I Wasn't the Best Coder. Here's How I Got My Team Using AI
How a manager learned AI coding firsthand, helped a team find useful workflows, and judged adoption by changes in the work rather than tool access.
The YouTube player loads only when you choose Play. YouTube then processes data under its own policies. Watch directly on YouTube ↗
The main idea
Start with friction in real work, support interested people, test an approved workflow, and check whether the result holds up in practice.
Key takeaways
- Access to a tool did not lead to regular use on Derek's teams. He first learned enough about AI coding to discuss its possibilities and limits from experience. Watch at 1:42 ↗
- The LEAD framework asks leaders to locate opportunities, enable people, activate workflows, and drive adoption. Watch at 4:45 ↗
- Measure whether a specific workflow improves and stays in use. Count quality, time, and the effort to review results, not just licenses or prompts. Watch at 8:15 ↗
Key numbers
- Regular AI adopters after five months
- 10 of 12 engineers
Derek's report about his own two engineering teams, from zero regular adopters at the start. This does not establish a result another team should expect.
Source ↗ - One legacy-code research comparison
- Two working weeks to four hours
Derek describes a measured comparison of comparable work for one activity. He explicitly says it does not mean every engineering task became that much faster.
Source ↗
Episode breakdown
0:00 Why engineers came to Derek about AI ↗
Engineers with stronger coding skills asked Derek for guidance on using AI. He traces that shift to his own hands-on learning, not to becoming the team's best programmer.
0:48 The adoption problem ↗
Giving a team access to a tool was not enough to make it part of their regular work.
1:42 Learning AI coding firsthand ↗
Derek explored where AI helped, where he struggled, and what changed when it became part of a real workflow.
2:42 Guidance from a different kind of experience ↗
He could help engineers evaluate new approaches even though they remained the stronger coders.
3:39 Different paths for different managers ↗
A manager can learn enough to guide adoption without copying Derek's personal AI coding path.
4:04 People, process, and technology ↗
The team needs support, the workflow needs improvement, and the tool needs a useful role.
4:45 The LEAD AI Adoption Framework ↗
Locate friction, enable interested people, activate an approved workflow, and drive adoption by checking regular use and results. Derek names security, privacy, accuracy, and human review as conditions of experimentation.
6:46 Five months of adoption ↗
Derek reports ten of twelve engineers using AI regularly, three advanced users, and one measured comparison in legacy-code research. These are his team's examples, not a general forecast of AI results.
7:40 Three practical starting steps ↗
Find recurring friction, support someone interested in learning, and measure whether one workflow actually improves.
8:35 The leadership lesson ↗
The manager's job is to create conditions for better ways of working, not to have every technical answer.
9:07 AI team readiness scorecard ↗
Derek points viewers to the free readiness scorecard as a starting resource.
AI in practice: improve one real workflow
Derek describes AI use in his engineering teams and a legacy-code research example. The video is a spoken account, not a recorded software walkthrough or a reusable prompt demonstration.
- Ask where repeated work or hard-to-find information slows the team.
- Choose one bounded problem and give an interested teammate time and support to explore it.
- Use an approved AI solution with security, privacy, accuracy, and human review addressed.
- Compare the complete workflow and result, then see whether people keep using the approach.
The leader’s review
The manager and team judge usefulness and quality. Tool access or prompt counts alone do not show adoption.
Try it with your team
Optional companion exercise: ask your team for one recurring friction point. Agree on an owner, an approved experiment, a reviewer, and one way to tell whether the whole workflow improved.