Mo prompts, same problems.

Since the debut of Terminator 2, we’ve all been primed to watch for signs of the impending skull-crushing robot apocolypse. As we’ve drawn closer to that pivotal moments of robots replicating forms of human intelligence and motor skills, the horror isn’t about destroying human kind, but another kind of doomsday—ending our careers. What was once dubbed the future is not present day. We’ve been here for a while now and that reality was abruptly introduced to the world through an internal company policy memo that reads “Start using AI, not more humans, now.”

A week ago, Shopify’s internal AI directive was leaked. It came from CEO Tobi Lütke, and it didn’t mince words.

I expect everyone at Shopify to become AI-native. Your job is to figure out how AI is going to make your work better. You are expected to use it to its maximum capability. I expect you to be the best in the world at this. Be world-class at prompting. Treat it like the most talented person you’ve ever worked with and you get to collaborate with them all day.

The memo goes into more detail with a bulleted list of edicts that all employees must become AI-native, use AI to its fullest extent, master prompting, treat AI as a top collaborator, prioritize AI over new hires, and consider AI use a fundamental part of their job. There’s no sense in looking back now, the line has been crossed.

You can bet a lot of money what Shopify unintentionally just shared with the world has been a building conversation and prioritization in board rooms of every Fortune 500 company and large enterprise around the world. It was only a matter of time before something like Shopify's leaked memo would give a shape and voice to what's coming out the executive leadership—find more ways to harness AI to be more productive without increasing our largest cost, headcount.

Do more with less ASAP—PDQ.

After reading the memo, I don’t agree with how Tobi planted the flag at Shopify but knowing what I do about enterprise company leadership I’m not surprised. The gospel of digital has always been: ship faster, cheaper, more often—come hell or high water. Make more faster, cheaper in search for better—quantity over quality. Now they see AI as the way for teams to fail faster and arrive at product nirvana in record time.

My challenge with Shopify’s direction—the path that all companies seem to be on—is to stop looking at AI fueled quantity as the best method for discovery and innovation. It's absolutely a great use for the technology, especially for digital products, but it’s not the only way. And I would argue, not the best.

For starters Tobi has made it clear that everyone will be held accountable to using AI. Engineers and product leaders will no doubt be measured against usage metrics.

Out: Number of code deployments.
In: Number of AI prompts.

We’re not evolving by working this way—we’re just shifting the scoreboard. Same shit, different OKR.

The problem with most AI adoption efforts right now—like what Tobi handed down—is that they stop at the prompt. Learn how to ask better questions, generate more stuff, and somewhere in the pile, the good ideas will reveal themselves. Consider that generating more logos is not the best path to create an amazing brand. Prototyping is a solid practice but we can only get to quality and real innovation if we change our perspective of using AI like a vending machine.

And that’s the trap. We think if we just prompt harder, we’ll get smarter. But more output doesn’t mean better outcomes.

Prompting isn’t a strategy. It’s a tactic. And when companies treat it as the whole game, they flatten what AI can actually do. Especially across the entire company with many teams and departments that have zero to do with making products, but have the potential to innovate all the same.

You’re not “working smarter” because you asked Claude to write your email thirty seconds faster. That’s not innovation—it’s intellectual laziness dressed up in a productivity metric. The real value of using this technology shows up when you actually think with it. When you use it to pressure-test your assumptions, push your thinking into unfamiliar territory, model decisions before you make them. But that requires effort. And most teams don’t want effort. They want shortcuts. They want vending machine magic.

When the telephone was invented, people used it to send short messages like a spoken telegram—completely missing the point that it enabled real-time conversation. That’s exactly how I see most people pitching AI: blindly recreating the past with a tool that was built for something radically different.

Framing is the hinge that decides whether AI is a shortcut or a thinking tool. Not all tasks—most I would argue—need a flood of outputs. The only way to work smarter is through convergence. The smarter move is narrowing the question and creating better constraints. Sometimes it’s running five tightly defined experiments instead of generating 50 half-baked ideas.

The best way for us to harness AI unilaterally is to use it like a second brain. A partner in the thinking process. Our robots are incredibly helpful in helping us go wide (generate options) and then go deep (refine, structure, decide). Sure, they can chase volume but we’re really cooking with gas when we use them to build clarity.

And that’s the shift I hope more leaders will prioritize. Not to just “use AI” because that’s too surface and like I said, not a strategy. We all should learn how to use AI well, because that’s the super power. This is how AI policy actually improves the whole company—not just the product org, but marketing, HR, compliance, legal—everyone.

So sure, tell your teams to use AI. Push them to experiment, to get familiar, to stretch their workflows. But don’t confuse usage with understanding. The real opportunity isn’t in generating more, it’s in thinking differently. The teams that figure out how to use AI with intent—across every department, not just product and engineering—are the ones that will actually move the company forward. Not because they used it more. But because they used it well.

I’ve been in on this thread a while. There’s more coming. Hang on to your butts.