plori

plori gives you an AI agent that runs on its own cloud computer. The agent has a real shell, a persistent disk, network access, and memory that carries over between sessions. Ask it to clone a repo and run the tests: it does the work on its own machine, and the checkout is still there tomorrow. That one design choice changes what an agent can do: - It installs its own tools. If a job needs ffmpeg or a Python library, the agent installs it once and keeps it. - Long jobs keep running after you close the tab. You can also schedule runs, or set up a cron or webhook workflow and let the agent build the steps itself. - It can publish. Files the agent writes can go live on its own plori.site subdomain. - Its disk is a real filesystem, so notes, datasets, and repos accumulate instead of vanishing when the chat ends. You can talk to the agent in the web app, drive it from the CLI (@plori/cli on npm) or the REST API, or connect it as an MCP server so Claude and other MCP clients can hand it work. The same agent answers on every surface. Pick the model per conversation: hosted options include Claude, GPT, Gemini, and DeepSeek, or connect your own OpenRouter account and route through your own key.