Cloudflare Open-Sources AI Workspace Platform to Secure Enterprise Agent Workflows
The new Cloudflare OS replaces broad API key access with a capability-based model for secure internal data integration.
Cloudflare has open-sourced Cloudflare OS, an AI workspace platform designed to provide employees with secure, governed access to internal company systems and data. The release aims to solve the critical security gap enterprises face when integrating AI agents with sensitive corporate knowledge.
Originally developed for internal use at Cloudflare, the platform is now available under the Apache 2.0 license. Cloudflare OS allows organizations to build AI agents and full-stack applications without requiring professional developers, integrating directly with the Cloudflare AI Gateway to support a variety of AI model providers. To ensure security, the system utilizes a "Gatekeeper" layer—implemented via service-specific Cloudflare Workers—that interprets APIs and resources. This architecture ensures that agents only access explicitly granted data, removing the need for raw API keys which often grant overly broad permissions.
The Enterprise Context Gap
Most enterprises have struggled to deploy AI agents because standard authentication methods, such as OAuth or API keys, frequently lead to data leakage or excessive access. Beyond security, most AI tools lack "business context"—the specific terminology, internal policies, and unique workflows of a company. This deficiency forces employees to manually re-explain corporate context in every session, hindering productivity.
Rita Koslov, VP for developers and AI at Cloudflare, noted that traditional API keys give agents too much breadth. "A capability-access-based approach lets us grant one specific resource, then record exactly what the agent observed, and verify that anyone who sees its work is also allowed to access the source," Koslov said.
Why It Matters
Cloudflare OS attempts to solve the "last mile" of enterprise AI by providing an audited environment where agents have precise access to company knowledge. By moving beyond simple LLM wrappers to a full operating system for agentic workflows, Cloudflare is positioning itself as the underlying infrastructure layer for the agent era. This shift prioritizes the ownership of corporate context over the raw power of the model.
Vladimir Beskorovainyi, enterprise AI architect and founder of Besk Tech, argues that the industry's focus on token-based pricing is misplaced. "The model is the commodity part," Beskorovainyi said. "What costs real money is the curated context, and nobody budgets for the fact that it starts decaying the day it is written."
What's Next
As organizations adopt the platform, the focus will shift toward how effectively companies can curate and maintain their internal business context within the Cloudflare OS framework. While the core architecture utilizing Cloudflare Workers and Durable Objects is now public, the industry will be watching to see if this capability-based model becomes the standard for securing agentic access across the broader enterprise ecosystem.