Shopify CEO: AI 'Slop Grenades' Are Making Work Harder (2026)

What Are AI 'Slop Grenades'?

AI 'slop grenades' is a term coined by Shopify CEO Tobias Lütke to describe low-effort, AI-generated material — drafts, code, documents, or messages — that employees rapidly produce and pass along to colleagues without proper evaluation. The recipient is forced to spend time verifying the output's usefulness or correctness, effectively shifting the burden of quality control from the generator to the reviewer. Lütke first used the phrase publicly in September 2026, calling the phenomenon a growing problem at Shopify despite his earlier aggressive push for AI adoption.

The AI Mandate That Backfired

In April 2025, Tobias Lütke issued a company-wide memo making AI use a "baseline expectation" at Shopify. The directive required employees to prove that AI couldn't do a job before they were allowed to hire for it. The goal was to accelerate productivity and leverage AI agents to handle repetitive tasks. But less than 18 months later, Lütke is publicly lamenting an unintended consequence: employees are now "tossing" AI-generated slop grenades at each other, creating more work, not less.

As reported by Business Insider, Lütke described the problem candidly during an internal meeting. "We have a culture where people are generating stuff and tossing it over the fence," he said. The irony is sharp: a CEO who mandated AI fluency now has to teach his staff when not to use it.

How Slop Grenades Undermine Productivity

The core issue is that generating AI output is nearly free, but verifying it is expensive. When an employee uses a tool like ChatGPT or an internal agent to write a draft, they may spend 30 seconds generating it and 10 minutes reviewing it — or they may skip review entirely. The recipient, however, might need an hour to fact-check and polish. This imbalance is what Lütke calls the slop grenade effect.

According to Fortune, the problem extends beyond simple text. About half of Shopify's production pull requests now originate from conversations with River, the company's internal AI agent. While River can write functional code quickly, those AI-generated code changes still require human review. If the submitting engineer hasn't thoroughly vetted the code themselves, they are effectively delegating a significant portion of their responsibility to a colleague.

A Futurism article captured the irony: "The cost of generating cheap AI output is being transferred as an expensive verification burden to someone else." The term 'slop grenades' has gone viral because it perfectly describes a scenario many knowledge workers now recognize.

The Role of River: Shopify's AI Agent

Shopify's internal AI agent, River, plays a central role in this story. In a podcast interview with The Fabulous, Lütke explained that River now drafts up to half of the company's pull requests. He described the technology as transformative, but warned that "taste, judgment, and disciplined pruning remain crucial." As The Fabulous reported, Lütke emphasized that machines cannot take responsibility — only humans can.

The table below compares the intended benefits of River and similar AI agents against the slop grenade problem:

Aspect	Intended Benefit	Slop Grenade Outcome
Code generation	Faster output, less boilerplate	Unreviewed code passes bugs or security issues downstream
Drafting documents	Quick first drafts	Superficial content that looks professional but lacks substance
Customer responses	Consistent, fast replies	Generic answers that miss nuance or contain errors
Meeting summaries	Time savings	Incomplete or inaccurate summaries requiring redo
Pull requests	Higher velocity	Half of PRs originate from River, but verification burden increases

The table highlights that the same tools can either boost productivity or create hidden costs, depending entirely on how humans apply oversight.

Broader Industry Implications

Shopify's experience is not unique. Lütke's term 'slop grenades' has resonated widely because it captures a systemic risk across organizations that rushed to adopt generative AI. A Search Engine Journal piece noted that the problem is especially acute in companies where AI usage is mandated rather than encouraged with guidelines.

247Wall St. reported a discrepancy between Shopify's public messaging to investors — which touts AI integration as a core growth initiative — and the internal reality of slop grenades. This tension suggests that even leading tech companies are still learning how to manage the human side of AI adoption.

Lessons for Businesses Using AI

What can other companies learn from Shopify's experience? Here are key takeaways based on Lütke's own warnings:

1. Mandate use, but set boundaries. A blanket requirement to use AI can backfire if employees aren't taught when human judgment is irreplaceable.
2. Verify before you pass. Every AI output should come with a "trust but verify" tag. The generator bears responsibility for initial screening.
3. Measure verification cost. Track not just how much AI is used, but how much time colleagues spend cleaning up AI-generated work.
4. Cultivate taste and pruning. Lütke repeatedly stresses that taste, judgment, and disciplined pruning are skills that matter more in an AI-augmented workflow.
5. Create a culture of accountability. As Lütke told The Fabulous, "Machines cannot take responsibility. People must."

What Tobias Lütke Said About Responsibility

In the same podcast, Lütke expanded on the philosophy behind Shopify's AI strategy. He argued that the best AI users are those who know when to ignore the AI's suggestions. "The temptation is to take whatever the machine spits out and run with it," he said. "But that's not creation — that's just distribution of noise." His comments, covered by Startup Fortune, underline a deeper point: AI amplifies both human intelligence and human laziness.

The Future of AI at Shopify

Despite the slop grenade problem, Lütke remains committed to AI. River continues to evolve, and Shopify still expects AI to drive significant efficiency gains. But the company is now investing in training and processes to reduce verification waste. Lütke has asked teams to treat AI output like a rough first draft from a junior employee — something that always needs a second look before it goes anywhere.

The lesson for the rest of the industry is clear: AI adoption is not just about deploying tools. It requires new norms, new skills, and a culture that values "no" as much as "yes" when it comes to AI-generated work.