Built browser use using @typesafeai's new jev model It's really fast and very cheap and absolutely mogs a traditional LLM for most browser tasks.
