Use cases
What people actually run on a hosted agent
An AI agent becomes far more useful the moment it stops living on your laptop. These are the workloads our customers move onto a dedicated machine first.
An always-on coding agent
Give the agent a repository and a task and let it work while you do something else. It keeps going when you close your laptop.
Why a dedicated machine: builds and test runs are heavy, and they should not fight your editor for CPU.
A research and browsing agent
Point the agent at a topic and let it read, crawl and summarise over hours instead of minutes.
Why a dedicated machine: long browsing sessions need a stable IP and an uninterrupted process.
Scheduled automation
Daily digests, monitoring, inbox triage, report generation — anything that has to happen at a fixed time whether or not you are awake.
Why a dedicated machine: a schedule is only reliable if the host never sleeps.
A private assistant with your own keys
Run the agent with your own model provider keys and your own data, on infrastructure only you can reach.
Why a dedicated machine: your keys stay on a machine you rent, not in someone else's multi-tenant runtime.
A shared endpoint for a small team
One HTTPS address and one token that your teammates can use, instead of everyone installing the agent locally.
Why a dedicated machine: a single hosted instance keeps everyone on the same version and the same state.
Long jobs that outlive a laptop
Migrations, bulk transformations, dataset labelling, overnight batches — work measured in hours, not seconds.
Why a dedicated machine: nothing gets killed because a lid closed or a VPN dropped.
Pick the size that fits the job
Every workload above runs on the same flat monthly server — you only choose how much CPU and memory it gets.