Comparison
Both European — different answers
Langdock and Bee Flow answer the same question — how does a European team use AI without mailing its data across the Atlantic — from opposite ends. Langdock is a polished hosted platform out of Berlin that advertises ISO 27001 and SOC 2 certification; Bee Flow is a Dutch workspace you can run on your own hardware and read line by line. Which answer you need depends on what your organisation has to be able to prove, and this page tries to make that decision faster rather than tilt it.
Run it yourself All comparisonsCertification, polish and adoption. Langdock advertises ISO 27001 and SOC 2 — audits of the company, which Bee Flow has not commissioned — and a hosted product with one deployment shape is easier to keep polished than a stack every customer runs differently. If your procurement checklist starts with a certificate and ends with a signed DPA, they clear it faster today. Their public customer list is longer than ours, and pretending otherwise would cost this page its credibility in the first paragraph.
A certificate says an auditor reviewed the vendor’s processes. It does not say where a specific prompt went — only the operator of the infrastructure can answer that, and with a hosted product the operator is not you. Self-hosting moves you into that seat: the code is source-available, the PII detector runs on your own CPU, and the Compliance Center runs GDPR, EU AI Act and ISO 27001 checks against your live install. Instead of inheriting a vendor’s certificate, you build evidence about your own deployment — which, for some organisations, is the thing the auditor actually wants.
What we bring
One command stands up the whole workspace from public images — no licence key, no account, no phone-home. Compose profiles let you start small, Kubernetes manifests are documented, and licences stay valid offline, which is what makes an air-gapped install with local models possible. There is no user, agent, message or knowledge-base cap in any tier, including the free Community tier, and all built-in integrations ship in it.
Once the detector is installed, personal data is found across 21 categories — with a Dutch fine-tune that validates a BSN by its checksum — and tokenised before any prompt reaches a model provider, then restored in the reply. Detection runs on your own CPU; no text goes to a third-party detection API, and an unreachable detector fails closed.
OAuth SSO with Google, Microsoft Entra ID and Nextcloud, group sync from Entra ID, six built-in roles and an access audit log built for Article 30 evidence. The identity page spells out the whole chain, boundaries included.
Assistants share the building with automations — 27 step types, six trigger kinds, approval gates and dry runs — plus knowledge bases that cite the passage they used, meeting notes with speaker identification, and internal apps from App Studio. One permission model and one audit trail run under all of it, so the answer to "who did what, with which data" is one query rather than a reconciliation project.
Side by side
Their column is drawn from public documentation. Treat their docs as authoritative — packaging changes faster than comparison pages do.
| Bee Flow | Langdock | |
|---|---|---|
| Deployment | Self-hosted — Docker Compose or Kubernetes — or hosted by us on EU infrastructure | Hosted platform; their documentation lists the current deployment options |
| Company certification | None yet, and we say so — the product ships compliance tooling instead | Advertises ISO 27001 and SOC 2 certification |
| Source access | Source-available under the Sustainable Use License — auditable line by line | Proprietary |
| PII protection | Detector on your own CPU, 21 categories, tokenise-and-restore, fails closed | Platform DLP features, per their documentation |
| Model choice | Six provider adapters plus any OpenAI-compatible endpoint, local models included | Multiple hosted model providers |
| Beyond chat | Automations, meeting notes, internal apps, knowledge bases — one permission model | Chat, assistants, workflows and integrations |
Questions
No. Langdock advertises certification and we do not, and pretending a product module equals a company audit would be exactly the kind of claim this site refuses to make. What ships instead is a Compliance Center that runs 15 GDPR, 6 EU AI Act and 23 ISO 27001 checks against your live configuration, plus an ISMS layer — Statement of Applicability, risk register, internal audits, evidence bundles. That is tooling for certifying your own installation, which is a thing no vendor certificate covers.
Their documentation is the authority on current deployment options — check it there rather than trusting a competitor’s page to stay up to date. What we can say about our own side: self-hosting at any scale is a supported, documented, one-command path on the free tier. If running it yourself is a hard requirement, that difference settles the evaluation quickly.
A hosted product with one shape is easier to polish than a self-hostable stack, and Langdock has clearly invested in theirs. Where we spend that effort instead is inspectability: source you can read, audit trails you own, a sovereignty report with real destination countries. Try both — polish is precisely the property you can judge in an afternoon.
The two products count differently, so compare capabilities rather than catalogue lengths. We ship 44 built-in integrations — Google Workspace, Microsoft 365, Nextcloud, ITSM, payroll, Dutch legal sources — and then the ceiling comes off, because Bee Flow speaks the Model Context Protocol. Any MCP server becomes a set of tools your assistants and automations can call, over stdio or Streamable HTTP: the public ones, the ones your own vendors publish, and the one you write in an afternoon for the line-of-business system nobody has ever built a connector for. When you self-host, installing those is your decision and nobody has to add your system to a catalogue first — the install surface is an enterprise feature, and once a server is installed, using it is an ordinary integration with the same permissions and audit trail as the rest. Where a large catalogue genuinely wins is the long tail of SaaS you want working in ten minutes without writing anything, so check the specific apps you depend on before you switch.
Source-available, and we would rather be precise than borrow the word. The server and frontend are published under the Sustainable Use License v1.0: read it, modify it, run it internally without paying — what you may not do is resell Bee Flow as your own hosted service. It is not an OSI-approved open-source licence. We are working toward AGPL-3.0-or-later; today the Nextcloud connector is the only part that carries it.
Stand Bee Flow up on your own hardware next to a Langdock trial and run the same week of work through both. Which one fits will be obvious faster than any feature grid can make it.
Self-host it All comparisons