Introduces new infrastructure to help developers build, deploy, and scale autonomous AI agents for enterprise workloads
Cloudflare has announced a major expansion of its Agent Cloud, aimed at accelerating the shift from experimental AI applications to production-grade autonomous agents. As software development evolves beyond traditional applications toward intelligent, self-operating systems, the company is positioning its platform as the backbone for the emerging agentic web.
The latest enhancements are designed to address key challenges in scaling AI agents—particularly around compute efficiency, deployment, persistence, and security. While early AI adoption focused heavily on chatbots, the industry is now moving toward advanced agents capable of reasoning, executing multi-step tasks, and interacting across systems. However, existing infrastructure models remain costly and inefficient for large-scale deployment.
To overcome these limitations, Cloudflare has introduced Dynamic Workers, a purpose-built compute model that enables agents to execute code in isolated, secure environments at significantly higher speeds and lower costs compared to traditional containers. This allows developers to run millions of concurrent agent tasks without performance bottlenecks.
“The way people build software is fundamentally changing. We are entering a world where agents are writing and executing code, and they need a secure, scalable home to operate seamlessly,” said Matthew Prince, co-founder and CEO of Cloudflare.
The company has also launched Artifacts, a Git-compatible storage system tailored for AI-driven development. This enables scalable code and data management, allowing agents to create, manage, and access repositories seamlessly. Complementing this is the general availability of Sandboxes—isolated Linux environments that provide agents with full operating system capabilities for more complex workflows.
Another key innovation is Think, a framework within the Agents SDK that supports long-running, persistent workflows. This ensures agents can handle complex, multi-step tasks beyond simple prompt-based interactions.
Additionally, Cloudflare is expanding its AI ecosystem through deeper integrations with models from providers like OpenAI, offering developers flexibility to switch between models with minimal effort.
With these advancements, Cloudflare aims to simplify the development of scalable, secure, and cost-efficient AI agents, enabling enterprises to unlock the full potential of autonomous systems in real-world applications.