91����

Like many founders and CEOs, I worried that my company was too far behind on AI. So at first, I followed what I’d seen other companies do: I planned to put an executive in charge and a team underneath to lead our AI transformation. Then I got to thinking: It’s not enough for one team to use AI. How do I get all my 900 employees to lean in — and fast?

So I shut the company down for nearly a week. It inspired a wave of innovation.

For context, I cofounded Fetch in 2013. When shoppers use our app to scan their receipts, we give them rewards. We’re a profitable, fast-growing company with a $700 million gross revenue run rate and 13 million monthly active users, and I was confident that AI could help us innovate and move even faster. 

I started asking my people why they weren’t using AI at work, and the answer was always the same: They were too busy doing their jobs. So I thought: What if we just gave them uninterrupted time to try it out? We’d put all other work on hold and only focus on AI.

I called the idea AI Week. My executives pushed back and said, “Let’s do it for a day.” But I needed to do something big. It was important for everyone to understand how serious this was. We ended up agreeing to four days. Unless there was something urgent like a deal closing, we canceled all meetings, told clients we would be unavailable, and warned users that response times would be slower. 

For true innovation, you have to remove barriers so people don’t hide behind the limitations. So we did a survey asking what types of projects and solutions they wanted to work on, and then assigned them to 200 teams. I also got a small task force to build a few examples in advance. In addition, we created a Slack channel for employees to submit requests for any AI tool up to $500 a month. (We’d later cancel those that weren’t useful.) 

Then, starting on June 10, 2025, we held . On the first day, we showed the task force’s examples and brought in speakers from OpenAI, Anthropic, and Google to talk about their tools. Then we sent everyone off in their teams to do their hackathon, with experts on hand to provide assistance. At the end of the week, each team submitted a three-minute video of their solution, which we split up between 10 judges. The idea was to choose the best three — but the results were so good that we couldn’t even narrow it down past the top 20.

Along with a massive culture shift, two products came out of that week, both in beta testing now. One is a shopping agent where users can ask, “What am I about to run out of?” And it knows the answer. That’s great for customer retention. The other solution was more internal and foundational. Before, if I wanted to know whether our users had ever scanned a receipt at Danny’s Pizza Shop in Des Moines, humans would have to go look and it would take a week to find out. Now I have my answer in seconds. It means we can extend into the SMB market, because we can go to Danny’s and say, “We know there are about 2,000 Fetch users near you who spent $10,000 on pizza last quarter, of which you’ve only captured 5%. By offering points, we can help you grow to 7%.” It unlocked a scalable way to access our own data. 

Since shutting down the company for AI, every measure of productivity has increased significantly. Our revenue per employee is up around 30%, while the cost of the tools (we learned that we can almost exclusively use Claude) is negligible. But this is not a “one and done.” Because AI is evolving rapidly, we now have AI Fridays to give people the afternoon to experiment. And we’re shutting down again this July. People now have big ideas for AI, so this time we’re calling it Innovation Week. Because that’s the real point.