Azure Adds AI Agents With No Cold Start

Azure Functions adds serverless AI agents that run with no cold start delays or extra costs on its Flex Consumption plan.

Microsoft has released a new serverless agents runtime for Azure Functions, now available in public preview following its announcement at the Build 2026 conference. This new capability allows developers to create and run automated, intelligent agents directly within the serverless environment. The design emphasizes simplicity and integration. Developers define their agents using straightforward markdown files (`.agent.md`) and configure their behavior with YAML-based triggers. This approach lowers the barrier to entry for creating complex automations. The runtime is built for extensive connectivity, providing access to the Microsoft Cloud Platform (MCP) server and a library of over 1,400 connectors for various services and APIs. For security, each agent executes within a sandboxed environment, ensuring that its operations are isolated and cannot interfere with other system components. This framework is designed to streamline the development of powerful, event-driven workflows that can orchestrate tasks across a wide range of enterprise and cloud services.

The most compelling features for developers and businesses are the performance and cost implications. The Azure Functions team confirmed that the new agents runtime introduces zero additional cold start overhead. Cold starts are a notorious pain point in serverless computing, causing delays when a function is invoked after a period of inactivity. By eliminating this latency, Azure makes its serverless agents far more practical for real-time and user-facing applications. Equally important, Microsoft is not charging a premium for this advanced functionality. The agents run on the standard Flex Consumption plan, meaning users pay only for the compute resources they consume, with no extra fees. This combination of high performance and predictable, usage-based pricing removes major barriers to adoption. It enables developers, CTOs, and IT teams to experiment with and deploy sophisticated AI-driven automations without risking performance bottlenecks or unexpected spikes in their cloud bills, positioning Azure as a highly competitive option for building next-generation serverless applications.

The new runtime removes two major barriers to adopting serverless for AI and automation: cold start latency and unpredictable costs. This makes it cheaper and more practical for developers to build responsive, intelligent applications on Azure.

Companies can now build and scale sophisticated AI-driven automations more affordably. The elimination of cold starts improves user experience for time-sensitive applications, while the lack of a price premium keeps cloud spending predictable under the Flex Consumption model.