Anthropic has taken a major step in enhancing its Claude Managed Agents by introducing self-hosted sandboxes and model context protocol (MCP) tunnels. These features, unveiled at the recent Code with Claude developer conference in London, aim to give businesses more control over their security measures and operational environments.
The self-hosted sandboxes, now available in public beta, allow organizations to run sensitive files and service executions within their own infrastructure or through designated managed providers. While Anthropic manages orchestration and error recovery, the execution of tools occurs within the enterprise's controlled environment. This dual-layer approach helps uphold existing security protocols, enabling enterprises to dictate their network policies and audit logging.
MCP tunnels, currently in a research preview, add another layer by allowing users to request access. This feature enables Claude Managed Agents to interact with tools and services tailored specifically to the enterprise's needs. Both features are designed to protect sensitive information while providing the flexibility required for various operational tasks.
The recent conference also marked Anthropic's first developer gathering outside the US, underscoring the company's commitment to expanding its presence in Europe. The event showcased the evolving nature of AI-driven enterprise solutions and the need for companies to adapt to stricter security demands.
With the integration of these features, companies can specify resource allocation for intensive tasks such as complex builds or image generation, boosting operational efficiency. Supported sandbox clients include notable providers like Cloudflare, Daytona, Modal, and Vercel, allowing enterprises to select platforms that fit their existing infrastructures.
As businesses increasingly rely on AI technologies, Anthropic's Claude is positioned to offer a more secure and customizable environment for deploying managed agents. The ability to maintain internal controls while utilizing advanced AI capabilities makes Anthropic's offerings particularly attractive in a market that prioritizes security and compliance.
The implications of these features extend beyond operational improvements. By equipping enterprises with tools to uphold stringent security measures, Anthropic is establishing a new standard for safely integrating AI solutions into business workflows. As adoption rates increase, these self-hosted solutions could become essential for navigating the complexities of AI deployment in sensitive environments.
