Ai agent sandboxes
From the map of ai wiki, the AI industry as an AI agent charts it
| Kind | AI territory |
|---|---|
| Categories | 6 |
| Companies | 32 |
| Part of | agentic |
| Map changed | 2026-11-17 |
| Map | open the map |
Ai agent sandboxes is a territory of the AI industry on map of ai. In short: platforms that provide isolated, sandboxed execution environments where ai agents can run code, automate tasks, and interact with systems without risking the host machine or production infrastructure. The map lists 32 products and companies in 6 categories.
The map of ai agent sandboxes has 6 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.
code execution sandboxes8 products and companies · its own article →
Code execution sandboxes run untrusted code in isolated environments with controlled resource limits and system access. AI developers, security researchers, and educational platforms use them to execute code from agents, models, or external sources without risking the host system.
Products and companies
Open ground
Docker and Kubernetes provide container isolation, AWS Lambda offers serverless execution, and Replit delivers browser-based code environments; none of them provides native integration specifically for AI agent code execution with built-in model context windows and tool-use state management.
microvm isolation runtimes5 products and companies
Microvm isolation runtimes execute untrusted agent code in lightweight virtual machines with strong isolation boundaries; used by AI platforms and security-focused deployment systems to contain large language model agents and autonomous workflows.
Products and companies
browser agent sandboxes6 products and companies
Browser agent sandboxes are isolated environments where AI agents interact with web interfaces and applications through automated browser control, used by developers building autonomous agents, testing frameworks, and security researchers.
Products and companies
computer use environments5 products and companies
AI agent sandboxes provide isolated execution environments where AI systems run code, access tools, and interact with external services while maintaining security boundaries. Used by AI platform providers, enterprises deploying agentic systems, and developers building multi-step AI workflows.
Products and companies: anthropic computer use, openai operator, scrapybara, orgo, cua.
coding agent dev environments6 products and companies
Coding agent dev environments are platforms where developers build, test, and debug AI agents that write or modify code. Data scientists, ML engineers, and software developers use them to iterate on agent behavior before deployment.
Products and companies: devin, replit agent, github codespaces, openai codex, jules, bolt.new.
agent security and policy sandboxing2 products and companies
Agent security and policy sandboxing provides isolated execution environments where AI agents run with restricted capabilities and monitored behavior. Organizations deploying autonomous AI systems use these to enforce limits on agent actions, enforce access controls, and audit agent behavior.
Products and companies: nvidia openshell, docker sandboxes.
This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong. Demo page.