When ChatGPT launched, Zapier did what most companies do: shared it in Slack, called it cool, moved on. No mandate, no urgency. Just “hey, this might be interesting for your roadmap.” Then GPT-4 dropped six months later and everything changed.

In a recent episode of Modern CTO Podcast with Joel Beasley, Zapier CEO Wade Foster shares how a company-wide “code read” and week-long hackathon increased AI adoption from 10% to 50% in just one week.

The secret? Hold everyone’s hand at the keyboard. Fear goes away when people actually build with these tools.

Table of Contents

How One Week Changed Everything

When the jump from GPT-3.5 to GPT-4 made it clear that AI wasn’t slowing down, Foster called the company-wide “code read” — a term Zapier had never used before. That was the point.

“The delta between 3.5 and 4, both in terms of model quality and cost, was pretty big. It didn’t take a genius to see it and go, if it continues in any way, it has a big impact on the way we work, the products we offer, everything.

As a CEO, you have to be very careful because if you pull that lever all the time, people learn to ignore it. But if you pull it into moments that are important, it’s annoying like people are like, well, it’s important.”

The most impactful thing Zapier did wasn’t memos or mandates. They shut down the entire company for a week and asked everyone — engineers, marketers, sales people, finance — to build something with AI.

“We went from less than 10% of people using AI weekly in their jobs to over 50% in one week. It was easily the single most important thing that drove the company forward.”

Why did this work? Foster says the fear of AI goes away once people use it. They stop imagining worst-case scenarios and start seeing what is possible. They also learn boundaries on their own, which makes conversation more practical.

This pattern shows up in every company Foster talks to on his podcast. Scale agents. Those who succeed with AI change share three traits: leadership that’s hands-on (not just preachy), frictionless acquisition so people can actually access the tools, and hackathons for everyone — not just engineering.

What does “AI fluency” actually mean?

In May 2025, Zapier began requiring all new hires. AI Fluency. But they didn’t just arm him. Foster’s team worked by function, studying how the top performers in each role were using AI. Then they built serviceability tests around those use cases.

They rank candidates on a four-level scale: Unacceptable, Acceptable, Adoptive, and Changeable.

“When you say, ‘Hey, show me what you’re doing with AI,’ you can hear it quickly. There’s a lot of people who are like, ‘Well, I asked ChatGPT to rewrite that memo.’ You’re like, well, you’re not doing anything, but there’s a whole world you can explore.”

The trait that matters most? Hyper curiosity. And because AI moves so quickly, they update their evaluation criteria every six months.

Why you should be automatically fired.

There is a fear that if you automate your work, you will become redundant. Foster says it’s backwards.

“I don’t know of any CEO or leader where someone has come up to me and said, ‘I’ve automated my entire job,’ whose response isn’t, ‘Holy crap, show me how.’ And I have nine more projects I’m ready to cut you loose.”

If you can systematically automate tasks, you’ve become one of the most valuable people in the company. He has the talent to run towards, not away from.

Related reading: