AIGoogle's planned open agent runtime aims for billions of agents per cluster
A developer involved in the work says the team is rethinking Kubernetes for stateful agent workloads and fast resumption, alongside building Google's open agent orchestrator and runtime.
TLDR
A developer working on the effort announced two projects: a Kubernetes rethink focused on preserving state and resuming agent workloads quickly, and Google's planned open agent orchestrator and runtime. In a follow-up reply, the developer outlined goals for a compute runtime that scales to billions of agents in a cluster and fits 10–20x more sandboxes into an existing setup.
Combined views
53.5K
3 Sources, first seen 5h ago
Google's planned open agent runtime aims for billions of agents per cluster
A developer involved in the work says the team is rethinking Kubernetes for stateful agent workloads and fast resumption, alongside building Google's open agent orchestrator and runtime.
TLDR
A developer working on the effort announced two projects: a Kubernetes rethink focused on preserving state and resuming agent workloads quickly, and Google's planned open agent orchestrator and runtime. In a follow-up reply, the developer outlined goals for a compute runtime that scales to billions of agents in a cluster and fits 10–20x more sandboxes into an existing setup.