AX – Google’s Open Agentic Orchestrator

On September 18, 2026, Google unveiled AX, an open-source "Agentic Orchestrator," during its I/O developer conference. This platform allows developers to compose, coordinate, and supervise autonomous AI agents across various environments using a single declarative API. AX is built around a modular graph engine that treats each AI component, such as language models, vision encoders, or external services, as a node in a directed acyclic graph.

These nodes communicate via typed messages over a lightweight protobuf channel, with a central scheduler managing dependencies, resource quotas, and safety policies. Unlike AX, AX does not include its own large language model; instead, it can attach to any model compatible with the Gemini-compatible inference endpoint, including open-source alternatives like LLaMA-3.2 or Anthropic’s Claude-3.

Three pre-packaged agents are included with the reference stack: a research assistant capable of web browsing, a customer-service bot integrated with Google Dialogflow, and a data-pipeline manager that orchestrates BigQuery jobs. These examples showcase AX's capability to manage context, maintain state, and invoke external APIs without human intervention.

In benchmark tests, the research assistant completed a 15-step literature-review task in an average of 42 seconds, a 27 percent improvement over a similar workflow built with LangChain on comparable hardware.

The development of agentic AI has transitioned from academic prototypes to commercial products over the past three years. Earlier efforts, such as OpenAI's AutoGPT series, have spurred a plethora of hobbyist projects. By mid-2025, the ecosystem had become fragmented, with developers selecting between LangChain's Python-centric SDK, Microsoft's Semantic Kernel, and a blend of proprietary orchestration layers.

Google's own advancement into agentic AI began with Gemini’s "function calling" feature in 2023, followed by the launch of "Agent Builder" in 2024, a low-code UI for enterprise customers. However, these tools remained closed-source and tied to Google Cloud.

Google's decision to release AX under the Apache 2.0 license marks a strategic shift, attracting over 150,000 developers to its GitHub page within the first 48 hours post-announcement. Contributions already include language bindings for Rust and Go, a WebAssembly runtime for edge deployment, and a safety plug-in enforcing OpenAI’s emerging "agentic alignment" guidelines.

This open-source approach addresses the need for transparency and compliance, particularly in line with the European Union’s AI Act, which mandates the auditability of high-risk AI systems. By providing a public, inspectable orchestration layer, Google offers organizations a compliance path previously unavailable with its closed-source alternatives.