Build your own AI team.
No code. Job descriptions.

A skill is a job description. An agent is an employee.

Building AI agents sounds like an engineering project. It isn't. If you can write down what a good assistant would do for you, in plain English, you can build one.

Enquire about the workshop See how it works

Stop thinking about tools.
Think about the team you need.

If you could hire three people tomorrow, what would they do? Write that down and the hard part is done.

01

Roles, not prompts

A prompt is a favour. A job description is an employee: same job, every time.

02

Most jobs can run on the clock

Morning briefs, weekly reports, monthly packs. Once a role is written, it runs on a schedule.

03

The human gate

Nothing goes out without a person approving it. The AI team drafts. You decide.

This is my actual team.
It runs my business.

I run my training business on one AI assistant and a team of written job descriptions: a training studio, a client desk, a content house and a back office. Five jobs run on schedules. Every output waits for my approval.

Org chart of Beatrix's AI team: one human at the top making every final call, a Chief of Staff routing work, a Council and Workshop role, and four studios (Training Studio, Client Desk, Content House and Back Office) with scheduled tasks marked
Flow diagram: work starts either when Beatrix asks or on a schedule, Claude opens the right playbook and does the job, and every path ends at the human gate where she reviews and approves, with a learning loop feeding improvements back into the playbooks

Your team leaves with
their own AI team.

A build session, not a lecture. Everyone works on their own job, with at least one role running before we finish.

The format

Half-day to design.
Full day to build.

The half day goes from "AI is a chatbot" to a designed team with first job descriptions drafted. The full day ends with at least one role live in each participant's own account.

The arc
1: Map your repeatable work
2: Name the team you'd hire
3: Write the job descriptions
4: Put one to work, live
5: Schedules & the human gate
What you leave with

A working system, not notes

  • An AI team designed around each participant's actual job
  • Two to three roles written as complete job descriptions
  • At least one role live in your own account before the end
  • A template for adding the next role without me
Who it's for

Any knowledge worker. Any tool.

  • Teams who already use AI and want to stop starting from scratch
  • Leadership groups who want to see what AI actually changes
  • Works with Claude, ChatGPT or Microsoft Copilot
  • No code, no technical background needed
Enquire about the workshop See Claude training

Questions about
building your AI team.

No. The method is writing job descriptions in plain English. No code at any point.

Claude, ChatGPT or Microsoft Copilot. The richest build uses Claude, where roles live as reusable skills and jobs run on schedules. See my Claude training page.

An AI team designed around their own job, two or three job descriptions, at least one role working live, and a template for adding more.

Training teaches the tools. This workshop builds a system that keeps working after everyone goes home.

Scheduled roles prepare work on their own, like a morning brief or a draft reply. Nothing goes out without a human approving it.

Yes. It works as the capstone of a training series, or on its own for confident teams.

It depends on group size, format and delivery. Tell me about your team and I'll send a quote.

Sessions that
actually build.

Every session with Beatrix is amazing, content relevant and easy to follow. I simply soak up knowledge.

Programme participantAI training, Brainster

The practical approach with live examples. Learning how to prompt properly, nothing I would change.

Programme participantAI training, global luxury retailer, London

Nice friendly atmosphere, nice to be well-informed and keep in the loop. Nicely executed course.

Programme participantAI training, Brainster