Embedded function calling in Workers AI: easier, smarter, faster

Cloudflare introduces embedded function calling in Workers AI and the open-source @cloudflare/ai-utils package, enabling LLM inference and function execution within the same runtime to reduce latency and complexity. The announcement highlights tools like runWithTools for multi/recursive tool calls, autoTrimTools to reduce token usage, and createToolsFromOpenAPISpec to dynamically generate tool schemas from OpenAPI specs, aiming to streamline building AI agents with a better developer experience.