I blocked off an entire day yesterday for open office hours. Eight hours. The idea was simple: anyone in the company who wanted to talk about Claude Code, agentic AI, what they were exploring, what they were stuck on, or just wanted help getting started could drop in.
I expected a handful of engineers. What I got was a parade of people I didn't expect, building things I didn't expect, asking questions I didn't expect.
The sales guy who never had time for top of funnel
A member of our sales team showed up with a custom built top-of-funnel review dashboard. Not a bunch of technical questions. Not asking me how to pitch that work to product and engineering. A real working prototype using real data.
Our sales team has always known that top-of-funnel content would help their pipeline. But there was never enough time to dedicate to a thorough review of all this in addition to the other deals that were further along. The time it takes to spec, request, wait, review, and iterate on a landing page was time we couldn't justify pulling away from active deals in the middle of the funnel. The deals in front of them always consumed the available attention. And it would have been pretty hard to justify the expense of product and engineering going and building something.
So we just never did it. At least not to the degree we should. For years.
He sat down, described what he wanted, and had a working page. Not a mockup. Not a wireframe. A real page with real copy tailored to the need to quickly assess top-of-funnel opportunities. Now all we have to do on the engineering side is help make it deployable, secure, and easy to modify. The hard work has already been done.
The head of product streamlining busywork
My head of product came in with a problem that had been sitting in her backlog for months: software capitalization reporting. Finance needed structured data about how engineering time mapped to capitalizable development work, pulled from real Jira data, formatted in a way that finance could actually use.
This is the kind of task that's important but never urgent. It requires pulling data from Jira, understanding the capitalization rules, structuring the output for a finance audience, and nobody on the product or engineering team is excited about doing it. So it sits.
She walked out with a working solution. Real data, real reports, ready to compare against the data from doing it manually. A task that had been languishing because it fell in the gap between teams, solved in a single sitting because the barrier to building it dropped to nearly zero.
The person who said "wait, can we add SSO?"
This was maybe my favorite moment of the day. Someone was watching a demo of one of these internal tools being built and jumped in with "hey, can we add SSO with our corporate Google logins to this?" Since we're building internal tools for us to use, this is the perfect way to think about securing these applications.
Not "can we file a request for SSO." Just: can we do it? Right now? In this room?
And yes, we could. And we started.
That's the thing that's hard to describe about what I watched happen over eight hours. It wasn't just people building things. It was the speed at which ideas turned into follow-up ideas. Someone sees a tool taking shape and immediately starts thinking about what it could become, what it should connect to, how it fits into the broader picture. The gap between "what if" and "let's try it" collapsed to nearly nothing.
What I actually learned
I went into this thinking I was going to help people learn Claude Code. And I did. But the bigger thing I noticed was what happens when you remove the friction of getting started.
Every one of these people had ideas they'd been sitting on. Not because they lacked ambition or creativity, but because the cost of execution was too high relative to the perceived priority. The sales guy knew top-of-funnel mattered. The head of product knew finance needed that report. The person asking about SSO knew these tools needed proper auth.
They all knew what to build. They just couldn't justify the time, the resources, or the cross-team coordination it would have taken to build it the old way.
What I watched yesterday wasn't people learning a new tool. It was a room full of people realizing that the backlog of ideas they'd been carrying around in their heads, the ones that were always important but never urgent enough, those were now buildable. By them. Today.
The energy in the room wasn't "AI is cool." It was "I can finally do this thing I've wanted to do for months." That's a fundamentally different kind of energy, and it's the one I think matters.
The collaboration part
The other thing I didn't expect was how collaborative it got. This wasn't people sitting in isolation with headphones on, prompting in silence. People were talking to each other. Sharing what they'd tried. Riffing on each other's approaches. An engineer would see what the sales guy was building and suggest a better way to structure the data. The product person would see the SSO conversation and start thinking about access patterns across all these new internal tools.
Eight hours in, the room had more energy than it started with. That almost never happens.
Hard-core engineers, client support, client success, product, sales. People who don't usually sit in the same room working on the same kinds of problems were doing exactly that. And the ideas that came out of those collisions were better than anything any one of them would have come up with alone.
Just give it a spin
I know there's a lot of noise right now. AI hype cycles, vendor pitches, think pieces about whether this is real or a bubble. I get the skepticism. I've been skeptical about plenty of technology trends that deserved it.
But here's what I'll say after watching a room full of people for eight hours yesterday: stop reading about it and try it.
Not "evaluate it." Not "build a business case." Not "wait for the enterprise plan." Just sit down with a problem you've been putting off and give it a spin. The sales guy didn't need a strategy. The head of product didn't need a roadmap. They needed twenty minutes and a willingness to see what happened.
I've run a lot of workshops, hackathons, and office hours over the years. The thing that usually kills momentum is the gap between the idea and the ability to execute. People get excited, sketch something out, and then reality sets in. You need a designer, a backend engineer, an infrastructure ticket, three sprints of runway.
Yesterday, that gap barely existed. And when it shrinks to nearly zero, people stop self-editing their own ideas based on what they think is "feasible" and start thinking about what's actually useful.
That's the shift I watched happen in real time. Not people learning AI. People dusting off the ideas they'd been carrying around for months and finally building them. All it took was sitting down and trying.