← Back to blog
•6 min read

How to Train Your Team on AI: Lessons from 30+ Workshops

Most AI training is forgettable. Here's what we've learned from running 30+ hands-on workshops with engineering and ops teams.

We've run over 30 AI workshops in the last year—with engineering teams, ops teams, sales teams, even finance teams.

Some were wildly successful: 6 months later, 80% of participants are daily AI users.

Some flopped: people left inspired, then never used the tools again.

Here's what made the difference.

The Problem: Most Training Is A One-Time Event

The typical pattern:

  1. Company buys AI tool licenses
  2. Runs a 1-hour lunch-and-learn
  3. Sends a follow-up email with links
  4. Wonders why adoption is low

This fails because:

  • It's too short: 1 hour isn't enough to build muscle memory
  • It's too generic: Examples don't map to real work
  • There's no forcing function: No reason to keep practicing
  • There's no follow-up: Questions go unanswered

Result: people try it once, get mediocre results, and revert to old habits.

What Works: Training As A System, Not An Event

Here's our model (tested 30+ times):

Phase 1: Hands-On Workshop (90-120 mins)

Format: Live coding, not slides. Use their codebase or workflows.

Structure:

  1. Context setup (15 mins): Show how AI understands codebases, docs, tickets
  2. Live demos (30 mins): 2-3 real tasks (refactor, debug, document)
  3. Pair practice (30 mins): Participants code with AI while we circulate
  4. Pattern extraction (15 mins): Document what worked (becomes team playbook)

Key principle: Real work, real code, real PRs. Not toy examples.

Phase 2: Automations (Weeks 2-3)

Training builds skills. Automations remove friction.

We ship 1-3 workflow automations so AI becomes infrastructure, not a choice. Examples:

  • Auto-generate PR descriptions from commits
  • Turn meeting notes into Jira tickets
  • Post weekly status updates from GitHub + Slack

Why this matters: People forget to use tools. But if the tool is part of the workflow, adoption is automatic.

Phase 3: Async Support (Ongoing)

We set up:

  • Slack channel for questions
  • Office hours (weekly 30-min drop-in)
  • Prompt library (50+ examples from the workshop)
  • Internal champion (one person who becomes the go-to expert)

Why this matters: Adoption happens over weeks, not days. People need a way to ask "how do I...?" without scheduling a meeting.

The 5 Mistakes That Kill Adoption

1. Making It A Presentation

If you're sharing your screen for 90 minutes, you're doing it wrong. People need to use the tool.

Fix: Structure it as 50% demo, 50% hands-on practice.

2. Using Generic Examples

"Let's build a todo app" doesn't teach someone how to debug a production incident in their stack.

Fix: Use their actual codebase, tickets, and workflows. If they use Django, demo Django. If they use Jira, demo Jira.

3. Skipping The "When Not To Use AI" Talk

If you oversell AI, people will try it on the wrong problems, get bad results, and lose trust.

Fix: Be explicit about failure modes. AI is bad at: novel algorithms, architecture decisions, high-security code, legal/compliance. Teach people the boundaries.

4. No Follow-Up

One workshop doesn't change habits. Without follow-up, people forget.

Fix: Schedule 3 check-ins over the next month. Ask: "What worked? What didn't? What questions do you have?"

5. No Measurement

If you don't track adoption, you won't know what's working.

Fix: Measure:

  • Active daily users (not just licensed users)
  • Time saved (via workflow automation)
  • Use case breadth (how many tasks, not just one person doing everything)

If those numbers don't move, iterate.

What Success Looks Like

After a good training program:

  • Week 1: 60-80% of participants use the tool at least once
  • Month 1: 30-50% use it daily
  • Month 3: It's invisible—people don't think about it, they just use it
  • Month 6: New hires learn it as part of onboarding

We also track:

  • PRs with AI assistance (should be 40-60% after 3 months)
  • Time spent on repetitive tasks (should drop 20-40%)
  • Questions in Slack (more = good; means people are experimenting)

Our Workshop Playbook (You Can Steal This)

Here's the exact format we use:

Pre-Work (1 week before)

  • Survey: "What repetitive tasks do you hate?"
  • Pick 3-5 tasks to demo (refactor, debug, test, document, ticket)
  • Prep a branch in the team's repo

Workshop Day (90-120 mins)

  1. Intro (5 mins): Why AI, what to expect
  2. Context setup (15 mins): How AI reads code, how to write .cursorrules
  3. Live demo #1 (20 mins): Refactor a component, show prompt iteration
  4. Live demo #2 (20 mins): Fix a bug, show debugging workflow
  5. Pair practice (30 mins): Participants pick a task and code
  6. Debrief (15 mins): What worked? What patterns can we document?

Post-Workshop (Same day)

  • Share:
    • Prompt library (30-50 examples)
    • .cursorrules file for the repo
    • Slack channel for questions
    • Link to office hours

Week 2-3

  • Build 1-3 automations (e.g., PR summaries, ticket generation)
  • Run one office hours session
  • Collect feedback

Week 4

  • Measure usage
  • Create rollout plan for wider team
  • Document lessons learned

Training Different Roles

AI adoption looks different across roles:

Developers

  • Focus: Code generation, refactors, tests, debugging
  • Tools: Cursor, GitHub Copilot, Claude Code
  • Metric: PRs with AI assistance

Ops / SREs

  • Focus: Incident triage, runbooks, automation
  • Tools: OpenClaw, Warp, ChatGPT for scripting
  • Metric: Time to resolve incidents

Product / Managers

  • Focus: Meeting notes → tickets, status updates, research
  • Tools: ChatGPT, Notion AI, OpenClaw agents
  • Metric: Time spent on meta-work

Sales / Support

  • Focus: Email drafts, lead research, knowledge base search
  • Tools: ChatGPT, HubSpot AI, Intercom AI
  • Metric: Response time, email quality

Want Us To Run A Workshop For Your Team?

We've done this 30+ times. We know what works.

We'll run a 2-hour hands-on workshop with your codebase, then ship 1-3 automations over the next 2 weeks.

Book a 30-min discovery call →

Bonus: Our Pre-Workshop Survey Template

Copy this into a Google Form:

  1. What's your role? (Developer / Ops / Manager / Other)
  2. Have you used AI coding tools before? (Never / Once or twice / Weekly / Daily)
  3. What's one repetitive task you'd love to automate?
  4. What's one thing you're skeptical about with AI?
  5. What does success look like for you in 3 months?

Use the answers to customize the workshop.

Get started

Want help with AI adoption?

We run hands-on workshops and ship workflow automations for engineering and ops teams.

Book a 30-min discovery call →