# The Attention Bottleneck 🧠 > Source: > Published: 2026-10-20 09:29:21+00:00 👋 *Hey, Linas here! Every day, I break down 3 stories shaping the future of FinTech & Artificial Intelligence - plus the money movements and trends worth tracking. First time here? 400k+ FinTech and AI leaders get this daily. Join them:* Anthropic, the company behind Claude, has just published **the most detailed evidence yet of what happens to work when AI agents do most of it**. Its October 2026 report, *How Employment is Changing at Anthropic*, tracks a company where AI agents now log 6 hours for every hour its employees work. Four numbers stand out: - **🚀 Productivity:** code commits per engineer rose 5.1x in 6 months, with no rise in code later reverted or deleted. - **👥 Hiring:** junior roles fell from 27% to 9% of Anthropic’s technical hires. - **⏳ Attention:** AI agents spend about 80% of their session time idle, waiting on a human. - **🧠 Skills:** hiring managers now rank judgment and autonomy above domain expertise. For anyone building or investing in AI, by far the biggest lesson here is that **the scarce resource is now human attention and judgment, not labor.** And that shift *rewrites* how AI-native companies should hire, how they should be organized, which AI products will win, and where the returns will go. One company’s data matters here more than ever because **Anthropic is a leading indicator**. It works with frontier models before anyone else and has rebuilt its workflows around them, so what happens there is likely to reach AI startups next and the wider economy after. And unlike most future-of-work commentary, this report runs on ***telemetry***, not *forecasts*: agent logs, code commits, calendars, Slack networks and thousands of hiring scorecards. We analysed the full 20-page Anthropic report in depth, tested it against our earlier research on [Sequoia](https://linas.substack.com/p/sequoia-pat-grady-ai-briefing-agi), [a16z](https://linas.substack.com/p/a16z-state-of-markets-2026) and Claude Code creator Boris Cherny, and turned it into practical, actionable strategy for founders, operators and investors. Inside: - **The 12 numbers that matter** , in one table, with what each one means for your company - **6 strategic insights** , from the attention bottleneck to the breaking junior talent pipeline - **Where Anthropic’s data backs, and where it contradicts, [Sequoia](https://linas.substack.com/p/sequoia-pat-grady-ai-briefing-agi), [a16z](https://linas.substack.com/p/a16z-state-of-markets-2026) and [Boris Cherny](https://linas.substack.com/p/ai-future-of-work-five-roles)** , including one finding that cuts against the popular “AI companies hire more juniors” story - **A 9-point founder playbook** for hiring, org design and AI product strategy in the AI agent era - **An investor thesis map:** where value is likely to accrue, which business models are under pressure, and 7 diligence questions to ask - **What the data can and can’t prove** , so you know how much weight each finding can carry If you are building a company, hiring a team, or deploying capital into AI, what follows is basically your **operating manual for the Agentic AI Era**. It starts with the 12 numbers that matter.