Computer use in Muse: a vision-plus-actions loop that generalizes to any website
armand_ruiz · x · 2026-09-17
Armand Ruiz explains how computer use works in @muse: a vision-plus-actions loop — see the page, click, type, read the result, adjust. No special APIs or scrapers, which is why it generalizes to any site, even an hour-long flight hunt. The quoted post from Matt Deitke adds that computer use has improved dramatically over the past year and works out of the box in Muse, which ships with a browser for tasks that can't be solved via APIs or connectors, citing progress on the OSWorld 2.0 benchmark.
More from coding & agent
- Dev builds a fridge-mounted Hermes voice home hub that recognizes family members and sets reminders — Teknium · 2026-09-17
- SpaceXAI Marketer to Demo Running End-to-End Campaigns with Grok Bots — chaitu · 2026-09-17
- Teknium challenges critics: replicate a full repo faster and cheaper in one agent session — Teknium · 2026-09-17
- Cloning a million-view YouTube video with Codex and open-source Hypit — songguoxiansen · 2026-09-17
- Teknium: astra agent loses coherence within 30 minutes on long-context tasks — Teknium · 2026-09-17
- LLMs Solved Machine Understanding, but Machine Accountability in Agentic SDLC Is Still Unsolved — omojumiller · 2026-09-17