Building an ambient personal agent with Jev as its tendrils
The author says Jev lets the agent stay with you, see your screen and write on it via Haiku, while Opus 5.5 handles reasoning, forming an ambient intelligence.
you are *not* ready for what i've concocted for this app since posting this. hot damn Opus 5.5 is changing my life, dude. ambient intelligence, with @typesafeai's Jev as your agent's tendrils. meaning: your agent is always with you, sees what you see, uses your computer through Jev, writes and dictates text on screen through Haiku (so it's fast and cheap), and reasons // acts as your agent. for me that means Opus 5.5 via CC. turns out when you are building something and tell claude "my primary goal is for this to feel magical for the user. like it shouldn't be possible"...it works. 😅
This post quotes another post: x.com/rileyralmuto/status/2105009201332236292
I’ll post a demo of it when a few other bits of the app are done! computer use and augmented memory recall are the two greatest so far. it’s also woven into a cognitive architecture I’ve been building for a very long time, and it significantly improves recall speed/on-the-fly retrieval. I see vision-related use cases being among most obviously game changing implementation tho. imagine openworld games and sims where agents operate autonomously, with jev~ocr running visual processsing in some way. seems ripe to me ¯\_(ツ)_/¯ it feels like with computer use the thing I’m asking to touch or edit gets highlighted before I finish the sentence though. it’s nutty