Cloudflare's New Dynamic Workers Promise A 100x Faster Sandbox For AI Workflows

As AI has become more engrained as part of the software development pipeline, there’s a need for tooling that enables developers to get the most out of deployed AI agents. Cloudflare is doing its part to meet this need with the release of a Dynamic Worker Loader that creates a lean sandbox environment for AI agents, which is latest feature the company has implemented to its Code Mode offering.

With Code Mode, AI agents are able to access tools using a TypeScript API instead of having direct access to them. This made tools more accessible to these agents and made it possible to string together several API calls, leading to a more effective process that’s easier to scale at the level Cloudflare customers need. However, the downside to this method is that it requires having to run these agents in a container, which can be slow to start up and potentially eat up a lot of resources, to ensure security.


The new Dynamic Worker Loader looks to drastically reduce the pain points of using Code Mode in a secure way. It’s an API that “allows a Cloudflare Worker to instantiate a new Worker, in its own sandbox, with code specified at runtime, all on the fly.” The company says that moving to this new kind of sandboxing will net developers a 100x speed increase alongside using up 10-100x less memory than typically seen with container approaches.

The benefits extend beyond the speed boost, though. Developers will no longer have a limit to how many sandboxes are running at the same time, which usually occurs when using containers. Moreover, Cloudflare is boasting “battle-hardened security” as it claims its systems deploy security patches with rapid speed and utilize a security architecture that employs a “custom second-layer sandbox with dynamic cordoning of tenants based on risk assessments.”

The Dynamic Worker Loader feature is currently in open beta but will likely be available to all developers sooner rather than later.