Claude, me and April Fool’s

Claude, me and April Fool’s

I am not an engineer. I want to get that out of the way upfront, because it's kind of the whole point of this story.

For the past year or so, I've been on a quiet journey with AI. Not a dramatic one — no big pivot moment, no lightbulb scene —  more a gradual, consistent effort to figure out how to use these tools to move faster at work. Writing, research, synthesising information, pulling together briefs that used to take me half a day. You get the idea..

But somewhere along the way something shifted.

I stopped thinking about AI as a way to do existing things faster and started thinking about it as a way to build things I couldn't build before. I'm no longer just a user of software; I'm making it do things specific to my role and needs. I’m building tools personalized to how I actually work—integrated systems, not a dozen SaaS apps bolted together that don't talk to each other. It’s what engineers call a "composable tech stack." Except I'm not an engineer. And that used to matter a lot more than it does now.

Anyone who works with Tom Chavez knows two things about him.

One: he talks about velocity constantly. How fast you move, how fast you learn, how fast you ship. This isn’t a buzzword he uses but moreover a way of thinking and doing. In a world where the pace of AI development is compressing timelines that used to take years into weeks, velocity isn't a nice-to-have. It's the whole game.

Two: he loves a prank. Not a small one. Tom goes big on April Fools. Elaborate, well-executed, and a complete commitment . The kind where people genuinely don't know what hit them until it's too late. And every year, the brainstorm starts early — what are we going to do this time?

This year, I was in on the prank. We were throwing ideas around and I suggested the Knowledge Vault.

On the surface, completely serious — a place where anyone across our portfolio companies, or what we affectionately call our Hive, could tap into our General Partners’ combined wisdom on demand. No calendar invite. No scheduling back and forth. Just: ask Tom or Vivek a question, get an answer. The thing people in the Hive have genuinely been asking for.

The twist? The vault would be full of completely useless, ridiculous, gloriously unhelpful advice. Delivered with total conviction.

People would think “ah, we've finally built the real thing”. With a simple and beautiful interface, anyone inside the hive could submit their most pressing questions. And they'd get an immediate response, in all seriousness, that they needed to go make Vivek a gin martini (dry, with a twist), place it on his desk, and then come back and ask the question again.

I had a week to pull it together. No grand timeline, no project plan. Just an idea, a deadline, and  Claude. So I opened Claude and started prompting.

Twenty minutes later, I had a working app. A dark, polished portal — the kind that looks like it took a design team weeks — where anyone in the Hive could select Tom or Vivek, type in their question, and receive a response. Complete with a spinning loader, a dramatic "consulting 25 years of wisdom…" message, and an answer delivered word by word, as if someone on the other end was actually thinking about it.

On April Fools morning, we sent it out to the team.

They fell for it. Hook, line, and sinker.

Messages started coming in. People were submitting questions. Real ones — about hiring decisions, product strategy, even fundraising. There were screenshots going around the Hive. Someone asked about their churn rate and got told, with complete confidence, that the answer was the same as why hot dogs come in packs of ten. Nobody knows. Move forward anyway.

The laughter that came back was the best part. Not just a chuckle — proper, shared, collective joy. The kind that's hard to manufacture and impossible to schedule.

Here's what I keep coming back to.

AI gets talked about in very serious terms most of the time. Rightly so — the implications are significant, the stakes are real, the pace of change is genuinely dizzying. I read Tom's posts and I feel the urgency he describes. Velocity matters. If you're not building with AI, someone else is. If you don’t jump in now, you will fall seriously behind.

But I think we sometimes forget that AI can also be fun. That it's a tool for creativity and play, not just productivity and disruption. That you don't have to be an engineer, or a founder, or a technical person to build something real and delightful. You just need a little imagination and the willingness to try.

I had never built a piece of software in my life. I built a fully functioning, beautifully designed, joke AI app in twenty minutes. It made an entire company of smart, busy people stop what they were doing and laugh together on a Tuesday afternoon.

That's velocity too.

Tom talks about the pace of change constantly — about how the tools available to us now have compressed what used to take months into days. He's right. But what I'd add is this: the speed is available to all of us. Not just the engineers. Not just the technical founders. All of us.

The only thing you need is a great idea and the courage to start prompting.