Tobi Lutke just gave every corporate email chain a name, and honestly, it’s the most accurate piece of tech jargon I’ve heard all year. “Slop grenades.” That’s what the Shopify CEO is calling the flood of half-baked, AI-generated garbage that employees are apparently lobbing into shared docs, Slack channels, and inboxes across his company. And once you hear the phrase, you can’t stop seeing it everywhere.

Wait, What Exactly Is a “Slop Grenade”?

Here’s the thing – it’s not that people are using AI at work. Everyone’s doing that now, that ship has sailed. The problem, according to Lutke, is what happens when someone uses AI to generate a huge wall of text (a report, a strategy doc, a Q3 planning memo, whatever) and then just… tosses it into the world without reading it first. It looks polished. It sounds confident. It’s got bullet points and headers and all the trappings of “real work.” But underneath it’s basically nonsense dressed up in a suit.

And then someone else has to actually deal with it. Read it, figure out what’s real, what’s fluff, what’s just AI hallucinating with extreme confidence. That person didn’t create the mess, but now they’re stuck cleaning it up. Lutke’s whole point, from what I can tell reading through the Fortune piece making the rounds on Reddit, is that this shifts the burden of thinking from the person who “wrote” the thing to whoever has to read it. That’s… kind of brilliant framing, not gonna lie.

This Isn’t Really New, It’s Just Faster Now

We’ve all worked with someone who sends a novel-length email when three sentences would’ve done it. That guy has existed since email existed. The difference now is that guy has a superpower. He can generate five pages of jargon in about eleven seconds, hit send, and move on with his day feeling productive. Meanwhile you’re sitting there at 4:45pm on a Friday trying to figure out if there’s an actual decision buried somewhere in paragraph fourteen.

Why Is Shopify’s CEO Even Talking About This Publicly?

Because Lutke has been weirdly candid about AI at Shopify for a while now. Remember when he told employees that using AI effectively was basically a baseline expectation for keeping your job? That memo went viral back in April. So this isn’t some random complaint, it’s a continuation of a pretty aggressive company-wide push to get everyone using these tools constantly. And now, months later, he’s basically saying “okay, we told you to use AI, but some of you are using it badly and it’s making everyone else’s life harder.”

That’s an interesting turn, if you ask me. Companies spent the last two years basically begging employees to adopt AI tools. Now we’re entering phase two, where leadership has to say “yes, but not like that.” It’s a pretty predictable pattern actually. I’ve seen this before with other tech rollouts – first there’s a mad rush to adopt something, then a slower, messier phase where everyone figures out the actual rules of using it well.

“The volume of work created has outpaced the volume of thinking behind it” – which, if that’s not the most 2026 sentence ever written, I don’t know what is.

The Real Problem Isn’t AI, It’s Accountability

Let’s be honest about what’s actually happening here. AI didn’t invent laziness. AI just gave lazy work a really convincing costume. Before, if you wanted to slack off on a project, your output looked bad. People could tell. Now you can generate something that looks thorough, looks smart, looks like you spent hours on it, when really you spent four minutes typing a prompt and zero minutes checking whether the thing actually makes sense.

The scary part is how this scales. One slop grenade is annoying but survivable. A whole team lobbing them at each other, every single day, in every single meeting prep doc and status update? That’s death by a thousand mediocre paragraphs. Productivity theoretically goes up (look how much “content” we’re producing!) while actual usefulness quietly goes down. It’s a weird inversion, and I think a lot of companies are going to hit this wall before they even realize what’s happening.

What’s interesting here is that this creates a totally new kind of office politics. Who’s the person whose docs you can trust versus whose docs you need to fact-check line by line? That’s becoming a real distinction between coworkers now, in a way it kind of wasn’t before. You used to trust people based on their track record. Now you’re also silently rating them on whether they actually read what the AI spit out before hitting send.

So What’s Shopify Actually Doing About It?

From what’s been reported, it sounds like Lutke isn’t trying to ban AI use (he’d be contradicting his own earlier mandate if he did that). It’s more about pushing a cultural norm: use the tools, but own what you produce. Read it. Edit it. Make sure it reflects actual thinking, not just prompt output. Basically, treat the AI draft the way you’d treat a rough first draft from an intern, not a finished product ready to ship.

Which, yeah, makes sense. But good luck actually enforcing that at scale across thousands of employees who are all under pressure to look productive. There’s a real incentive problem baked into this. If your performance review rewards volume and speed, why would you spend the extra twenty minutes cleaning up your AI output? Nobody’s measuring “thoughtfulness per document.” They’re measuring deliverables.

What This Actually Means

I think Lutke’s comments are a preview of a much bigger conversation that’s coming to basically every company that’s rushed to adopt AI tools without thinking through the second-order effects. The first wave was “can we use this?” The second wave, the one we’re entering now, is “okay we’re using it everywhere, but is it actually making things better or are we just moving garbage around faster?”

My honest take? This is going to get worse before it gets better. Companies are going to keep pushing AI adoption because the pressure to look innovative is enormous, and in the meantime the slop grenades are going to keep flying. Eventually there’ll be some kind of correction, maybe new norms, maybe tools that flag unreviewed AI content, maybe just a return to old-fashioned “did you actually read this before sending it” accountability. But we’re probably not there yet.

What I keep coming back to, though, is that phrase. Slop grenade. It’s funny because it’s true, and it’s going to stick around a lot longer than this particular news cycle. Next time you get a five-page memo that says nothing in five pages, you’ll know exactly what to call it.