--- title: "Introducing and supporting Langdock" lang: en type: service canonical: https://www.arocom.de/en/services/langdock alternate: https://www.arocom.de/leistungen/langdock last_updated: 2026-09-09 --- # Introducing and supporting Langdock Langdock is an AI platform for organisations: one shared, secured way into the large AI models, connected to the organisation's own documents. On top of that you can set up assistants that take on recurring work, such as reviewing a tender or sorting the responses to a survey. arocom checks with you whether it is worth it, sets the platform up, builds the first assistants together with your departments and trains your people. We do not sell licences, so our recommendation may equally come out against Langdock. ## Introducing and supporting Langdock Many employees already use an AI chat at work, usually through a private account. Langdock is the company version of that: one access point for the whole organisation, with a contract under German law, connected to your own documents, and with an administration where you decide who may do what. arocom sets it up, builds the first assistants together with your departments and stays your point of contact afterwards. - [Request an initial call](https://www.arocom.de/en/services/langdock#kontakt) ## Why organisations are taking this on now In many organisations people have long been working with AI, just privately and unnoticed: through free accounts, into which documents also wander that have no business being there. Four in ten companies in Germany assume this is happening in their own house. A ban changes little, because the reason remains: the work gets done faster that way. More on this in the article on [shadow AI](https://www.arocom.de/en/knowledge/ai/shadow-ai). An official access point solves this from two sides. Usage runs through your central sign-on, which makes it contractually covered and traceable. And it becomes more useful, because unlike a private account a platform reaches the organisation's documents and works inside your applications. **42%** of companies assume private AI accounts are in use in-house ([Bitkom, 2025](https://www.bitkom.org/Presse/Presseinformation/Beschaeftigte-nutzen-Schatten-KI)) ### Assistants belong to the organisation Set up once, usable by everyone. An assistant stays even when the person who built it leaves. ### No single vendor everything hangs on Several model vendors behind one interface. Which model does the work is your decision, at any time and by hand. ## What an assistant takes on An assistant is a fixed role in the platform that you set up once and that is then available to everyone. It gets an instruction, the documents it should know and, where needed, access to your applications. It is set up by the department that knows the task, and nobody has to program anything for it. Three examples we often start with. The quantities are rules of thumb from experience; what comes up in your organisation we count together beforehand. Every result is approved by a person. **Three typical use cases** | Who works with it | Where the work starts today | What the assistant delivers | How often | | --- | --- | --- | --- | | Sales | a tender of 40 to 80 pages | every requirement and deadline as a list, with the source reference, plus an outline for the bid | 2 to 4 a month | | Support team assistant | a customer question much like many before it | a draft reply from your manual, with the source reference, for approval | 20 to 60 a week | | HR | free-text answers from a staff survey | the themes sorted and grouped, with sample quotes | 60 answers in one pass | ## Who it pays off for We work with organisations from around ten people upwards: law firms, associations, social care providers, mid-sized businesses, public authorities. What they have in common is less their size than their situation: nobody in the organisation has the job of dealing with AI, and at the same time there is work that repeats every month. No IT team of your own We take on the administration or train someone at your end for it. Where there is an IT lead, we work to them. Training alongside daily work Volunteers and shift workers are rarely available at the same time, so there are evening sessions with a recording. Changing staff Assistants and knowledge folders belong to the organisation, and we document the handover to a successor. Board approval and co-determination For the works council, staff council or employee representation we supply the paperwork, as context, not as legal advice. ## How we go about it We work in four steps, and the decision only falls at the third. Anyone who goes through the [ten questions before a rollout](https://www.arocom.de/en/ai-readiness) beforehand and brings them along filled in shortens the first step. 1. Initial call 45 minutes, free of charge, by phone if you like. Afterwards you know whether the next step is worth it for you. 2. Trial in your organisation We set up a trial workspace, hand over real tasks and record which of them held up. 3. Decision One page for the management, board or head of authority: benefit, annual cost, risks, alternatives. 4. Rollout Only after the decision. We build the first assistants, and from the second onwards the departments build along. **Who takes on which part** | arocom takes this on | Langdock takes this on | | --- | --- | | Selection, trial, decision paper | Platform, data centre, availability | | Setup, permissions, the first assistants and knowledge folders | The models of several vendors and switching between them | | Training, refresher training, tidying up what has accumulated | Product support and documentation | | Input for a works or service agreement | One contract, one DPA, one invoice, German law | ## What this costs Three items, and only two of them are invoices from us. You pay the licence directly to Langdock. On monthly billing the list price is EUR 25.00 per person and month, and on annual billing Langdock grants a 20 per cent discount. For an organisation with 25 seats that comes to around EUR 6,000 a year. Not everyone in the organisation needs a seat of their own, though. It pays off where texts, enquiries or analyses come up regularly, say in sales, in support or in HR. Anyone who only wants to rephrase a sentence now and then is better served by a simple single subscription, and the two run side by side without any trouble. What else the vendor charges for is set out in [What is Langdock?](https://www.arocom.de/en/knowledge/ai/langdock). The rollout we bill by time spent. How much that is depends above all on how many assistants should be in place at the start, how well ordered your documents are, and whether we connect a line-of-business system. You get a concrete figure after the initial call and before any commitment. The support afterwards is optional. What comes up regularly we settle together after the first few months, based on actual use. **6,000** euros of licence cost a year, at 25 seats paid annually ([langdock.com, retrieved 7 August 2026](https://langdock.com/pricing)) ### Free of charge up to the decision The initial call, the trial and the decision paper cost you nothing. Only the rollout is billed. Book a meeting: Pick a slot straight away (on the page: https://www.arocom.de/en/services/langdock) ## What we do once the rollout is over A platform nobody looks after usually looks the same after six months: thirty half-finished assistants, two contacts have left the organisation, and nobody knows any more which knowledge folder is still being maintained. So that it does not come to that, you have a named contact with us who knows what you have. They tidy up regularly, train new staff and, when models or terms change, review the model selection and the seats. If the person looking after it on your side changes, we hand over to their successor with documentation. ## Why arocom, without a partner badge We are not a Langdock reseller, and that has an upside for you: we earn nothing on the licence, so our recommendation may equally come out against Langdock. What we bring is our own practice. We work with AI agents ourselves every day, in development, in content work and in our own analyses, and so we know which use cases hold up and which nobody touches after three weeks. What we dropped again is in the [workshop report](https://www.arocom.de/en/blog/ai-in-agency-work). On top of that, we have spent years bringing systems into the daily work of organisations without a large IT team, for instance the workflows at [Fohhn](https://www.arocom.de/en/references/fohhn) and the job import at [GOPA](https://www.arocom.de/en/references/gopa). That is where rollouts fail, not on the technology. One distinction, because the terms get muddled easily: this is about a platform next to your website. AI inside the website itself we build as [AI in the Drupal editorial process](https://www.arocom.de/en/drupal-ai) or as [AI integration](https://www.arocom.de/en/services/ai-integration). ## When Langdock is not the right fit Below around ten seats it does not pay off, because the administrative problem a platform solves does not yet exist at five people. Anyone who requires every processing step to happen in Germany will not be happy, since some of the models compute outside the EU. Which ones those are and how to switch them off, we clarify beforehand. And the platform does not replace automation software. Anyone wanting to connect line-of-business systems permanently needs a tool such as [n8n](https://www.arocom.de/en/knowledge/ai/n8n) alongside it. The evidence for all of this is in [What is Langdock?](https://www.arocom.de/en/knowledge/ai/langdock). Download: AI platform review guide (on the page: https://www.arocom.de/en/services/langdock) 24 questions to put to any vendor, with a findings sheet. Vendor-neutral and useful without us. ### Does our data stay in Germany? No, and anyone promising that has not read the terms of use. What is promised is storage in the EU, encryption, no training on your content, ISO 27001 and SOC 2 Type II. We list beforehand which model computes where, and switch off the ones without an EU region. ### What if we want to get out again? A monthly contract can be cancelled up to seven days before the end of the billing period, an annual one up to 14 days before it runs out. Your documents and prompts are yours and can be taken with you; the assistants themselves stay tied to the platform. ### What happens if a model vendor goes down? Then you switch to a different model yourself, in chat for the conversation at hand and for assistants via the administration, both in seconds. Nothing happens automatically: Langdock does not promise an automatic switch when a vendor has an outage. ### Who sets up the assistants, us or you? The first two or three we set up, so that they run cleanly from the start, and after that the departments take over. Before an assistant gets write access to a line-of-business system, we take another look. ## What working with us is like These quotes come from web and platform projects; for Langdock rollouts we have not been at it long enough. They show how we bring a system into everyday work. > Dear arocoms, dear 21TORRs, everything has been said. Thank you for the outstanding job. > > — Dirk Danker, Head of Marketing, Corporate Communications & Fundraising > We thank you for your reliable and well-structured work. We would like to particularly highlight your patience and professional demeanor. Your feedback and assessments were of extraordinary value for the further decisions in the Leando project. We can make an unreserved recommendation for arocom GmbH, and we are happy to do so publicly. > > — Melanie Schütt, Public-Sector Client > After rather negative experiences with other agencies, we switched to arocom this year and are fully satisfied with the services provided. arocom impresses with good communication, transparency and speed. We are happy to recommend them. > > — Jan Gebauer, Client ## The whole topic, in reading order This page explains what we do. The questions behind it are answered by six further pages, exactly one each, here in the order in which they are best read. ## Pages on this topic - [What is Langdock?](https://www.arocom.de/en/knowledge/ai/langdock): What the product does, what it costs, where it hits limits - [Shadow AI](https://www.arocom.de/en/knowledge/ai/shadow-ai): Why private AI accounts in the organisation are a problem - [AI platform: the six building blocks](https://www.arocom.de/en/knowledge/ai/ai-platform): What makes an AI platform, and how to compare vendors - [Six decisions before an AI platform](https://www.arocom.de/en/blog/rolling-out-langdock-smaller-organisations): What you need to settle before you start - [Are we ready for AI?](https://www.arocom.de/en/ai-readiness): Ten questions to print out, for your next meeting - [One contract instead of four](https://www.arocom.de/en/blog/one-contract-instead-of-four): How to recognise a contract that holds up ### Talk to us about Langdock Quickest by phone: +49 711 633 77 960, weekdays 9am to 5pm. The initial call costs nothing. Form: Request an initial call (on the page: https://www.arocom.de/en/services/langdock)