AI consultant in Europe: Claude rollouts, training and managed AI

An AI consultant in Europe gets the right AI tool into your business, trains your team to use it on their own work and stays long enough for the habit to stick, all inside the GDPR and the EU AI Act. With us that means rolling out Claude, the AI assistant from Anthropic, training teams on their own documents and running the platform afterwards. What we skip is the 60-page strategy document and the hourly invoice that doubles quietly.

With us the price is fixed and published in euros, whichever EU country your team sits in. The full launch costs €560 per user as a one-time fee from 8 users (€4,480), training on its own runs from €280 to €560 per user, and managed Claude costs €490 for setup plus €14 per user per month with the licence on top, all excluding VAT. A team of 10 doing the full launch pays €5,600 once, whatever the number of questions on the day.

The GDPR puts the obligations on you as the controller, so the useful question is whether the contract and the set-up let you meet them. Anthropic says it does not use inputs or outputs from its commercial products to train its models by default, and its Data Processing Addendum with the EU Standard Contractual Clauses is built into its Commercial Terms. On top of that we set permissions to match how you classify data, write down what may go into a prompt and can keep processing inside the EU through EU cross-region inference on Amazon Bedrock or Google Cloud Vertex AI in EU regions.

The AI literacy obligation has applied since 2 February 2025, according to the European Commission, and since the 2026 amendment it asks businesses to take measures that support AI literacy among their staff, which our training covers in practice. Most of the Act has applied since 2 August 2026, while the rules for high-risk systems in areas such as employment and education follow on 2 December 2027. We classify each planned use before we build it, and we flag anything that touches a high-risk area so a lawyer can look at it first.

Yes, for teams that need it. Amazon Bedrock offers Claude through EU inference profiles you can call from Frankfurt (eu-central-1), and AWS documents per model which Regions a request may be routed to, so we check that list before we promise anything. Google Cloud Vertex AI offers Claude through European regional endpoints and an EU multi-region endpoint, and both routes can mean a narrower feature set than the standard Claude apps, which we walk you through before you choose.

Remote delivery is the default across the EU, and it works for webinars, workshops and the build work itself, because we design the sessions as workshops and not as lectures. When a full day in the same room is worth it, we agree the date and any travel cost in writing before we book. The method, the exercises and the fixed prices stay the same in Madrid, Vienna or Copenhagen.

Depth beats breadth when someone trusts you with the working habits of their whole team. Claude from Anthropic is the platform we know inside out, with commercial terms that keep your data out of model training by default. For teams that already run ChatGPT or want a European model provider, we run satori-chatgpt and satori-mistral on the same fixed-price terms.

The first step is a free 30-minute call, usually within the same week you ask for it. A typical launch then runs over four to six weeks: the webinar first, the full-day workshop next, and the first five skills going live one by one while the questions are still fresh. Most teams use Claude on real work within a fortnight of the workshop.

It depends on team size and appetite for change, and we will say plainly which one fits instead of pushing the bigger package by default. A pilot with one team suits a business that wants results before it commits budget for everyone, and it usually runs two to four weeks. satori-launch is built for the other case, where the whole team starts on the same day and nobody ends up as the only team that ever learned the tool.

ROI, the return on the time and money you put in, is tied to three numbers we agree before the first workshop: hours saved per person each week, how many of the five skills are still in daily use a month later, and how quickly a new starter catches up. We check back at 30 and 90 days to put a figure against each one. A skill that saves nobody time by day 90 gets rebuilt or removed, and we tell you which.

A consultant fits when you want the team trained and the platform running within weeks, with someone who has already made the early mistakes. An agency fits when you want a finished build handed over and run for you, which is why we offer that too as our AI agency service. Building in-house works when someone on your team has the time and the technical grounding, and most businesses we speak to do not have that person free for the six to eight weeks a proper rollout takes.

Yes, through MCP, short for Model Context Protocol, the connector standard that lets Claude read from and write to your tools instead of someone copying data across by hand. Martin builds the MCP servers behind our managed platform, and satori-mcp covers custom connector work when your stack needs something new. Nothing gets ripped out, because the point is to put Claude on top of what already works.