● Covered by 2 sources · 2 reportsMedium impact2 neutral
Google Introduces AX, an Open Agentic Orchestrator for Scalable Agent Workloads
Key points
- AX orchestrates agent workloads with isolation and scalability.
- It provides primitives for sandboxing, workspace setup, and network control.
- AX manages model configurations and secrets in a centralized manner.
- It runs on Agent Substrate for high density and fast actor lifecycles.
- AX is Apache 2.0-licensed.
- AX is hosted at agentexecutor.io and on GitHub as google/ax.
- AX provides sub-second task suspension and resumption.
- AX has four core declarative primitives: Task, Workspace, Gateway, and Model.
Introducing AX for Agent Workloads
Google has launched AX, an open agentic orchestrator. This new system is built to manage and scale agent-based tasks, which are characterized by their stateful nature, need for strict isolation, and interactions with model APIs and tool servers. AX aims to provide a dedicated solution for these workloads, distinguishing them from traditional microservices or batch jobs.
Core Capabilities of AX
AX offers four primary primitives for declarative management of agent tasks. These include isolated execution within sandboxes with defined CPU and memory limits, making tasks cheap to create, suspend, and discard. It also simplifies workspace setup by allowing users to declare Git repositories, MCP servers, and skills an agent requires, which AX then configures before task execution.
Network and Model Management
The orchestrator includes gateway features for defining and managing network policies, enabling traffic to be locked down to explicit allowlists of hosts and ports, and injecting credentials. Additionally, AX provides a centralized place for configuring models, their parameters, and secrets, allowing for easy key rotation or model version pinning.
Scalability and Underlying Infrastructure
AX is designed to scale to billions of tasks. It operates on top of Agent Substrate, a compute runtime specifically developed for high density and rapid stateful actor lifecycles. This foundation allows AX to efficiently handle the unique demands of agent workloads at scale.
Updates
- AX is Apache 2.0-licensed.
- AX is hosted at agentexecutor.io and on GitHub as google/ax.
- AX provides sub-second task suspension and resumption.
- AX has four core declarative primitives: Task, Workspace, Gateway, and Model.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
How outlets covered it
Google has open-sourced AX, an Apache 2.0-licensed orchestrator and declarative runtime designed to execute and scale autonomous AI agent workloads. AX addresses the operational challenges of stateful, bursty, and long-running AI agents by providing sub-second task suspension and resumption, optimizing compute resource utilization.
Google has released AX, an open agentic orchestrator designed to manage and scale agent-based workloads. AX provides sandboxed execution, workspace setup, network policies, and model configuration, addressing the unique requirements of agent workloads that differ from traditional microservices or batch jobs.