From a single assistant to an operating system of collaborating assistants that drives your recurring work processes. Not a loose chatbot, but your platform, ready to deploy in your organisation. You choose the platform that fits you: Dust (low-code) or Claude Code (code-native).
Not a trial model, but the start of your own operating system you can deploy directly in your organisation.
A system of collaborating assistants that uses your knowledge and executes real actions, on the platform that belongs to your organisation.
Not just giving answers, but also processing documents, looking up information and executing actions based on your instructions and data.
With Dust you take your working assistants in your own workspace. With Claude Code you take a code repository you fully own and can run anywhere.
Who is this workshop for?
You already work with AI assistants and now want to build a system that drives multiple processes at once.
Whether you prefer building in the browser (Dust) or working in real code (Claude Code): both paths lead to the same result.
Departments that are done with loose tools and now want a coherent AI platform that belongs to their organisation.
No basics yet? Start with Build your own AI agent (beginners) → or first build an assistant via AI Personal Assistant →
So you can keep building independently in your own environment after the workshop
With Dust: your working assistants in your own workspace. With Claude Code: a real code repository you fully own and can run anywhere.
The most important steps, patterns and settings per platform, so you can quickly pick it up again later.
The key points and framing per block, to read back at your own pace after the session.
A selection of tools, documentation and next steps to keep building your AI Operating System on your own.
The expert concepts behind a working AI Operating System, explained in a platform-neutral way and applied immediately in your own variant.
The leap from one assistant to a network of collaborating assistants that drives your recurring processes.
How to connect an assistant to your own documents, databases or sources so it answers based on your data.
An assistant that not only gives answers but also does something: retrieve information, make a calculation or send a message.
The difference between steps you can fully automate and steps where you build in an extra check, especially for financial actions or irreversible steps.
With Dust: save a version of your assistant before a risky step. With Claude Code: git as a version control system, rollback is always possible.
How multiple assistants hand off tasks to each other, so a main assistant can delegate work to specialists.
Built modularly: short framing, then keep building. Anthonie sets the context per block, Stefan leads the build on both platforms.
For those who want to automate without code. You build in the browser and take working assistants in your own Dust workspace.
For those who want to write real, fully owned code (skills, subagents, MCP tools). You take a code repository that runs anywhere.
Anthonie sketches the leap from one assistant to a system of collaborating assistants that drives your work, and links the platform choice (Dust vs Claude Code) to your organisation and tech landscape.
You build one assistant with a clear role, clear instructions and the right boundaries. With Dust: set role, tone and boundaries and test right away. With Claude Code: write a skill or subagent, review the result and commit. Goal: get the build rhythm into your fingers.
The assistant gets access to your knowledge and executes a real action. You also learn when to trust the AI for a fixed step, and when to build in an extra check for financial actions or irreversible steps. With Dust: connect connections and actions. With Claude Code: add MCP tools or scripts.
Done with Block 1 and 2? Have multiple assistants collaborate. With Dust: assistants call each other and hand off work. With Claude Code: a main agent that delegates subtasks to specialists via subagents. Room to keep building at your own pace.
Block 3 is optional and intended for fast workers. Stefan walks around so no one gets stuck.
A clear division of roles: Anthonie sets the context per block, Stefan leads the build.
Background in Management, Economics & Law (Hanze), former COO at Ancora Health and an entrepreneur since 2010. Builds AI into existing organisations and sets the context per module: why you build this and where it lands.
Background in Human Technology (Hanze). Builds AI agents and AI infrastructure daily, and guides participants step by step as they build their own system.
Choose Dust if you want to automate without code and see results quickly: low-code, browser-based and works alongside Microsoft or Google. Choose Claude Code if you already work in a Claude or terminal environment and want to write real, fully owned code that runs anywhere. The workshop content is identical; only the platform differs. Not sure? During the intake Anthonie and Stefan help you choose.
This is an expert workshop. The medium workshops (AI Personal Assistant or AI Automation) are recommended as preparation. You already know the basics of AI assistants and now want to build a complete system of collaborating assistants. Absolute beginners are advised to start with the beginners workshop first.
No. This is the expert rung of the workshop ladder. You are already familiar with AI assistants and understand the basic concepts. Want to start from the basics? Start with Build your own AI agent (beginners).
Your own laptop and 2 to 3 work processes you want to automate. We provide the working environment for the session: for Dust a configured workspace with credits and connections, for Claude Code a cloud dev environment per participant with Claude Code authenticated and a token budget monitored.
We work with 8 to 20 participants. Expert hands-on building needs personal guidance so everyone stays on pace and no one gets stuck.
The workshop starts from €149 per person (open enrolment) or from €2,500 excl. VAT for an in-company session. The exact price depends on group size. Request a quote for a tailored proposal.
That differs per variant. For Dust: a Dust workspace with topped-up credits and connections, prepared the day before (indexing may take time), participants added as members. For Claude Code: a cloud dev environment (Codespace) per participant, Claude Code authenticated with API access and a token budget monitored, sample data and MCP tools ready. We provide the environment. If you want to work with your own production credentials, we align on that during the intake.
Yes. With Dust you take your working assistants in your own Dust workspace: they just keep running. With Claude Code you take a real code repository you fully own and can run anywhere. You keep building independently after the session, with the cheatsheet, summary and list of sources as a reference.
Request a no-obligation quote for a tailored in-company session.
