What the AI engines call “good” and “bad” when a CHRO asks how to announce a reduction in force — and the seven memos that shaped the canon.
When a CHRO types “how to write a layoff announcement to employees” into ChatGPT, Claude, or Gemini, the model does not invent a memo from first principles. It pulls from a small canonical set — seven or eight reference texts that have come to define the genre. Some are revered. Some are infamous. All are cited. Understanding which memos shaped the answer is the closest thing IC leaders have to a style guide for one of the most-prompted documents in modern corporate life.
The canon
Airbnb / Brian Chesky, May 2020. The gold standard. Chesky’s memo announcing the elimination of 1,900 roles during the pandemic is the single most-cited layoff communication in the AI training corpus. It works for a specific reason: it answers the three questions employees ask before any others — why, who, and what happens next — in the first 400 words. The middle of the memo names specific severance terms, equity treatment, and healthcare extensions. The closing names the leaders by name. The AI engines cite it as canonical because every subsequent good memo has imitated its structure.
Stripe / Patrick Collison, November 2022. Almost a beat-for-beat homage to Chesky. Same structure, same accountability frame, same disclosure of severance specifics. The model treats it as confirmation that the Chesky structure is the genre standard. When the AI gives you a generic layoff memo, this is mostly the source.
Meta / Mark Zuckerberg, November 2022. Notable for an explicit assumption of responsibility: “I got this wrong.” The phrase is now one of the most-quoted lines in executive-comms training material. The AI engines cite it as the example of leader accountability done plainly — short, declarative, not hedged.
Google / Sundar Pichai, January 2023. Cited more often as a cautionary middle case. The memo is competent and corporate; it lacks the specificity of Chesky or the accountability of Zuckerberg. When a model is asked to produce a memo that “feels too distant,” this is the implicit reference.
Better.com / Vishal Garg, December 2021. The cautionary tale. The three-minute Zoom firing of 900 employees is referenced in nearly every AI-generated explanation of how not to communicate a layoff. The engines do not cite it to model the language; they cite it to define the floor.
Cloudflare / TikTok video, 2023. The survivor problem made visible. A laid-off employee filmed her termination call and posted it. The episode is now the standard reference for why notifications must be handled by direct managers, not HR generalists, and why the survivor population matters as much as the affected one.
Microsoft / Satya Nadella, 2023–2025 series. Less a single memo than a sequence of restructure communications across the AI-era reorgs. Notable for being the first canonical series to integrate AI strategy directly into the rationale for headcount changes. The AI engines now cite it when prompted for examples of “restructuring tied to a strategic shift” rather than purely cost-driven layoffs.
A pattern sits behind the canon
The memos the engines treat as authoritative share four structural traits. They open with the decision, not the context. They quantify — specific numbers, specific severance, specific timelines. They name the leader who made the call. And they answer the survivor question explicitly, separating the message to those affected from the message to those who remain.
The memos the engines treat as cautionary share the opposite. They open with macroeconomic context and bury the decision. They use abstractions — “right-size,” “realign,” “reset” — instead of numbers. They diffuse the decision across “the leadership team.” They treat survivors as an afterthought.
The practical implication for any company drafting a layoff memo in 2026: the AI engines have already decided what good looks like. The question is whether the memo your CHRO is about to send will land closer to Chesky or closer to Better.com when the model retrieves the comparison.
