E2B

E2B provides AI agents with isolated, fast machines and real-world tools. The platform emphasizes secure, on-demand execution environments where agents can run code, access data, and use tools within separate microVMs. The product originates from an open-source project and targets large enterprises and developers who require trusted compute for agent workflows.

E2B runs a full machine per agent session, offering a sandbox that boots from a snapshot, includes a shell, filesystem, browser, and network, and supports lifecycle operations like pause and resume. The environment is designed to integrate with customers’ clouds (AWS, GCP, Azure) or E2B’s own cloud infrastructure, providing a consistent API and SDKs for control.

The platform supports multiple agent stacks and runtimes, including OpenAI Agents SDK, Claude Code, Codex, and custom harnesses. It emphasizes isolation, security, and observability, with per-sandbox secrets, per-sandbox network controls, and telemetry to monitor metrics and logs across sessions.

E2B outlines various deployment options (cloud or BYOC), a wait-free sandboxes model, and plans that include different regions and compliance features. It highlights SOC 2 Type II and HIPAA compliance, with deployments in US, EU, and APAC regions and a range of integration points in its documentation and product pages.

Use cases

• Build and train agents with isolated compute
• Ingest and analyze data with secure compute
• Run apps and workflows inside agents’ environments