A Plain Memo
AI Agents "Run" Companies Now. Even Anthropic Won't Hire One.
Anthropic put an AI in charge of a real snack fridge in its own office — real inventory, real money, a name and a title: Seymour Cash, CEO. Meanwhile, Salesforce cut support staff from 9,000 to 5,000 and said AI agents made it possible; Cloudflare followed with 1,100 jobs in a record-revenue quarter, calling it "the agentic AI era."
So is AI quietly running companies right now? We check the one honest test anyone has actually run — Anthropic's own AI CEO experiment ("We would not hire Claudius") — then follow the receipts: the rehiring reversals (55% of leaders now call their AI layoff decisions a mistake), Gartner's finding that of thousands of vendors selling "agentic AI," only about 130 are the real thing, the Workday discrimination lawsuit a federal judge kept alive this March, and the restructuring filings where AI is named as the reason less than 10% of the time. The finding: the AI didn't do the firing. It just got the byline.
ℹ️ AI voice disclosure: This episode uses an AI-generated voice (ElevenLabs). The host is a pseudonymous byline persona (Reid Conlin).
In this memo
Sources
- Anthropic, "Project Vend: Part 1," 2025:
- Anthropic, "Project Vend: Part 2," 2026:
- CNBC, "Salesforce CEO confirms 4,000 layoffs 'because I need less heads' with AI," Sep 2025:
- TechCrunch, "Cloudflare says AI made 1,100 jobs obsolete even as revenue hit a record high," May 2026:
- CNBC, "Employers who laid off workers for AI are reversing their decisions," Jul 2026:
- Orgvue, "55% of businesses admit wrong decisions in making employees redundant when bringing AI into the workforce":
- Gartner, "Over 40% of agentic AI projects will be canceled by end of 2027," Jun 2025:
- MIT NANDA, "The State of AI in Business 2025":
- Forrester, "The State of Agentic AI in 2026: Companies Are Chasing, Few Are Catching":
- Duane Morris, "California federal court ruling in Mobley v. Workday," Jun 2026:
- Rest of World, "The downfall of Builder.ai's AI apps," 2025:
- PwC, "AI agent survey":