Your rules, in front of the model.
Steinkauz AI is a policy and evidence layer for AI operations.
Connect public cloud, private cloud, and on-premises at the same time. The same policy sits in front of them. Integrate the layer into the tools and workflows you already run.
Routing, budgets, and a record of what was allowed to leave sit in front of every request.
Same routing, budgets, and evidence: in chat, in the tools you already run, and on the API.
Consumers
Inference environments
Public managed
OpenAI
Anthropic
Mistral
- And many more
Private managed
Amazon Bedrock
Azure OpenAI
Google Vertex AI
- And many more
Self-hosted
- OpenAI-compatible
Control
Your data only reaches environments you allow
Policy runs before inference. Optional budgets cap spend. Every use and every change stays reviewable.
Data routing policy
Label the sensitivity of each conversation, file, and how much you trust each provider. Steinkauz AI blocks incompatible requests before inference runs.
Learn moreBudgets
Optionally cap spend per member or API key, and per provider. Steinkauz AI does not include inference.
Learn moreTransparency and auditability
Usage, cost, and policy decisions in one place, with a reviewable trail of configuration changes.
Learn moreDeployment
Run Steinkauz AI on our cloud, or inside your boundary
Steinkauz Cloud if we host it. Private Deployment if the software has to run inside your boundary. Same application, same controls.
Self-service
Steinkauz Cloud
Hosted by us in the EU. Same product as Private Deployment. You connect your own model keys.
- One monthly organisation price, members included
- Your own inference keys (BYOK)
- Chat, API, routing, and audit on the shared EU stack
Contact
Private Deployment
The same software as Cloud, in your data centre or your own cloud account. You operate it. We license it, help you go live, and ship updates.
- Your infrastructure, identity, and data
- We help you deploy and keep it updated
- Optional support. Quoted annual licence
Note
Where Steinkauz AI runs is independent of where inference runs. You can stay on Cloud and still send sensitive work to a private or self-hosted provider. See Cloud and Private Deployment