Linda
Your local AI coworker for the Mac, like Dots but free!
58 followers
Your local AI coworker for the Mac, like Dots but free!
58 followers
Linda is a free, private AI coworker for your Mac. Say or type what you need and it does the work in its own browser and your folders: research, price checks, pages, sheets, renamed files, replies. Set up specialists, like a Researcher or a Bookkeeper, with their own instructions and skills. Watch every step in the Computer column and take over any time. It runs local AI on Apple Silicon or uses your own key, and asks before it sends, buys, deletes or posts. No account.
Hi Product Hunt! I'm the maker of Linda.
I love what AI agents do for me. But I hated two things about them: my files and data live on someone else's server, and the bill grows every month.
Instead Linda runs open models on your Apple Silicon. No cloud, no account, nothing sent anywhere unless a task needs the web.
Fast on small models. A decision layer picks the right model for each step, compacts the context and keeps only what matters. Small specialist models handle quick choices like clicks and form fields, so most steps finish in few seconds on an M-series Mac.
Real work. Linda browses the web, organizes your files, builds websites and runs routines while you keep working.
No token bills. Your Mac does the work, so Linda is free to use.
You stay in charge. Linda asks before she sends, buys or deletes anything.
Would you trust a local AI with your daily work, or does the cloud still win for you?
@jackrossi Fantastic!
Seems like an interesting product, although the market is flooded. What's unique about this?
thx@mikerubini Native AI local inference, small models + fast decision models (Jev-like) to route the skills and guide the agents. Local sandbox env for executing the task, everything run locally. Think of Claude Cowork, but 100% free and no cloud or api keys bills, 100% privacy!
the "asks before she sends, buys, deletes or posts" line is the part that actually matters here, not the local-vs-cloud question. we run an AI agent that takes real-world actions (phone calls) at Dial, and the lesson we keep relearning is that the scary failures aren't about where the model runs, they're about which actions are reversible. a wrong file rename costs you a minute. a sent email or a completed purchase doesn't. local inference solves the privacy and cost problem, but the confirmation gate is doing the actual safety work here, and you could build that same gate on a cloud agent too. so for me the answer to "would you trust it" depends less on local vs cloud and more on whether that gate covers everything irreversible or just the obvious ones
@galdayan agreed. Local solves privacy and cost, and the gate does the safety work. A cloud agent could have the same gate.
In Linda the gate lives in code, not the prompt, so no setting, specialist or web page can skip it. It covers kinds of action: send, buy, delete, post and submitting personal data. Every app plugin goes through it too. Everything else is contained: files stay in the project folder, commands run in a sandbox, and the browser is its own profile.
The quiet cases worry me most, like a form that commits something without saying so.
See the architecture overview: https://www.getlinda.app/features.html#overview