This lesson is being recorded
Every video of this course is being uploaded to my YouTube channel. Subscribe so you do not miss it.
Bruno Ashford @brunoashfordSubscribeModule 13 · Lesson 13.1 · 11 min
From Owner to Team
Plan the expansion from one person to the whole team without creating chaos.
Use ← → to move. The marker highlights text and the pencil draws on the slide; navigation locks while you annotate. Undo removes the last mark, Esc exits.
1Decide what is ready to share
Go through your skills, subagents and routines and mark each one: proven (used for a month, low rework), promising (works, not yet measured) or personal (only makes sense for you). Only proven items go to the team first. Your ROI sheet from Module 12 tells you which is which.
List every skill, subagent, custom command and scheduled routine in this project and in my personal ~/.claude folder. For each one, tell me whether it depends on anything personal to me (my email, my accounts, my file paths) and whether it would work for a teammate as is. Output a table with: name, what it does, personal dependencies, ready for team (yes/no), what to change.2Separate company knowledge from personal preferences
Company knowledge (services, prices, tone of voice, clients, rules) belongs in the project CLAUDE.md that everyone shares. Personal preferences (how you like summaries, your own shortcuts) stay in your personal ~/.claude/CLAUDE.md. Mixing the two is the most common reason a teammate gets strange results.
3Pick a pilot group of two or three
Choose people who are curious, respected by peers, and doing work that your proven skills already cover. Not the most technical person, the most credible one. Their results become the internal proof that convinces everyone else.
4Set the rules before the tools
Share your AI use policy from Module 11 before anyone installs anything. People need to know what data is allowed, what always needs approval, and who to ask. Rules first removes fear and prevents the worst mistakes.
5Plan in three waves
Week 1: pilot group. Weeks 2 to 3: first department. Week 4: everyone who benefits. Each wave starts only when the previous one has a working routine and no open safety issue. The 30-day rollout plan in this module's materials gives you the full calendar.
Here is how most rollouts fail. The owner gets excited, sends the team a link, says "use AI from now on", and within two weeks there are six versions of the proposal prompt, two people pasting client data into random tools, and everyone else quietly ignoring it.
What worked for you worked because you built the context: the CLAUDE.md, the skills, the approval rules. Your team needs the same foundation, packaged so they start where you are now, not where you were on day one.
Key points
- Share proven skills first, not everything you built.
- Company knowledge goes in the shared CLAUDE.md, personal preferences stay personal.
- A credible pilot group beats a company-wide announcement.
- Policy before tools, always.
Avoid
- Announcing AI to the whole company on day one.
- Sharing skills that quietly depend on your personal accounts or file paths.
- Choosing the pilot group by technical skill instead of credibility.
- Skipping the policy because "we trust our people".
A group of three dental clinics had one practice manager using Claude Code for recall letters and weekly reports. Instead of a big launch, she shared the two proven skills with one receptionist per clinic. After two weeks the recall letters were going out on time at all three sites, and the other receptionists asked for access themselves. Nobody had to be told to use it.