THE NUMBER: $30 — the monthly cost of a Microsoft 365 AI agent seat. Your junior analyst costs more than that before they finish their first cup of coffee.
In 1914, Henry Ford raised wages to $5 a day — double the going rate — and set the exchange rate for human industrial labor that defined the 20th century. This week, Microsoft may have done the inverse. When you price an AI agent as a human seat in your workforce management software, you aren't launching a product. You're publishing a conversion table. Agent costs X. Employee costs Y. Agent doesn't take PTO, doesn't negotiate equity, and doesn't ghost you after the second interview. Every CFO in America now has a spreadsheet cell they didn't have last Monday.
Dario Amodei — CEO of the company whose chatbot just hit #1 on the U.S. App Store and crossed $20B ARR — told an audience this week that AI will eliminate half of all entry-level white-collar jobs. Note what he isn't doing. He isn't speculating. He isn't hedging. He's describing his sales pipeline. The conventional framing is that this hits the kids fresh out of college. It doesn't stop there. Block replaced departed staff with AI doing the same tasks this month. Salesforce is back from the dead on an agent story. Apple just announced a $599 MacBook that runs local inference. Apple isn't selling a laptop. They're selling the terminal that replaces your SaaS stack.
Yesterday we called them ostriches. Today we're being generous — at least an ostrich is fast. The executive version buries its head, schedules a working group, and calls it a strategy. You know the type. Three AI conferences this year. Forwarded this newsletter with "interesting read" in the subject line and zero follow-up. ChatGPT Enterprise license that six people use. Currently "monitoring the space." The space is not waiting to be monitored. The board is already running the math. The only question is whether you're presenting the restructuring plan or starring in it.
There's a second problem nobody is talking about: everything your organization has written since 2022 — your training materials, competitive research, institutional playbooks — may have AI fingerprints on it. Researchers confirmed this week that AI models fabricate scientific papers on demand. The trust infrastructure of knowledge was built for a world where producing credible work took months. When it takes 30 seconds, the whole system breaks. Pre-2022 institutional knowledge isn't just old. It's clean. Tag it. Date it. It's appreciating while everything else is suspect.
One more signal worth watching: Anthropic got blacklisted by the Pentagon this week and hit #1 on the App Store the same day. Apple's 1984 ad didn't sell specs — it sold rebellion against IBM. Anthropic just ran the same play at zero marketing cost. "We said no to the Pentagon" is currently the highest-ROI brand strategy in AI. The company's own investors have gone quiet about it. Show me the incentives and I'll show you the behavior.
Pull your head out. The math already ran.
-Harry & Anthony
On the site today: The full Signal/Noise briefing — labor repriced, knowledge contaminated, OpenAI governance exposed → getcoai.com
From the Scroll: The OpenAI Files resurface — amplified by Elon Musk to 22M views, originally compiled by Rob Wiblin. Governance isn't a checkbox. It's a procurement criterion. → getcoai.com
Deep read: The Roman Lead Problem — why pre-2022 knowledge is the new Scapa Flow steel → getcoai.com