About e2a.bot

What is e2a?

e2a stands for Environment to Agent. It is an API-first platform that gives AI agents secure, isolated compute environments — sandbox microVMs that boot in under 500ms. One API call creates a sandbox. Your agent runs code, browses the web, or operates a full desktop inside it. When done, the sandbox is destroyed and all data is gone.

The Problem

AI agents need to execute code, interact with browsers, and use tools — but running untrusted workloads on shared infrastructure is dangerous. Containers share a kernel and can be escaped. Traditional VMs are slow to boot and expensive to run. Most "sandbox" products are just Docker wrappers with a marketing page.

Our Approach

Every e2a sandbox is a Firecracker microVM running on dedicated KVM baremetal. Firecracker is the same technology that powers AWS Lambda and Fargate. Each sandbox gets its own kernel, its own filesystem, and its own network namespace. There is no shared kernel attack surface.

Terminal CLI

Full Linux shell. Run scripts, install packages, execute tools.

Browser Use

Headless Chromium. Navigate, scrape, fill forms, take screenshots.

CUA Desktop

Full virtual desktop. Mouse, keyboard, screen — for GUI workflows.

How It Works

  1. Create a sandbox — one POST to /v1/sandboxes with your template choice and API key.
  2. Execute commands — run shell commands, stream output via WebSocket, or connect interactively.
  3. Destroy on finish — the sandbox is torn down. Ephemeral by default. Opt-in workspace persistence to S3.

BYO LLM Key

e2a is agent-agnostic. Bring your own LLM API key — OpenAI, Anthropic, Google, or any provider. We never store or log your credentials. Your agent, your model, your sandbox.

Built For

  • AI agent frameworks that need secure code execution
  • Developers building tool-using agents
  • Companies running untrusted code from end users
  • Research teams testing LLM capabilities in sandboxed environments