AI Training & Workshops for Professional Teams
I’m an AI engineer, and alongside the systems I build I run practical AI training for professional teams: hands-on sessions designed around the work the participants actually do, from confidential use of AI and local models through to Claude, ChatGPT, MCP connectors and the workflows that make them repeatable. People leave with tools they use the next morning, not a folder of slides.
Workshops and teaching delivered to law firms, accounting professionals, business leaders and university audiences.
What the workshops cover
Four areas, in the order teams usually need them. Each session draws on them in the proportion that fits the room.
Foundations and everyday workflows
Using Claude and ChatGPT properly: what the interfaces actually offer, which tool or model suits which task, and how Projects, custom instructions and reusable context let an assistant start each task already knowing the work. Then the output professionals need, from documents and analyses to presentation decks, and a method for telling which recurring tasks in the business are worth handing to AI in the first place.
Confidential and responsible use
What can reasonably be sent to an external AI provider and what cannot, how to anonymise or pseudonymise material before it leaves the organisation, and how your own professional secrecy obligations and internal AI policy bear on that decision. When a cloud tool is the right choice, and when a local model is better: we install one with Ollama or LM Studio and see first-hand what it can and cannot do. Running a model locally changes where the data goes; it does not by itself make a use case compliant. I’m an engineer, not a lawyer, so the training gives your team the technical reasoning and the questions to take to counsel or to your DPO.
Beyond the chat window
Connecting assistants to the tools and information a team already works with, through MCP connectors, including services such as email where that is appropriate. Reusable capabilities, Skills, recurring automated tasks, and the persistent instructions that turn a good result into a repeatable one. Taught for professionals rather than developers: it is configuration and judgement, not code.
Applied to your team’s own work
We work on the participants’ real tasks, with their own tools and constraints, and build the two or three workflows that will still matter once I leave. No two workshops are the same: the use cases worth having in a law firm, an accounting practice and a research team have almost nothing in common.
How I design a session
I prepare each workshop for the people who will be in the room. It usually goes like this:
- A scoping call: who is in the room, what they already use, how their work flows, and what they are not allowed to share.
- A shortlist of the use cases where AI would genuinely save them time, agreed with you rather than assumed.
- A session built on that shortlist, with hands-on exercises on real material instead of generic demonstrations.
- Technical depth adjusted as we go, so the confident participants keep moving and nobody is left behind.
- A written recap of the workflows, instructions and settings we built together, so the team can keep using and extending them.
Why me
I teach what I build. My daily work is engineering AI systems for research and regulated environments, including on-premise and air-gapped deployments at LocusLab where data cannot leave the client’s infrastructure, and freelance AI engineering for biology and TechBio teams. That is what the sessions carry back into the room: where these tools are genuinely strong, where they fail quietly, and what that means for work that has to be right.
Who I work with
The sessions adapt to very different profiles: independent professionals, SME leadership teams, consulting and professional-services firms, specialist practices, and university audiences. What changes between them is the pace, the examples and the technical depth, never how hands-on the session is.
Format
Half-day or full-day sessions, on site or remote, in French or English. Longer programmes run as several sessions, spaced so participants can put each one into practice before the next. None of that is fixed: a short call beforehand scopes the content, the depth and the format to your team’s tools, constraints and hardware.
Common questions
Do participants need a technical background?
No. The sessions are built for professionals, not developers. Everything hands-on, including running a model on your own machine, is done step by step together.
Can you work within our confidentiality constraints?
Yes, and it is usually the first thing we scope. Your constraints decide which tools are on the table, which material can be worked on live, and whether we run anything locally, so they belong in the design of the session rather than in a disclaimer at the end of it.
How does an engagement start?
A 30-minute call. If the fit is there, I send a concrete proposal within a few days: content, format, dates. If it isn’t, I’ll say so directly.