글로벌 커넥티비티 클라우드 기업 클라우드플레어(Cloudflare)가 업계 최초 원격 모델 컨텍스트 프로토콜(Model Context Protocol; MCP) 서버를 공개하며, AI 에이전트 구축을 위한 워크플로우(Workflows) 기능의 정식 출시(GA)와 듀러블 오브젝트(Durable Objects) 무료 제공을 통해 개발자들이 쉽고 빠르게 AI 에이전트를 구축할 수 있도록 지원한다.
Accelerate AI Agent Development with Remote MCP Servers
Cloudflare, a global connectivity cloud company, announced innovative capabilities to accelerate the development of AI agents.
Cloudflare announced on the 14th that it is releasing the industry's first remote Model Context Protocol (MCP) server, and is supporting developers to build AI agents easily and quickly by announcing the general availability (GA) of Workflows for building AI agents and providing Durable Objects for free.
AI agents are AI-based systems that operate autonomously without direct user intervention, making decisions, and adapting to environmental changes. On the other hand, many companies have struggled with the development of existing AI agents due to high costs and complex implementation tasks.
To address these issues, Cloudflare enables AI agents to securely connect to external services via remote MCP servers, running on the cloud to enable fast and reliable interactions.
Additionally, we have simplified secure authentication and authorization management for AI agents through partnerships with Auth0, Stytch, and WorkOS.
Cloudflare offers Durable Objects for free to enable AI agents to maintain persistent context and operate more intelligently. This feature, previously only available on paid plans, is now available for free, greatly expanding accessibility for developers.
In addition, the officially released workflow function has enabled AI-based work to be processed more systematically and reliably. Workflow can maintain work for a certain period of time or support automatic execution, allowing for more efficient implementation of complex work flows such as travel reservations, financial analysis, and product recommendations.
Existing cloud AI deployment methods have high costs due to the difficulty in predicting usage. In contrast, Cloudflare's serverless AI platform reduces resources to zero when there is no usage and can immediately expand to a global level when needed, contributing to reducing operating costs.
Cloudflare currently deploys GPUs in over 190 cities around the world, enabling AI to run closer to users, delivering AI experiences with minimal latency.
“AI agents are the future of AI technology, and Cloudflare provides the optimal platform to build them quickly and scale them massively,” said Matthew Prince, CEO of Cloudflare. “We will drive innovation in agentic AI on the world’s most interconnected network.”