vaaya / workflows / Standalone compute

Standalone compute

Give your agents compute that scales past your laptop: fast sandboxes, isolated code execution, and GPUs from T4 to H100, spun up on demand, metered by the second.

What you get

Fast sandboxes and isolated code execution that scale past your laptop.
GPUs from T4 to H100, spun up on demand and metered by the second.
Untrusted or generated code runs safely, with results read straight back.

Just tell your agent

In Claude, Cursor, ChatGPT, or any MCP agent, say it in plain English. Two ways in:

Simple

Use Vaaya to run this Python script in a sandbox and show me the output.

Detailed

Use Vaaya to spin up an A100 GPU sandbox, run this training job, poll it until it finishes, read back the results, and shut it down so I stop paying — all metered by the second.

Type it at vaaya.ai/chat with nothing to install, or say it to your agent. Same balance, same tools, either door.

Tools this workflow uses

Every call runs through Vaaya. No per-vendor signup, API keys, or billing.

Modal

E2B

Daytona

Vercel

Part of a bigger ecosystem

These and hundreds more tools are reachable through Vaaya.