Google Open-Sources AX: An Open Runtime for AI Agents
Google has released AX, an open platform for running AI agents at production scale, published on GitHub under Apache 2.0. It sets up a Git and MCP environment from a plain-text description and scales to billions of tasks.
Google has released AX, a runtime for running AI agents at production scale. The code is on GitHub under Apache 2.0. The platform sets up a working environment with Git and MCP servers on its own. A developer describes what they need in plain language, and the infrastructure is assembled automatically. The authors say the system scales to billions of tasks. An idle agent resumes work in fractions of a second.
Standard container orchestrators are not built for this kind of load. An agent can sit idle for hours waiting for a response, then suddenly consume all available resources at once. If Workspace really does build an environment from a plain-text description, that saves me several hours on each new pipeline. For now these are words in documentation. Google Research as a source deserves attention, but it will need to be tested by hand.
Source: agentexecutor.io
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides

Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles