Shopify cofounder and CEO Tobias Lütke warned that employees using artificial intelligence to generate unexamined emails and code are tossing “slop grenades” that create substantial cleanup work for colleagues. Speaking on The Knowledge Project podcast on Tuesday, Lütke noted that uncritical AI adoption degrades corporate productivity, mirroring a broader tech-sector retreat from unchecked automation.
The Bottom Line
- Quantifiable Drag: BetterUp Labs and Stanford’s Social Media Lab surveyed 962 full-time American desk workers, finding that over a third receive workslop, costing an average of 3.4 hours per month in revision time.
- Financial Impact: In 2025, lost productivity tied to revising workslop cost single employees $186 per month and scaled up to $9 million annually for an organization of 10,000 people.
- Strategic Pivot: Major tech executives, including Shopify CEO Tobias Lütke and Duolingo CEO Luis Von Ahn, are shifting from aggressive “AI-first” mandates to demanding stricter human oversight over automated outputs.
From Baseline Expectation to Operational Friction
The corporate rush to integrate generative artificial intelligence into daily workflows has hit a significant friction point. Tech leaders who initially championed AI as an absolute productivity multiplier are now confronting the hidden costs of unexamined automation. Shopify cofounder and CEO Tobias Lütke addressed this shift directly during a podcast appearance on Tuesday.
https://x.com/tobi/status/1909231499448401946
Last year, Lütke established AI usage as a baseline expectation for Shopify employees, instructing staff to prove they could not accomplish a task using AI before requesting additional human resources. However, speaking on The Knowledge Project podcast, he revealed that employees have begun leaning too heavily on automated tools without taking responsibility for the final output. He described these unexamined submissions as “slop grenades” that colleagues must intercept and untangle.
Lütke pointed to long, AI-generated emails as a prime example of the phenomenon. Instead of synthesizing key arguments concisely, workers often forward massive text blocks that require recipients to deploy secondary language models just to decipher the original message. “This is terrible.”
Defining ‘Workslop’ and Its Measured Toll on Productivity
Researchers have formally labeled this friction “workslop,” defined as polished-looking AI output that ultimately destroys productivity because it requires extensive human revision. Unlike traditional low-quality human work, workslop masquerades as legitimate and structured on the surface while harboring critical errors, such as broken hyperlinks in professional correspondence or overly complex, unmaintainable programming code.
Data from a joint study by BetterUp Labs and Stanford’s Social Media Lab highlights the growing scale of the issue. The survey of 962 American full-time desk workers revealed that 52.7% of respondents reported sending workslop to colleagues. The phenomenon occurs more frequently within organizations that actively encourage unbridled AI utilization.
Furthermore, interpersonal relationships within corporate teams suffer when workslop enters the workflow. Employees who received workslop rated the senders as less competent and less friendly. Over a third of recipients—specifically 36%—reported actively wanting to avoid working with those colleagues in the future.
Comparative Industry Retreats Among Tech Executives
Shopify is not isolated in its reassessment of automated output. Duolingo CEO Luis Von Ahn has similarly walked back aggressive automation targets. Last year, Von Ahn announced an “AI-first” corporate strategy, which involved evaluating employees on AI integration, substituting human contractors with large language models, and freezing headcount expansions wherever automation was feasible.
By May, however, Von Ahn admitted to Fast Company that he had grown overly enthusiastic after seeing strong initial software demonstrations. He noted that automated text generation fails to match the nuanced creativity required when scaling operations. For instance, when producing thousands of unique stories for language learners, Duolingo found that roughly 20% of the AI-generated volume constituted pure workslop requiring strict filtering.
| Metric / Study Data | 2024 Survey Baseline | 2025/2026 Current Data |
|---|---|---|
| Employees Encountering Workslop | 40% (reported encountering workslop) | 52.7% (Sending) / Over a third (Receiving) |
| Average Monthly Cleanup Time | 2.0 Hours per Employee | 3.4 Hours per Employee |
| Estimated Individual Cost | Not Reported | $186 per Month |
| Enterprise Productivity Loss (10k staff) | Not Reported | Up to $9 Million Annually |
Macroeconomic Implications for Enterprise Efficiency
The quantifiable expansion of cleanup time—surging from an average of two hours per month last year to 3.4 hours per month according to the latest BetterUp Labs and Stanford findings—demonstrates a measurable drag on corporate margins. For a mid-to-large enterprise with 10,000 employees, the financial toll reaches $9 million annually in lost output.
As corporate leadership teams analyze return on investment for enterprise software licenses, the hidden labor cost of reviewing synthetic output is forcing a structural pivot. Companies are moving away from unguided employee experimentation toward strict governance models.
Disclaimer: The information provided in this article is for educational and informational purposes only and does not constitute financial advice.
