> Agents can make code changes in seconds, but it takes them several minutes to test the output. This makes iterating extremely slow and manual ... We’re fixing

> Agents can make code changes in seconds, but it takes them several minutes to test the output. This makes iterating extremely slow and manual ... We’re fixing this by designing our app architecture to work for both humans and agents. The core principle here is that business logic should be completely decoupled from the UI and be able to run headlessly on desktop. Looks like we're back to Uncle Bob! [1] [1] https://youtu.be/o_TH-Y78tt4?t=2203 https://youtu.be/o_TH-Y78tt4?t=2203