Many European companies have chosen Langdock, the enterprise AI platform from Berlin. So have UK and US subsidiaries of German, Austrian, and Swiss groups, which often inherit the platform from headquarters. The reasons are good: data protection, EU deployment, and the freedom to use models from OpenAI, Anthropic, Google, Mistral, and others instead of tying yourself to one vendor.
But the license alone does not change how people work. A few months after rollout, a handful of people chat every day and everyone else rarely. There are hardly any agents, no workflows, and the integrations are not connected. What helps then is not another product demo but training built on the work your teams do every week.
Langdock training for your teams
Langdock training from Specialty Tokens runs inside your own workspace. Specialty Tokens is independent of Langdock: not a partner, no license sales. Beforehand we agree which departments take part, which models are approved, and which data may go into which agents. The table shows the five building blocks Langdock names itself and what we do with each.
| Building block | What it is, per Langdock | What the training does with it | What we set up beforehand |
|---|---|---|---|
| Chat | Model choice (or Auto mode), web search, company knowledge, Deep Research, files | Pick the right model, work with your own documents, check results | Approved models, data rules, accounts |
| Agents | Instructions, knowledge, model, actions; shareable with groups | One or two agents per department, built and shared | A group per department, knowledge folders, who maintains each agent |
| Workflows | Visual editor: triggers, agent, condition and action steps, user input for approval | Automate one recurring process of the department | Workflows enabled, allowed step types (Code, HTTP Request) |
| Integrations | Built-in integrations, MCP servers, Integration Builder | Agents read and act in file storage, wiki, or CRM | Which integrations are allowed, who connects what |
| API | Connects agents and workflows to your own applications | Only for IT and developers | A workspace API key from an admin |
For systems without a ready-made integration, we build a custom integration with you, through MCP or the Integration Builder.
How to build Langdock agents and workflows for a department
- Pick a task that comes back every week: answering requests, preparing quotes, summarizing reports.
- Connect accounts: in Langdock, a connection belongs to the user who created it. So each participant connects their own accounts in the first session.
- Build the agent: write instructions, attach knowledge (according to Langdock up to 50 files plus synced folders, for example from SharePoint or Confluence Cloud), choose actions, share it with the group.
- Turn it into a workflow: what repeats gets a trigger (form, schedule, webhook, or an integration event) and an approval step before anything goes out.
- Assign upkeep: who keeps the agent current, and who checks after four weeks what is still running?
Langdock reviews: what makes a rollout succeed
People reading Langdock reviews usually want to know whether the rollout will pay off. From our work with AI platforms inside companies, the rollouts that work share the same patterns:
- One use case per department to start, not twenty.
- Agents built by the people who know the task, with one person who maintains each.
- Integrations approved before training, so nobody pastes content into a chat.
- Leaders who use it themselves, not just approve the license.
Langdock's own rollout playbook agrees: focus on one or two use cases first, find one or two champions per department, and don't overengineer the rollout.
TimeTac shows what this can look like: after a workshop and with an internal AI platform, its marketing and growth team had 10 automations live, with 117 runs in three weeks, 54 of them unattended. There are no engineers on the team.
Which Langdock training fits you?
Specialty Tokens is one of the providers of Langdock training and wrote this page. An honest overview:
- Langdock's own resources: the rollout playbook and docs are free and good. Fit when an internal team runs the rollout itself. Don't fit when nobody has the time to turn them into sessions.
- Certified Langdock partners, listed in Langdock's partner directory, know the product first-hand. Fit when you want rollout and training closely aligned with Langdock. Fit less when you want an outside view on the choice of tool.
- Open seminars and online courses with fixed dates and prices suit individuals, such as your future admin. Fit less when whole departments should practice in your own workspace.
- Specialty Tokens trains department by department in your workspace, on site in Europe, the UK, and the US, in English or German, with integrations connected beforehand. Fits when several teams start and integrations matter. Doesn't fit when you only want to send one person to an open course.
Also check which onboarding your Langdock contract already includes.
Still choosing a tool? See the comparison with meinGPT, companyGPT, and others in Langdock alternatives. If your working day happens in Outlook and Teams, read Microsoft Copilot training; for OpenAI, ChatGPT training; for long documents and writing, Claude training.
Everything built in the training is yours: the agents, the instructions, the workflows. No per-seat fees to us and no dependency. This training is part of our AI training for employees. Talk to us about your departments and systems.
