10 New Ways to Drive AI Adoption

Even AI-native companies struggle with this.

Some of my CEOs run AI companies with hundreds of millions in revenue.

Do you think everyone on their team is building AI agents?

Nope.

Remember that agents only became viable around 5 months ago. It's new for everyone.

Yes, there are early adopters in every department. But the majority of people are risk-averse, and want to 'wait and see'.

AI adoption is one of the biggest problems facing CEOs today... but I get to hear the most interesting solutions.

So here are my top 10 tactics for driving AI adoption at your company.

1. Token tracking

If you can measure it, you can manage it. Creating dashboards of token usage by team and leaderboards by individual makes usage visible. Yes, it's gameable, but it does reveal who your heavy users are—and your laggards. No one wants to be last.

2. Skills marketplace

In the AI world, a "skill" is a repeatable workflow that agents can look up and use when needed. Creating an internal marketplace where employees can create and share new skills allows agents across the business to level up and avoid reinventing the wheel.

3. 'Request an agent' boards

Some people love building agents more than others. Building a way for anyone to request an agent allows builders to pick up a ticket and help out. Seeing your agent idea come to life is often inspiring enough to want to build another by yourself.

4. App templates

Claude Code and Codex make it easy to vibe-code an app or dashboard, but putting it into production is another story. Rather than having your entire team learn Vercel and Supabase, you can create templates and guides so people don't have to build it from scratch.

5. Progression matrices

How do you know if you are a power user of AI? Outlining the behaviours at different levels of usage (beginner, intermediate, advanced, etc.) by function (sales, marketing, product, etc.) allows individuals to see what level they're building at, and how to go even further.

6. Training

Even if building agents is easy, if people don't believe they can build them, they won't try. Creating basic walkthroughs on how to build a simple agent demystifies the process and helps people build momentum. The best way to learn is to teach, so rotate the instructor so everyone gets the chance.

7. Hire an AI coach

An AI coach is a person who coaches leaders and teams to build agents. Part of the role is technical: building agents and the infrastructure needed to make it easier for everyone. The other part is working with people. The soft is the hard stuff.

8. Go to an AI dev day

If you can, get your tech team to go to the OpenAI/Anthropic Dev Days. These are inspiring events that will leave them feeling inspired and more confident about using AI in their work.

9. Create a "shocking rule"

I've written about shocking rules before, but rules can force people into action. One common one I'm hearing is "Demos, not memos", in which PRDs are banned in your product team and only working prototypes are reviewed.

10. Link it to performance

People act in their own interest. Adding AI usage to hiring scorecards, performance reviews, and a cash bonus creates a real incentive to adopt AI.

This barely scratches the surface of what I've seen, but I hope it inspires you that accelerating AI adoption is possible.

And if you're interested in joining my CEO community, where we hone our craft together, apply here.

Related Reading:

• Why I built an AI Organisation
• Why Every Startup CEO Needs a Chief of Staff
• What Your Team Meetings Are Missing

   

Originally published on May 27th, 2026