Daily curated AI insights you can't miss.
The 2x people and the 100x people are using the exact same Claude... the leverage is not in the weights, it's in how you wire the work.
When you sit down with Claude Code or Cursor, you're not writing software. You're hiring, training, and managing a workforce made of markdown.
The organization that captures what it learns like this gets smarter every single day. The one that doesn't wakes up every morning with amnesia.
The core tension in Garry Tan’s closing keynote is the shift from traditional software engineering—where you write every line of code yourself—to managing AI agents that do the heavy lifting. Tan reports a 400x increase in output by moving from writing code to orchestrating a workforce of AI agents. The problem is that most developers still treat AI as an autocomplete tool rather than a scalable workforce, missing the structural changes needed to capture that leverage.
Tan’s concrete technical path is to encode every organizational component—roles, processes, even performance reviews—as markdown-based skill files. This turns the company into a library of reusable skills, with a company brain (like his own project GBrain) acting as both library and librarian to supply context. The discipline of execution is to never do one-off work: every task must be refined into a skill file, preventing the organization from waking up with amnesia each day.
For serious builders, the takeaway is that the real leverage is in wiring the work, not in the model weights. The same Claude or Cursor used by a 2x person is also used by a 100x person; the difference is how you structure the system. Tan argues that abundance is not a policy paper but shipped software, and that AI-native companies can now ‘boil the ocean’—automating tasks that were previously impossible. The new physics of business means lean teams can achieve unprecedented revenue-per-head ratios by building on this infrastructure.
The core tension in Garry Tan’s closing keynote is the shift from traditional software engineering—where you write every line of code yourself—to managing AI agents that do the heavy lifting. Tan reports a 400x increase in output by moving from writing code to orchestrating a workforce of AI agents. The problem is that most developers still treat AI as an autocomplete tool rather than a scalable workforce, missing the structural changes needed to capture that leverage.
Tan’s concrete technical path is to encode every organizational component—roles, processes, even performance reviews—as markdown-based skill files. This turns the company into a library of reusable skills, with a company brain (like his own project GBrain) acting as both library and librarian to supply context. The discipline of execution is to never do one-off work: every task must be refined into a skill file, preventing the organization from waking up with amnesia each day.
For serious builders, the takeaway is that the real leverage is in wiring the work, not in the model weights. The same Claude or Cursor used by a 2x person is also used by a 100x person; the difference is how you structure the system. Tan argues that abundance is not a policy paper but shipped software, and that AI-native companies can now ‘boil the ocean’—automating tasks that were previously impossible. The new physics of business means lean teams can achieve unprecedented revenue-per-head ratios by building on this infrastructure.
