Cloudflare, the global leader in connectivity cloud infrastructure, has unveiled a powerful suite of tools aimed at accelerating the development and deployment of AI agents. This includes the first remote Model Context Protocol (MCP) server, general availability of Workflows, and a new free tier for Durable Objects. Together, these offerings radically simplify how developers build autonomous, intelligent systems—ushering in a new era of scalable, cost-efficient agentic AI.
The Rising Role of AI Agents
AI agents—autonomous systems capable of reasoning, decision-making, and task execution—are emerging as the next frontier of artificial intelligence. These agents can perform actions like sending emails, booking meetings, or adjusting codebases with minimal human intervention. However, developing truly scalable AI agents has remained a challenge due to infrastructure limitations and high deployment costs.
To build effective agents, organizations need access to three key pillars:
- AI models for intelligent reasoning
- Workflows for action execution
- APIs for real-world interaction
Cloudflare’s latest developer platform release addresses these challenges head-on, delivering the infrastructure needed to build AI agents in minutes—not months.
Announcements from Cloudflare
1. Remote MCP Server – Unlocking Smart AI Actions
Cloudflare now supports the industry’s first remote MCP server, making it possible for AI agents to interact with external services over the Internet, rather than just locally.
- MCP (Model Context Protocol) is an open-source standard that allows AI agents to take direct action rather than merely responding with instructions.
- Cloudflare’s MCP implementation enables persistent user context, ideal for long-term, personalized AI experiences.
- Through integrations with Auth0, Stytch, and WorkOS, developers can easily manage authentication and permissions for secure agent deployment.
This opens the door for secure, scalable AI agents that can autonomously perform tasks like sending emails or executing backend updates without user micromanagement.
2. Free Durable Objects – State Management for All
Previously locked behind paid plans, Durable Objects are now part of Cloudflare’s free tier, significantly lowering the barrier for developers to build context-aware AI agents.
- Durable Objects offer persistent, serverless compute with storage, perfect for agents that must remember past interactions.
- Cloudflare’s global network ensures low-latency performance and scalability to support millions of concurrent interactions.
- Ideal for use cases such as customer service bots, personalized AI shopping assistants, or dynamic recommendation systems.
3. Workflows – General Availability for Multi-Step Applications
Workflows, now generally available, allows developers to orchestrate complex, multi-step applications with AI at the core.
- Enables agents to retry, persist, and run processes over extended durations—from minutes to weeks.
- Example: A travel-booking agent that searches flights, confirms details, books tickets, and sends itineraries.
By combining Workflows with Durable Objects and AI inference, developers can create robust, long-running AI agents without the overhead of managing backend infrastructure.
Scaling Responsively: Pay Only for What You Use
Traditional cloud platforms require constant provisioning of infrastructure to accommodate peak usage—making AI deployment expensive and inefficient. In contrast:
- Cloudflare’s serverless AI platform scales automatically based on demand, from zero to global scale in milliseconds.
- Organizations benefit from elastic, usage-based pricing, optimizing cost-efficiency while maintaining performance.
- Cloudflare’s AI inference layer, powered by GPUs in 190+ global cities, ensures ultra-low latency by processing requests close to end users.
“These low-cost, easy-to-use options could empower more businesses to adopt and experiment with agentic AI.”
— Kate Holterhoff, Senior Analyst, RedMonk
Cloudflare’s Vision for Agentic AI
“Cloudflare is the best place to build and scale AI agents. Period.”
— Matthew Prince, CEO and Co-founder, Cloudflare
By integrating compute, context, and coordination into a single developer-friendly platform, Cloudflare is democratizing access to next-gen AI agent development tools. With its interconnected network, serverless architecture, and infrastructure-level support for AI workloads, Cloudflare is laying the foundation for a world powered by autonomous digital agents
With the launch of remote MCP servers, free Durable Objects, and the general availability of Workflows, Cloudflare is poised to lead the AI agent revolution. Developers can now rapidly build, deploy, and scale intelligent, autonomous systems—without the complexity or costs traditionally associated with AI infrastructure. As AI agents become foundational to future digital experiences, Cloudflare provides the tools to innovate at speed and scale.
