Grok Bot Dominates Figma While Muse Wins at Video Cuts
Claire Vo tested four AI agents on video cuts and Figma designs. Here is why Grok Bot and Muse split the creative crown while Claude Duet failed.
I tested 4 AI agents on 9 real tasks. Here’s who won.
With Claire Vo · Sunday, November 1, 2026
Claire Vo conducts the first How I AI personal-agent benchmark, testing Grok Bot, Dot, Muse, and Claude Duet across nine real-world work and life tasks using autonomous computer use. Grok Bot wins overall due to its breadth, Figma skills, and structured UI documents, while Muse emerges as the strongest agent for personal and creative tasks. Claude Duet places last due to sluggish response times and excessive permission prompts, though it excels at bug fixing and product launch videos.
Key takeaways
- Muse took first place on raw video editing by correctly applying brand overlays, setting on-brand captions, and pacing cuts tightly without human guidance. Read more →
- Claude Duet failed an Adidas sneaker price check because major retail storefronts repeatedly blocked its browser session, leaving it unable to sort items by price. Read more →
- Claire Vo tested four personal agents (Grok Bot, Dot, Muse, and Claude Duet) across nine real-world workflows to see which could run a founder's daily load. Read more →
- Grok Bot won Claire Vo's benchmark with a 93 percent completion rate across nine autonomous tasks, separating itself through Figma manipulation and structured UI documents. Read more →
4 articles from this episode
Claude Duet Fails at Shopping But Wins Product Launch Videos
Claire Vo benchmarks AI agents on retail shopping and launch videos. Claude Duet gets blocked on e-commerce sites but dominates product marketing.
Claire Vo Tested Four AI Agents with Computer Use
Claire Vo benchmarked Grok Bot, Dot, Muse, and Claude Duet across nine tasks using operator agents. Here is how she ran the test and what broke.
Claire Vo Tested 4 AI Agents: Grok Won Work, Muse Won Life
Claire Vo ran 9 tasks on Grok Bot, Muse, Claude Duet, and Dot. Grok scored 93% on work, Muse won creative life tasks, and Claude stalled on permissions.
More How I AI episodes
- How Reddit’s designers build ultra-realistic prototypes with Figma + CursorNov 1 · 4 articles
- I tested Claude Duet. Is it worth switching from Grok Bot?Oct 25 · 4 articles
- Claude Code for creators: AI editing is hereOct 25 · 4 articles
- Meta Muse: the AI agent people actually loveOct 18 · 5 articles