Teaching / Course 06
Team AI Enablement
One day to bring a whole team to a shared, safe level of AI use
A one-day workshop in which a team learns to hand real tasks to AI assistants safely, builds a shared library of reusable instructions, and leaves with a plan to keep practising.
A hands-on workshop for whole teams, technical or not: enough about how assistants work to judge their output, rules for safe use, practice on the team's own recurring tasks, and a shared, versioned skills library that a team admin maintains afterwards.
- 1 day · 7 contact hours
- Whole teams in any function, plus the person who will look after the team's AI setup.
- Participants work on their own recurring tasks; the instructor builds the team's first skills live and hands them to the team admin.
- Three automated tasks per participant and a versioned team skills library with a usage policy.
Feasible in one day when the organisation's AI tools are licensed and set up beforehand and the team brings real tasks. One day builds habits, a shared library and a common vocabulary, not engineering skill.
Outcomes
What a participant can do afterwards.
- Explain in plain terms how an AI assistant produces an answer, and why it can be confidently wrong.
- Apply the team's rules about which data may go into which tool.
- Hand a recurring task to a desktop agent such as Claude Cowork, ChatGPT or Microsoft 365 Copilot, with clear instructions, examples and a check of the result.
- Turn an instruction the team keeps repeating into a reusable skill everyone can use.
- Measure honestly how long a task took before and after, including the time spent checking.
- Follow a four-week practice plan, so the team reaches a similar level instead of splitting into experts and non-users.
Depth
Realistic depth within one day.
The difference between these three columns is the difference between a course description and a course. Everything in the third column is real content; it simply isn't something anyone builds from scratch in 7 hours.
Implemented
Written by participants, from scratch or near it.
- Three of each participant's own tasks handed to an assistant
- A shared team skills library
- A one-page team usage policy
Practised
Run, tuned and measured against prepared scaffolding.
- Writing instructions and examples
- Checking and correcting output
- Measuring time saved
Surveyed
Explained and demonstrated. Not built in one day.
- Automations that connect several systems
- Writing software with coding agents
- Organisation-wide governance programmes
The day
Working with AI as a team
Give everyone on the team the same safe working habits and a shared set of reusable skills.
-
How assistants work
Enough mechanics to judge the output: prediction, context, tools, why answers sound certain when they aren't, and what each tool can reach.
-
Safe use
Which data may go where, confidentiality, copyright, checking facts and figures, and when a person signs off. Drafting the team's usage rules.
-
Lab 1: first tasks
Each participant hands their recurring tasks to a desktop agent, checks the results and notes what went wrong.
-
From prompts to skills
Instructions, examples and reference files packaged as a reusable skill. The instructor builds the team's first skills live with Claude Code.
-
Lab 2: the team library
Participants turn their best instructions into skills, test them on each other's tasks and add them to the shared library.
-
The admin's job
Versioning, distributing and retiring skills; connectors and permissions; collecting feedback from the team.
-
Measuring and practising
Before-and-after timing for each task, the team usage policy and the four-week practice plan.
-
Review
Each participant shows one task, what it saves and what still needs a person.
Three automated tasks per participant, a versioned team skills library, a usage policy and a four-week practice plan.
Labs
Three deliverables, each one the input to the next.
- First tasks Three recurring tasks per participant handed to an assistant, with a note on what had to be corrected.
- Team skills library Reusable skills, tested on colleagues' tasks and added to a versioned library.
- Practice plan Before-and-after timings, the team usage policy and a four-week practice schedule.
Capstone specification
- Every participant has three of their own recurring tasks working with an assistant, each with a recorded check of the output.
- A shared skills library with an owner, a version history and a short description of each skill.
- A one-page usage policy: approved tools, data rules, review duties and who to ask.
- Before-and-after timings for each task, including the time spent checking.
- A four-week practice plan with one small exercise per week and a date for the follow-up check-in.
Stack
The tools, and why each one is in the room.
- The team's licensed assistant (Claude with Cowork, ChatGPT or Microsoft 365 Copilot) Teams learn fastest in the tool they will use the next morning.
- Agent Skills: folders of instructions, examples and reference files A reusable format the admin can version and distribute, instead of prompts copied between chats.
- Claude Code (instructor) Builds and tests the first skills quickly, so the day's time goes to the team's own tasks.
- The organisation's admin console or a shared repository One place where the current version of every skill lives.
- A timing sheet Makes time saved visible, including checking time, so the team keeps what works and drops what doesn't.
Discussed, not taught hands-on
- Automations that connect several systems are the subject of Course 4.
- Writing software with coding agents is covered in Course 3.
- Organisation-wide strategy and governance are covered in Course 5; the workshop produces a team-level usage policy only.
Logistics
Who it is for, and what it needs.
- Teams in marketing, sales, finance, HR, operations, customer support, legal or research, their managers, and one or two team admins who will own the setup afterwards.
- None beyond everyday office software. Participants need an account for the team's assistant and three recurring tasks prepared.
- 8–20 participants per instructor; above 20, a second facilitator or two sessions.
-
- The team's assistant (for example Claude Team or Enterprise with Cowork, ChatGPT Business or Microsoft 365 Copilot) is licensed, and every account works before the day.
- Each participant brings three recurring tasks from their own work, with example inputs and good outputs; confidential material is replaced by anonymised samples.
- The instructor builds the first skills live with Claude Code; the team admin learns to edit, version and distribute them.
- About 4.5 hours of hands-on work on participants' own tasks and 2.5 hours of explanation and demonstrations.
- A follow-up check-in after four weeks reviews the practice plan and the skills library.
One day builds safe habits, a shared library and a common vocabulary. It doesn't turn participants into engineers, and the time savings depend on the four weeks of practice that follow.
Book a training
Tell me which course and roughly when. You get a reply from me, not a sales sequence — dates, a quote, and an honest answer if the course is a poor fit for what your team actually needs.