AI, Employee Equity, and Other Listener Questions

Welcome to Rework, a podcast by thirty seven Signals about the better way to work and run your business. I'm your host, Kimberly Rhodes, and I'm joined as always by the cofounders of thirty seven Signals, Jason Fried and David Heinemeier Hanssen. This week, I thought we would answer some listener questions that we've received by text, voice mail, email. Let's dive right in. First question is a text message. It says, love the rework podcast and would like to ask Basement and David a question. My company is all in on cloud, driven by the need to support AI tools. I was 37 signals managing this as the company moved off the cloud? Thanks. David, do you wanna jump in on that one? The answer is actually very simple. We don't have any AI tools in the main products at this time. We use AI to build the tools because we use them in Cursor or Visual Studio or Code or wherever else people are making our stuff. I use AI a bunch. I just use ChatGPT or Claude, and I just paste in code whenever I want a second opinion or I ask it a question. But there's no AI that summarizes your email in hey or anything like that. And a little bit of that is us not wanting to jump on a fad. We wanna find something that actually helps. I absolutely believe that there are uses of AI that will help. One of the things that Jason and I have been discussing is essentially ways of querying all this data that you put into the tools that we built, like Basecamp, for example, and be able to have a conversation about. This is perhaps the most compelling use case I've seen in information products. I haven't really enjoyed the kind of the summary stuff that often feels very gimmicky. I've seen it go wrong a lot, and that doesn't feel like that has value. So when that was the first frontier rush, everyone wanted to summarize. They realized, oh, hey. I can take a long email and can boil it down to three bullet points, so let's do that. Do you know what? That hasn't been for us. So we don't have any of those mechanics currently inside the tool, but I think it is actually a very worthwhile use case for cloud if you have a very spiky use. For example, you're training or detailing your model, you're you're specializing your model to your domain, and that sometimes requires a lot of fancy hardware. You don't wanna buy all that hardware to do a one time training to come up with a specialized model. Use the cloud. Cloud is great for that kind of stuff. Cloud, in our critique, fails when you found your steady state, when you know, do you know what? I just need 200 cards. Yeah. You can rent those. You're gonna pay through the nose if you need those cards …