Current MacOS computer use tools aren't the best.
Thats why I built Hunch, a local MCP that lets your LLM operate your Mac in the background, without disturbing you.
To compare Hunch with other tools of similar capabilities, I made 54 MacOS native tasks, and benchmarked Hunch with @trycua's Cua Drivers and @openclaw's Peekaboo.
Cua Driver completed 31 tasks, and used $37.42.
Peekaboo completed 41 tasks, and used $31.03.
Hunch completed 53 tasks, and used only $6.03.
More info in the thread 🧵
https://t.co/GeB5haclfB
Announcing Discovery Loop!
I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. The four of us have worked together for 14 to 30 years, and have helped build some of the world’s most used products, infrastructure and AI models, and we’re excited to turn our attention to this ambitious endeavor.
♾
Learn more at: https://t.co/Rv3LMdLluK
Cool use case for Hunch:
When setting up a server, bucket, or anything on a 3rd party service like Vercel, Cloudflare, AWS, etc, there will inevitably be manual steps that YOU have to make (creating and account, create the bucket and get the URL, set up the rules, etc).
With the Hunch MCP, your agent can do those tasks for you without taking control of your Mac. Ship faster and don't break your flow!
Get Hunch now! https://t.co/GeB5haclfB
@francedot This is great feedback. I'll re-run my benchmarks with Cua's MCP skill. I wanted Hunch to operate more as an "assistant" that operates on the same Mac so that all the important apps are already available. Ill run Hunch in the Lume VM myself and see where it can improve. Thanks!
Current MacOS computer use tools aren't the best.
Thats why I built Hunch, a local MCP that lets your LLM operate your Mac in the background, without disturbing you.
To compare Hunch with other tools of similar capabilities, I made 54 MacOS native tasks, and benchmarked Hunch with @trycua's Cua Drivers and @openclaw's Peekaboo.
Cua Driver completed 31 tasks, and used $37.42.
Peekaboo completed 41 tasks, and used $31.03.
Hunch completed 53 tasks, and used only $6.03.
More info in the thread 🧵
https://t.co/GeB5haclfB