LOCAL AI
Local AI: DENKRAUM AI on your own computer or server
You decide where DENKRAUM AI runs, which models you connect and what your agents are allowed to do. With local models, the AI processing can also run within your own network – no cloud lock-in.
Three ways to run DENKRAUM AI
- Desktop app for Windows: installs like an office program. Your data sits in a folder of your choice, and backups run automatically every day.
- In a team: one switch shares DENKRAUM AI on your local network; your team works along in the browser – with roles from admin to reader.
- On your server: for larger teams, DENKRAUM AI runs on a server of your own, with local models if you wish.
Local models or the cloud – per task
Under “Models & providers” you connect Anthropic, OpenAI or a local model, for example via Ollama or vLLM. Agents and models can be combined depending on the task. During setup, we check which local or external models suit your use case; hardware requirements, quality and running costs depend on that choice.
If you use local models and connect no external services, the AI processing can also take place within your own network. With cloud models or external integrations, the data flows you set up and the providers’ terms apply.
Control built in
- Changes to knowledge are proposals; for critical steps, you build human approval into the workflow.
- Sources, note versions and the activity history make answers and changes verifiable.
- Markdown instead of a closed repository – Obsidian-compatible and connectable to your systems via MCP or HTTP.
Transcript
In Settings, you adapt DENKRAUM AI to the way you work. Under “General” are the name and description of the workspace – and which agent answers first in the chat. Under “Team” you add people and assign roles: admin, editor, member or reader – with access per knowledge space. You choose the design for your browser – light, dark or one of many colour themes. Under “Models & providers” you connect Anthropic, OpenAI or a local model, for example via Ollama or vLLM – then your data stays in-house.
How to get started
- In a 45-minute demo, we show you suitable examples.
- Together we define a first use case, the knowledge sources needed and the operating concept.
- We choose the models – local, external or both – to suit the task and the hardware.
- You receive a quote for the software and any support you want.
Frequently asked questions
Does our data stay within the company?+
DENKRAUM runs on your infrastructure. If you use local models and connect no external services, the AI processing can also take place within your own network. With cloud models or external integrations, the data flows you set up and the providers’ terms apply.
Do we need our own server?+
Not necessarily. The desktop app for Windows runs on a computer in the office; with “Share on the network”, your team works along in the browser. On the go, phones reach DENKRAUM AI through the office Wi-Fi, a VPN or your own server with HTTPS.
What hardware do local models need?+
That depends on the model and the use case. We assess hardware requirements, quality and running costs together during setup.
More topics
See DENKRAUM AI with your own example
In a 45-minute demo, we show you suitable examples. Then we define a first use case together, along with the knowledge sources needed and the operating concept.