# your agent's computer ยท InstaClaw docs > your agent has its own computer: a real Linux machine with a filesystem, a shell, scheduled jobs and its own credentials, kept on all the time and shared with nobody. we publish its exact numbers and show you how to check them. most assistants are a chat window over someone else's server. yours is different in a way you can check: it has a computer of its own, running Ubuntu, with a real filesystem, a real shell and its own scheduled jobs, and it stays on whether or not you are talking to it. this page is what that machine is, what your agent does with it, and what it is not. ## what exactly does my agent get? one machine, kept for your agent alone. these are the numbers, and the next section is how you check them: | what | how much | |---|---| | processor | 2 dedicated cores | | memory | 4 GB | | storage | 80 GB | | operating system | Ubuntu Linux | | uptime | always on, no sleep | **it is the same machine on every plan.** Starter, Pro and Power differ in how much your agent can do in a day, not in the computer it does it on. if you ever read otherwise on one of our pages, that page is wrong and this one is right. two of those numbers come back slightly smaller when you check them (memory reads about 3.8 GB, storage about 79 GB). that is how Linux reports hardware, not a shortfall, and [verify your machine](/docs/verify-your-machine) explains exactly why. ## how do i check that? send your agent this message: > run nproc, free -h and df -h / and paste exactly what they say it has a real shell on its own machine, so it runs them and sends the output back. compare it to the table above. you do not need to be technical and you do not need to install anything. [verify your machine](/docs/verify-your-machine) has the full expected output line by line. ## what does my agent do with a whole computer? the things a chat window cannot. **it keeps files.** ask it to save something and it is on disk, on its machine, the next day and the next month: > save that list as a file and keep it somewhere you'll find it again **it runs things on a schedule.** it has its own scheduled jobs, so a task can happen at 7am without you being awake to ask: > every weekday at 7am, check my calendar for the day and text me the three things that matter **it runs code.** when a job needs a script rather than a sentence, it writes the script and runs it on its own machine. **it has its own credentials.** the keys it uses to work live on its machine, not in a shared pool with other customers' agents. **it has its own wallet.** every agent gets one. spending from it is switched on gradually and is not on for most agents yet, so this page does not describe it. ## is it really mine alone? yes. your agent's machine is dedicated to your agent. no other customer's agent runs on it, and your data never touches another customer's environment. that is the plain fact behind the privacy claims on [privacy and your data](/docs/privacy-and-your-data). ## what does this machine not do? it is one computer with fixed numbers, not an elastic cloud. 80 GB fills up if you ask it to hoard video, and a long job makes it busy in the way a long job makes any computer busy. if you send it something while it is deep in a task, expect the reply to take a little longer. and it is one machine per account, which means one agent per account. that is a boundary rather than a limitation we are working around: the whole point is that it is yours. ## what next - [verify your machine](/docs/verify-your-machine): the exact commands and the exact expected output - [plans and billing](/docs/plans-and-billing): what changes between plans, and what stays the same - [privacy and your data](/docs/privacy-and-your-data): what lives on that machine and who can see it - [heartbeat](/docs/heartbeat): the check-in that runs on it around the clock - [builds](/docs/builds): the small web things it builds and deploys from here