AX is a declarative, distributed orchestrator designed to manage stateful, autonomous AI agent workloads on Kubernetes.
It treats agents as long-running, stateful actors rather than stateless microservices, providing native capabilities for task suspension, resumption, and isolated sandboxing.
- 01
Declarative primitives (Task, Workspace, Gateway, Model)
- 02
Sub-second task suspension and resumption
- 03
Isolated sandboxed execution with resource limits
- 04
Egress network control via host allowlists
- 05
Redis-backed state management for high-volume task handling
Open Source
Released under the Apache License 2.0; free to use, though users incur costs for the underlying Kubernetes infrastructure, compute, storage, and model API usage.
Best for
Enterprises and researchers managing large-scale, long-running, or bursty agent fleets that require stateful checkpointing and strict infrastructure governance.
Not ideal for
Small teams or solo developers looking for a quick-start agent framework, as it requires significant operational overhead including Kubernetes, Redis, and container registry management.
AX is a powerful, infrastructure-focused runtime for production-grade agent orchestration, but it is currently in early alpha development with expected breaking changes and a steep operational barrier to entry.
In September 2026 (v0.3.0), the project underwent a major architectural redesign, moving task state from Kubernetes custom resources to Redis Streams and splitting the system into three binaries (ax-server, ax-controller, and ax-task-runner) to improve scalability.
As of
Is AX (Agent Executor) free?+
Yes - AX (Agent Executor) is Open Source. Released under the Apache License 2.0; free to use, though users incur costs for the underlying Kubernetes infrastructure, compute, storage, and model API usage.
Is AX (Agent Executor) open source?+
Yes - AX (Agent Executor) is open source. Released under the Apache License 2.0; free to use, though users incur costs for the underlying Kubernetes infrastructure, compute, storage, and model API usage.
Who is AX (Agent Executor) best for?+
Enterprises and researchers managing large-scale, long-running, or bursty agent fleets that require stateful checkpointing and strict infrastructure governance.
Who is AX (Agent Executor) not ideal for?+
Small teams or solo developers looking for a quick-start agent framework, as it requires significant operational overhead including Kubernetes, Redis, and container registry management.
What are the best AX (Agent Executor) alternatives?+
The closest AX (Agent Executor) alternatives on ai.dosa.dev are Jules, OpenHands, SWE-agent - all listed under Autonomous & Async Agents.
Who makes AX (Agent Executor)?+
AX (Agent Executor) is developed by Google. It is listed in the Autonomous & Async Agents category on ai.dosa.dev.
Favorite this tool to revisit it later, or Zap it to contribute to the public vote count.