Lyto
AI that sees your screen and does the work for you.
Start new thread

Lyto CLI - AI that works on your computer while you sleep.

byβ€’
Lyto CLI runs on your computer, logs into your accounts, and handles your tasks on a schedule while you sleep. Tell it once, it runs forever. Install it, describe what you want, walk away. Full shell access, Playwright automation, your own API key, your real Chrome profile. OpenClaw without the self-hosting. Same memory and credentials as the Lyto browser extension. Now with full access to your machine. No DevOps, no config files, no babysitting.

Add a comment

Replies

Best
Maker
πŸ“Œ
Hey Product Hunt πŸ‘‹ Arystan here, co-founder of Lyto. We're 17, building this out of Almaty, Kazakhstan, and today we're launching the thing we've been most excited about β€” Lyto CLI. The browser extension was step one. You could ask Lyto to click, fill forms, research, build spreadsheets, send messages. But it was still limited to what was in front of you. CLI is step two. Same agent, same memory, same saved logins β€” now running on your actual machine. You describe a task, set a schedule, and walk away. It handles the rest while you sleep. We built this because we were tired of automation tools that either require a CS degree to set up or only do one thing. Lyto CLI does the full job and takes five minutes to install. Would love to hear what tasks you'd hand off to it first. Drop them below and we'll show you exactly how Lyto handles it. πŸ‘‡

Would love to see some kind of approval step before Lyto sends messages or hits send on forms from my accounts. Even a quick "approve or edit" popup would do the trick, especially for anything going out on WhatsApp or email. Feels safer to let it act, but I want a final check on anything that leaves my machine.

Maker

Β This is already how it works and I'm glad you asked because clearly we need to say it louder. Anything risky or outbound stops and waits for your explicit yes before it runs. In Telegram it literally sends you an approve or reject prompt and does nothing until you answer. Nothing leaves your machine silently. That said, an edit before send option is a great idea, right now it's approve or reject and letting you tweak the draft first would be even better. Adding it to the list, thank you!

Would love a way to set recurring tasks that Lyto handles automatically, like every Monday morning it grabs my unread invoices from Gmail and drops them into a Google Sheet with totals. Right now it seems like you have to kick off each task manually, but the real magic would be it remembering routines and just doing them.

Maker

Β You just described a feature we already ship and apparently need to market way better haha. Lyto has scheduled and background tasks, so "every Monday morning pull unread invoices from Gmail into a Sheet with totals" is exactly the kind of routine you can set once and forget. It remembers the routine and runs it without you kicking it off. If you try setting one up and anything feels clunky I genuinely want to hear about it, this is the part of the product I care most about.

The "one memory across everything" idea is genuinely cool, but I'd love to see a way to set per-task permission boundaries, like letting Lyto draft and send WhatsApp messages only to specific contacts or only operate inside certain sites unless I approve. Right now it sounds like it's all-or-nothing trust, which would make me hesitant to hand it real logins.

Maker

Β Really fair concern and honestly the kind of feedback we want. Today the boundaries work like this: Lyto only acts from your paired account, risky actions stop and wait for your explicit approval, and your logins and keys never leave your machine. So it's not quite all or nothing, the approval gate is the boundary. But per contact and per site allowlists like you're describing would be a much finer level of control and I think you're right that it would make handing over real accounts feel safer. Adding scoped permissions to the roadmap, this comment is exactly why we launched here.