Mitarbot or Langdock?
Both come from Europe and both want to make teams less dependent on US tools. The difference is the shape: Langdock is a chat front end over many models, Mitarbot is a workspace with its own file system, code and tools.
Langdock is an EU platform that bundles many models in one interface, with SSO, roles and enterprise features. If you mainly want a privacy-friendly chat for many employees, it is a sensible choice.
Mitarbot goes a layer deeper: every workspace is its own Linux machine with files, connectors and schedules. The agent runs real code and builds mini-apps your team uses in the browser — instead of only answering.
Head to head
| Mitarbot | Langdock | |
|---|---|---|
| Pricing model | No subscription — only tokens and running time; an idle month costs €0 | €25 / user / month, billed in idle months too |
| Unit of organization | Workspace — files, memory, connectors, mini-apps and budget per workspace | Org-wide assistant + workflows |
| Data stored in the EU | Yes — Germany | Yes — EU |
| AI inference in the EU | Yes — TensorX, Dublin & Helsinki | Platform EU, models usually US |
| No US company in the AI chain | Yes — open-weight models, EU inference, European GPUs | No — OpenAI / Anthropic / Google models |
| Zero retention / no training | Yes — contractual, provider-level ZDR | Yes — provider-level |
| Builds mini-apps / tools | Yes — real web apps, served from the workspace | No |
| Scheduled jobs | Yes — cron jobs with run history | Workflow triggers |
| Your own data sources | Drive, FTP/SFTP, email, HTTP, GitHub, MCP | SaaS connectors |
| Runs real code safely | Yes — each workspace is its own isolated machine | No |
| Metered usage billing | Yes — tokens + awake compute | No — seats + credits |
| Enterprise controls (SSO, SCIM, audit, roles) | No — workspace-first, no org layer (yet) | Yes — enterprise tier |
Prices and product facts from public information as of October 2026. Competitors change fast — verify before buying.
Where they are ahead
Where Langdock is ahead: more model choice in one interface, mature enterprise features (SSO, roles, org administration) and a broader SaaS connector catalogue. For large companies with a central IT department, that is often the shorter path.
Common questions
Does Langdock use US models?
Yes, typically OpenAI, Anthropic or Google models under European contracts. The platform is European; the models come from the US. Mitarbot runs open-weight models on European inference infrastructure under zero-retention terms.
Can Mitarbot chat too?
Yes — chat is the entry point. The difference is that the agent reads real files, runs code and leaves results behind as files or mini-apps.
What about SSO and roles?
Mitarbot does not have them yet. Workspaces have members and roles (owner, member) but no org layer with SSO, SCIM or audit logs. That is openly on the roadmap.
Mitarbot or Langdock — Mitarbot
Both come from Europe and both want to make teams less dependent on US tools. The difference is the shape: Langdock is a chat front end over many models, Mitarbot is a workspace with its own file system, code and tools.