E2B

What E2B does

E2B provides isolated sandboxes that let agents safely execute code, process data, and run tools. Their SDKs make it easy to start and manage these environments. E2B sandboxes are open-source, secure environments with real-world tools for enterprise-grade agents. They offer a fast, secure Linux VM created on demand for AI agents, with templates defining the environment a sandbox starts with. E2B also offers Desktop Sandboxes to provide secure virtual computers in the cloud for LLMs, allowing agents to see, understand, and control virtual Linux desktops. The platform supports building custom sandbox templates with custom tools and dependencies, and offers cloud browsers powered by Kernel for automated browser interactions.

Products

• E2B SandboxA fast, secure Linux VM created on demand for AI agents, providing an open-source, secure environment with real-world tools for enterprise-grade agents.
• E2B Desktop SandboxProvides secure virtual computers in the cloud for LLMs, enabling AI agents to see, understand, and control virtual Linux desktops with VNC streaming for real-time visual feedback.
• E2B TemplatesAllows defining custom sandboxes with base images, environment variables, files to copy, commands to run, and a start command for pre-configured sandboxes with zero wait time.
• E2B Desktop SDKGives agents a full Linux desktop with mouse, keyboard, and screen capture APIs for computer use agents.
• E2B CLIA command-line interface for creating new E2B templates.
• Kernel (Cloud Browsers)Cloud Chromium instances for automated browser interactions, offering sub-second browser startup, persistent sessions, and handling CAPTCHA solving, residential proxies, authentication, and observability.

Key capabilities

• Isolated sandboxes for safe code execution
• Secure Linux VMs created on demand
• Real-world tools for enterprise-grade agents
• SDKs for easy environment management
• Pre-built templates for popular agents
• Customizable sandbox templates
• Terminal, filesystem, and git access within sandboxes
• Scalability for concurrent sandboxes
• Persistence for sandboxes (pause and resume)
• VNC streaming for virtual desktops
• Mouse, keyboard, and screen capture APIs for desktop agents
• Cloud browsers for automated web interactions
• Sub-second browser startup with Kernel
• Persistent browser sessions with Kernel
• CAPTCHA solving, residential proxies, authentication, and observability for cloud browsers
• LTS 6.1 Linux kernel
• Enterprise-grade security

Use cases

• AI agents
• LLM Training
• MCPs (Multi-Agent Systems)
• Deep Research Agents
• Computer Use Agents
• Automations Agents
• Background Agents
• Reinforcement Learning
• AI data analysis & visualization
• Coding agents (write, debug, refactor code autonomously)
• Vibe coding (code runtime for AI-generated apps)
• Running and evaluating reward functions
• Providing agents with virtual computers
• CI/CD pipelines (review pull requests, generate tests, validate code changes)
• GitHub Actions workflows (testing, validation, AI code reviews)
• Clone repositories, manage branches, and push changes from sandboxes
• Screenshot app endpoints
• Agent data extraction (LLM autonomously browse and extract data)
• Live browser preview

AI approach

E2B provides isolated sandboxes (virtual Linux environments) for AI agents to safely execute code, process data, and run tools. It offers SDKs to manage these environments, pre-built templates for popular agents, and supports custom templates. The platform enables AI agents to conduct research, perform data analysis and visualization, execute code, and control virtual desktops. It integrates with LLMs and provides secure execution environments for CI/CD pipelines.

Tech named: LLM, Claude Code, Codex, Amp, OpenCode, OpenAI Computer Use API, Anthropic Claude, gpt-5.2-mini, LTS 6.1 Linux kernel, Chromium, CDP, BiDi, CUA API, Kernel SDK, Playwright, FastAPI, uvicorn

What it says sets it apart

• Open-source, secure environment
• Enterprise-grade agents
• Fast, secure Linux VM on demand
• Pre-built templates for quick start
• Full Linux environment with terminal, filesystem, git, and package managers
• Isolation from production systems or local machines
• Scalability to run many sandboxes in parallel
• Ability to pause and resume sandboxes to preserve state
• Customizable templates for any agent
• Cloud browsers with sub-second startup and persistent sessions
• Kernel handles CAPTCHA solving, residential proxies, authentication, and observability for cloud browsers
• Templates start faster and use fewer resources than snapshots
• Layering templates with fromTemplate for efficient custom builds

Funding rounds we track

From the AI funding tracker — rounds as reported by the linked publications.