3, Mistral 8x7B)
✅ Option for fully isolated inference servers (air-gapped)
✅ No automatic model forwarding or external model connection
✅ Optional rate limits, safety filters and individual content policies

Maximum control over model selection, access and safety levels.