Ts Floyd@ts_floyd

关注

A Google engineer argues that 90% of engineers misuse LLMs like ChatGPT and Claude. The common mistake involves running a single agent, watching it fail, and then labeling the model as dumb. In contrast, all senior engineers at Google are already using loops and graphs for their self-learning agentic systems. An 18-minute session explains how to build a self-learning agentic system from scratch.

2w

Ts Floyd@ts_floyd

关注

A Google engineer argues that 90% of engineers misuse LLMs like ChatGPT and Claude. The common mistake involves running a single agent, watching it fail, and then labeling the model as dumb. In contrast, all senior engineers at Google are already using loops and graphs for their self-learning agentic systems. An 18-minute session explains how to build a self-learning agentic system from scratch.

2w