Once AI Clicks Teams Get Excited

We’re doing an embedded AI engagement with a mid-size company. Twelve people across finance, operations, and HR. We did a kickoff session on Monday. By Thursday, the project lead called me.

"You really got some people moving on this stuff."

One of the team leads had been openly skeptical. Three days later, he's the one evangelizing Claude to the rest of the group. People who had struggled with every previous attempt at AI adoption were, in his words, "starting to rip."

Three days. I sat with that for a while.

We've been talking about AI adoption for two years now. Most of that conversation has been about strategy decks and executive buy-in and change management frameworks. BCG has a whole model for it. The "10-20-70 rule." Ten percent algorithms, twenty percent technology, seventy percent people and processes. They're right about the ratio. But they're wrong about what the seventy percent looks like in practice.

I’m conducting a free live webinar on March 25th at 12PM ET where I’ll build a lead generation skill in Claude Cowork that you can use to amp up your marketing function.

You’ll get all the tools you need to implement this for yourself, free and I’ll walk you through all the steps to operationalize it

It’s exactly the kind of thing we’ll be helping owners like you build in our next AI for Business Owners Cohort. Register for the live session >>

What moved people wasn't a training session. It was building something they cared about.

The original plan was to get everyone on a call twice a week. Teach general AI concepts, then work through a big cross-functional project together. After the first session, I realized that format was wrong.

The finance people don't care about the operations team's problem. The HR people don't care about the finance team's reporting workflows. Making twelve people sit on a call while we work through someone else's problem is a waste of everyone's time.

So we broke it up. General AI concepts on Tuesdays for the whole group. Then smaller sessions by function where we build skills specific to what each team actually does. Finance gets skills for their workflows. HR gets skills for theirs. Operations gets their project built.

The tools make this possible now in a way they didn't a year ago. Claude Cowork lets non-technical people work alongside AI on real tasks, inside their actual files and systems. It's not a chat window anymore. It's a collaborator sitting next to you while you work. And Microsoft just announced Copilot Cowork this week, which brings Anthropic's Claude into the M365 environment.

The real point is this: the tools are good enough now that three days of focused, team-specific work produces more adoption than months of general training ever did. And the best part is these teams will keep on building skills for themselves long after we’re gone.

Here's what I'm watching. This team-by-team approach works. It gets people productive fast and builds genuine advocates inside the organization. But it also creates something that's going to need attention later.

Every team builds their own solutions. Finance has their skills. HR has theirs. Operations has theirs. Each one is tailored to how that group works, which is exactly why it's effective. But six months from now, this company is going to have dozens of individually crafted AI tools with no coordination between them.

Only 43% of organizations have any formal AI governance in place right now. The ones that encouraged adoption first and worried about coordination later are already dealing with what people are calling "shadow AI." Individually useful tools that nobody has a map of.

So the work I'm doing now is Phase 1. Get people productive. Build advocates. Create real wins that justify the investment.

Phase 2 is AI rationalization across the organization. Taking all those individual solutions and figuring out how they fit together, where they overlap, where they conflict, and what the company's actual AI architecture should look like. I don't think you can do Phase 2 without Phase 1. The advocates have to exist first. But Phase 2 is coming whether you plan for it or not.

The good news is that everything we're building is portable. The skills are markdown files. They move between platforms. If this company decides Microsoft's bundle is worth the trade-off, the work transfers. If they stay on Claude Cowork, nothing changes. The platform decision can wait because the skills aren't locked in.

That's what I'm telling every company I work with right now. Get your people building. The coordination layer comes second. But it does come.