Launching today
Run Lobster
Each agent runs on its own cloud computer/isolated container
3 followers
Each agent runs on its own cloud computer/isolated container
3 followers
Each agent runs on its own cloud computer, completely separate from yours. It has a browser, apps, terminal, and file system, everything it needs to ship real work on your behalf, 24/7. Your agent can research the web and act on what it finds, build, deploy, and host live websites, manage your email, Slack, and calendars, read, write, and organise all your files, run any code and install any package, and keep working autonomously while you sleep. Try it FREE, no card needed.
Hey Product Hunt 👋
We're a small team based out of Oxford, and today we're launching RunLobster — and honestly, the pitch is the kind of thing that sounds obvious once you hear it:
Most "AI agents" are stateless wrappers around an LLM. We gave ours an actual computer.
Every RunLobster agent gets its own dedicated Linux container — 12 GB of RAM, 8 vCPUs, a real filesystem, a headful Chrome browser it can drive (and that you can watch live via noVNC), a code workspace tracked by git, persistent memory, cron jobs, and its own website at yourname.runlobster.com. It can install packages, scrape JS-heavy sites, write code, edit files, and ship apps. It's less "chatbot" and more "junior employee with a laptop."
A few things we're really proud of:
🦞 Cross-channel by default. The same agent is reachable on web, Slack, Discord, iMessage, and email simultaneously — and it's all one continuous conversation. A Slack mention can reference yesterday's email. iMessage can pick up a thread that started on web chat.
🦞 It's completely open source. So you can always export all your files and take your agent with you.
🦞 It reaches out to you. Within ~3 hours of signup, your agent does deep research on you (your GitHub, your site, the works) and sends you a personal message on whatever channel you're most likely to actually read. Within 24 hours, it builds you something — usually a personal website based on what it learned.
🦞 Live transparency. There's a side panel where you watch your agent work in real time — its actual Chrome screen, the tool calls it's making, the files it's editing. Less "trust me bro," more "look, it's doing the thing."
🦞 800+ integrations. Gmail, Drive, Notion, Linear, GitHub, Stripe, HubSpot, Salesforce, X, Reddit, LinkedIn — all OAuth-scoped per agent, no API keys for you to manage.
🦞 It's actually usable by non-engineers. Pick one of 20 role cards (sales, EA, recruiter, scraper, market analyst…), sign up, get useful output in five minutes. But if you're technical, you can drop into the terminal, edit the system prompt, install custom skills, and treat it like a coworker who never sleeps.
Why we built it: we kept seeing AI agents that forgot you the moment you closed the tab. That's not an agent, that's a search bar with extra steps. We wanted something that persists — that can spend two days researching something, save its work, schedule itself a follow-up, and come back to it. Something that earns the word "agent."
It's free to try (5,000 credits on signup, no card required). As far as we can tell we're the first to ship managed full-computer agents at this scale — 600+ live containers running right now on a single worker.
We'd genuinely love your feedback, would mean a lot. The team's around all day to answer questions 🦞