There is a bimodal productivity distribution among engineers based on AI tool adoption, with a large gap between AI-native and non-AI-native workers.

A single engineer today can do what two full teams of engineers could do two years ago.

“A single engineer can do what you know two teams of engineers could do two years ago.”

Bending Spoons is acquiring software companies, cutting 80% of staff, and making the remaining 20% AI-first employees to unlock existing revenue.

“They get rid of 80% of the employees and then put the 20% who are left as AI first employees and they figure out how to get all of this incredible revenue.”

With Fable 5, the quality of work is bottlenecked by the user's ability to clarify unknowns rather than by model capability.

“Fable is the first model where I find the quality of the work is bottlenecked by my ability to clarify its unknowns.”

With AI, two or three people can now build software that five years ago would have cost $2-3 million per year and that companies would not have built at all.

“I have two or three people who are building software that I would say 5 years ago we would have spent maybe two or three million dollars a year building this.”

At Uber, more than 70% of pull requests are now attributed to AI agents.

“More than 70% of pull requests are attributed to local or cloud agents.”

The actual ROI from AI spending for the S&P 493 is near zero, with EPS growth driven largely by pricing power and buybacks rather than AI productivity.

“The answer as far as all publicly available data was that the actual ROI was somewhere between zero.”

Non-technical GitHub employees are being given direct CLI access to run AI agents that read all internal communications.

“We've just been distributing even to the nontechnical folks the CLI and then effectively we're just giving it access to like read about everything that we're writing.”