An agent that only exists inside a chat window forgets everything and owns nothing. plori gives each agent a computer of its own in the cloud: a persistent disk, a real shell with tools, and memory that survives between conversations.
Opens your agent straight away, no signup needed.
Real files that survive across sessions, browsable and editable from the files panel. Repos, reports, datasets: they stay with the agent.
A browser, git, and code and research CLIs on PATH. The agent installs whatever else the job needs, and the installs persist.
Short notes the agent keeps about you and your projects, read at the start of every turn.
One agent per machine. Nothing shared between accounts, nothing leaked between agents.
Idle agents scale to zero. The machine's state is kept, compute stops, and disk within your plan is free.
Use it from the web chat, or drive it from your own tools over MCP.
Most execution environments for agents are ephemeral sandboxes: useful for running a snippet, gone when the session ends. That shape forces every task to start from zero and makes "pick this up tomorrow" impossible without you building storage around it.
A durable computer changes what you can delegate. You can hand an agent a codebase and let it keep working copies. You can leave a half-finished job overnight and resume it by saying so. You can schedule work for later and trust that the machine that wakes up is the one that remembers the context. The agent stops being a calculator with a chat window and starts being a colleague with a desk.
It is a persistent, isolated machine in the cloud that belongs to one agent: its own disk, its own shell and tools, its own memory. The agent works there the way you work on your laptop, and the machine outlives any single conversation. On plori every agent gets one automatically.
Sandboxes built for code execution are typically ephemeral: the filesystem is gone when the session ends, and persistence means wiring up snapshots yourself. A plori agent's computer is durable by default. Files written today are there next month, and the agent sleeps in place at zero compute cost instead of being torn down.
A real shell, a browser, git, and a baseline of code and research tools on PATH. Beyond that, the agent installs what a task needs, from language toolchains to CLIs, and those installs persist on its disk.
Nothing, within your plan. An idle agent scales to zero, so compute stops entirely, and the disk included in your plan is free. You only pay while it works, from one pool of credits.
Yes. Each agent's dashboard page has a files panel where you browse its directories, open and edit text files in a code editor, upload files for it to work on, and download anything it made.
Your account has one disk pool that all your agents draw from: 1 GB on Free, 20 GB on Pro, 50 GB on Power, and you can buy more at 10 credits per GB per month on any plan.