Jonathan Zeller’s new McSweeney’s satire, *“Why We Must Return to the Office to Use AI in Person,”* imagines a company where an employee’s job is largely to press “generate” and “approve” on an AI system—under the watch of thousands of executives, pervasive monitoring software and an ever-expanding return-to-office mandate.
It is plainly fiction, not a report about an actual company or policy. But its targets are recognizable: the tendency to wrap strict workplace controls in productivity language, to treat AI as a justification for more management oversight, and to cast physical presence as proof of loyalty rather than a decision tied to the work itself.
The joke is that the rationale never needs to make sense
The narrator works for the fictional conglomerate Mondo Mayo as an “Associate Slop Doula,” reviewing AI-generated sales material. Management insists that six-day, 14-hour office schedules—and later seven-day, 16-hour schedules—are essential. The stated reasons shift constantly: corporate principles, collaboration, construction-site distractions that supposedly build focus, and the need to outperform rival companies.
That instability is central to the piece. The office requirement is not presented as a practical operating choice supported by a clear business case. It is a loyalty test. The narrator eventually learns that the company’s aim is “never profit—it was always power.”
For real organizations, the useful takeaway is less dramatic: if leaders cannot articulate what in-person work is meant to improve, employees will supply their own explanation. And that explanation may be control, distrust or a desire to justify real-estate and management structures.
AI can amplify productivity theater
Mondo Mayo’s AI system, “SlurryHose,” turns the employee into a high-volume reviewer of generated output. Meanwhile, an “AI-powered Performance Enablement Software” monitors workers in the office. The result is a caricature of a real risk: deploying AI not to redesign work thoughtfully, but to intensify measurement around low-value activity.
Operators should distinguish between meaningful AI-enabled work and mere throughput. Questions worth asking include:
- Does AI reduce time spent on repetitive production, or simply increase the expected volume of it?
- Who is accountable for quality, accuracy and customer impact after a model generates material?
- Are managers measuring outcomes, or counting prompts, approvals, keystrokes and hours online?
- Does the technology give employees more judgment and autonomy, or make their work more scripted?
A system that generates more drafts can be useful. A system that turns every employee into an auditable button-pusher is unlikely to create much durable advantage.
Surveillance changes the employee proposition
The story’s “Best Buddy” monitoring tool is intentionally grotesque, but the underlying concern is serious. AI makes it easier to observe activity, summarize communications, score performance and trigger managerial intervention at scale. Those capabilities can help with security, compliance and operational diagnosis. They can also corrode trust when the purpose, limits and consequences of monitoring are unclear.
Leaders introducing workplace AI should be explicit about what data is collected, how it is used, who can access it and whether it affects performance reviews or employment decisions. They should also establish human review for consequential decisions. Calling monitoring “enablement” does not settle those governance questions.
What to watch next
The satire lands because return-to-office debates are increasingly entangled with AI strategy. Some teams may genuinely benefit from co-located training, faster feedback loops or secure access to sensitive systems. But “AI requires everyone in the office” is not, by itself, an operating model.
The companies best positioned to avoid Mondo Mayo comparisons will make narrower claims, test them against outcomes and adjust. They will define where presence matters, where flexibility works, and how AI changes roles beyond simply raising the quota. In an era when technology can make work more measurable than ever, credibility may depend on showing that measurement serves the work—not the other way around.
