The first personal agent with on-device execution and compute.
SF
Joined May 2026
- Tweets39
- Following2
- Followers1.1K
- Likes20
Result retweeted
We built the best experience for local AI on MacOS
Introducing Result for Mac
> Computer Use with local models
> One-click connection to private inference: DGX, Halo, etc
> Use open-weight cloud models
> Access models running on your Mac on your iPhone
result.dev/mac
Result retweeted
Switch to Result!
A simple task this could do but Claude duet could not: find my polling station in NYC. Duet refused because entering my birthday violated safeguards. That’s ridiculous! Let the agents do useful things!
My polling site lookup task wasn’t very hard, but I’m still impressed a model like qwen 3.8 27B managed
Result retweeted
yet another upside of switching to open weights: less censorship ⚡️
A simple task this could do but Claude duet could not: find my polling station in NYC. Duet refused because entering my birthday violated safeguards. That’s ridiculous! Let the agents do useful things!
My polling site lookup task wasn’t very hard, but I’m still impressed a model like qwen 3.8 27B managed
Result retweeted
Today we’re releasing Result for Mac beta!
goal is to build the best UX for interacting w/ open-weight models.
in v1, you can:
- Download & run models locally on your Mac
- Access the latest models in our cloud
- Connect your private inference (DGX, Ollama, etc.)
All our cloud models are free to use without limits for a limited time.
Written with Swift & Rust, with computer-use features built in.
try it today. feedbacks appreciated
We built the best experience for local AI on MacOS
Introducing Result for Mac
> Computer Use with local models
> One-click connection to private inference: DGX, Halo, etc
> Use open-weight cloud models
> Access models running on your Mac on your iPhone
result.dev/mac
Result retweeted
my favorite feature is connecting my iphone anywhere in the world and using my mac as compute
We built the best experience for local AI on MacOS
Introducing Result for Mac
> Computer Use with local models
> One-click connection to private inference: DGX, Halo, etc
> Use open-weight cloud models
> Access models running on your Mac on your iPhone
result.dev/mac
Result retweeted
A simple task this could do but Claude duet could not: find my polling station in NYC. Duet refused because entering my birthday violated safeguards. That’s ridiculous! Let the agents do useful things!
My polling site lookup task wasn’t very hard, but I’m still impressed a model like qwen 3.8 27B managed
Cloud VMs keep getting hacked, so we built the first on-device iPhone agent
Introducing Result
> Cards & passwords stay on-device
> Run openweight models on your compute or cloud
> Native Apple Pay & Passwords
> Works in the background even when locked
result.dev/ios
Result retweeted
Oopps! Somebody just cracked the code!
Cloud VMs keep getting hacked, so we built the first on-device iPhone agent
Introducing Result
> Cards & passwords stay on-device
> Run openweight models on your compute or cloud
> Native Apple Pay & Passwords
> Works in the background even when locked
result.dev/ios
Result retweeted
what siri should be like and is not
Cloud VMs keep getting hacked, so we built the first on-device iPhone agent
Introducing Result
> Cards & passwords stay on-device
> Run openweight models on your compute or cloud
> Native Apple Pay & Passwords
> Works in the background even when locked
result.dev/ios
Result retweeted
This is the personal agent Apple should’ve built. On-device, supports ApplePay, and even local passwords.
Cloud VMs keep getting hacked, so we built the first on-device iPhone agent
Introducing Result
> Cards & passwords stay on-device
> Run openweight models on your compute or cloud
> Native Apple Pay & Passwords
> Works in the background even when locked
result.dev/ios
Result retweeted
> if you like experimenting and breaking systems to create great implementations in a generational consumer product
come work @tryresult
new experiment: <LoginWithChatGPT />
> allow users to login to their personal ChatGPT account and use it inside your site
> never need to pay for API usage to OpenAI
demo: login-with-chatgpt.vercel.ap…
should I make this an open-source library?