EN FR
Editorial illustration of a workshop prep agent — a notebook at the center of a ring of email envelopes connected by curved lines

An agent that preps the room before I walk in

I run AI enablement workshops, and I always wanted to know what the people in the room needed from AI before I walked in. So I built an agent the participants email before the workshop — it collects what they want AI to do, explains what AI can and can't do for their task, coaches them through getting it working, and hands me a briefing built out of real, named pain points the participants already own.

The conference organizer wants to know what people in the room want AI to do for them. Participants email the agent with their task. The agent collects and triages every request for the organizer — but it also responds to each person, explaining what AI is good at in their task and where it falls short.

The agent reads a participant's task and responds with an honest assessment — AI is strong at these parts, not so good at others. No hype, no "AI can do everything."

Stylized email from the workshop prep agent to a consulting firm owner asking about AI for financial management, breaking down what AI can handle well and where it falls short, with blue and amber highlighted sections

When someone's task is something AI can help with, the agent offers to coach them through getting it working. Three exchanges later, the skeptic walks into the room already winning.

Two stylized emails side by side — on the left the agent coaching a participant through a specific AI task, on the right the same participant reporting a working result, framed in amber

A week later, the agent opens the next conversation like a colleague who was paying attention — no re-explaining, no "as a language model I don't remember."

Two stylized emails a week apart — the second one opens with an amber-highlighted reference back to what Marie said in the first

Morning of the workshop, I get a briefing built out of real, named pain points — not a generic agenda that could have been written last year.

Stylized facilitator briefing document with four numbered findings about what participants are stuck on, and an amber-highlighted recommendation for how to spend the first half hour

Before the first email goes out, the agent researches the organization's document types and decides ahead of time how each should be handled — data privacy under Law 25, corporate policy on what can and can't go to a cloud AI. When a participant asks about something that touches either one, the agent already knows the answer.

Stylized email where the agent answers a question about uploading documents and includes an amber-highlighted section covering both Law 25 data privacy and corporate policy considerations

The reusable part is the pattern — an agent that learns in the spaces between formal interactions, and hands its learning to the facilitator instead of hiding it in a dashboard nobody checks. The architecture is designed so a new client gets their own agent, their own inbox, their own branding, and their own tuned knowledge base. Quebec French or English, bilingual if that's what you need.

If you run workshops, offsites, or training sessions and you want the agenda to write itself out of what your people are actually stuck on, drop me a line. I'll show you what the agent already knows about a problem you bring to it.

Let's talk →

Related projects