A computer for every agent.

Each agent gets a Linux pod with its own disk and root. Pay only while it runs.

Install the CLI Read the docs

macOS, Linux, and Windows · no sudo required

Give your agent a machine that remembers.

Install once. Then pod create, pod shell active, and start work.

$ curl -fsSL https://tipod-control-plane.ramasamy-vignesh.workers.dev/install | sh
# then: pod create && pod shell active

Persistent compute / APAC first

Same category as agent sandboxes — different contract: abort-stop, per-second billing, disk that survives pause.

Durable disk

Each pod keeps its own disk across pause and wake. Stopping never loses the disk.

Linux with root

Isolated KVM pod per agent. Install what you need. SSH in like any machine.

Pay while it runs

Prepaid machine-time by the second. Shapes: lite, core, and max.

Pricing

Starter includes carry-forward seconds. Open the dashboard when you’re ready to top up.

Starter$20 / mo
Included1,000 hrs (3,600,000 secs)
lite1×
core2×
max4×

Open the dashboard

FAQ

Short answers. Details live in the docs.

When are disk snapshots taken?

While a pod is active, a snapshot is taken about every one hour. A snapshot is also taken when you stop the VM. tipod does not snapshot every minute.

What machine shapes are there?

lite (1×), core (2×), and max (4×) — billed in lite-equivalent seconds per wall-clock second while running.

Do I pay when the pod is stopped?

No. Prepaid machine-time is consumed only while the pod runs. Unused seconds carry forward. Credit is shown as hrs (secs) — for example Starter includes 1,000 hrs (3,600,000 secs) lite-equivalent.

What happens if a stop cannot save the disk?

Abort-stop: a stop that cannot complete its stop snapshot does not succeed — the pod stays running so you don’t lose the disk.

Where do pods run?

APAC-first. Each agent gets an isolated Linux pod with root access.