--- title: Google releases GKE Agent Sandbox and Agent Substrate... description: Google has open-sourced Agent Executor, a distributed runtime for production AI agents that provides durable execution, secure sandboxing, session consistency,... canonical: https://daily.dev/posts/google-releases-gke-agent-sandbox-and-agent-substrate-for-running-ai-agents-at-scale-8my5a98qr twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Google releases GKE Agent Sandbox and Agent Substrate for running AI agents at scale | daily.dev og:description: Google has open-sourced Agent Executor, a distributed runtime for production AI agents that provides durable execution, secure sandboxing, session consistency,... og:url: https://daily.dev/posts/google-releases-gke-agent-sandbox-and-agent-substrate-for-running-ai-agents-at-scale-8my5a98qr og:image: https://api.daily.dev/og/posts/8My5a98QR.png og:image:alt: Google releases GKE Agent Sandbox and Agent Substrate for running AI agents at scale og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Google releases GKE Agent Sandbox and Agent Substrate for running AI agents at scale **[Collections](https://daily.dev/sources/collections)** · 3 min read · 1 upvotes · 0 comments ## Summary Google has open-sourced Agent Executor, a distributed runtime for production AI agents that provides durable execution, secure sandboxing, session consistency, connection recovery, and trajectory branching. It works with LangChain/LangGraph, Google's ADK, and the A2A protocol. Alongside this, GKE Agent Sandbox is now generally available, using gVisor-based kernel isolation with warm pools supporting 300 sandbox allocations per second at under 200ms latency — growing 16x in under five months. A third project, Agent Substrate, is in preview as a Kubernetes extension targeting extreme-scale workloads with hundreds of millions of agents. Together, the three form Google's infrastructure stack for AI agents in production. ## Content Google has made two significant moves for running AI agents in production: open-sourcing Agent Executor, a distributed runtime for reliable agent execution, and releasing GKE Agent Sandbox as generally available. A third project, Agent Substrate, is also in preview and targets extreme-scale workloads. ## Agent Executor Agent Executor is Google's answer to a real gap in the current AI agent ecosystem. Frameworks like LangChain and AutoGen work well for prototyping, but they struggle with the demands of long-running production agents — outages, human-in-the-loop approvals, disconnected clients, and the need to audit or replay execution paths. The runtime addresses these with a handful of concrete capabilities: - **Durable execution** — workflows resume after failures or pauses for human approval, rather than restarting from scratch - **Secure sandboxing** — agent workloads run in isolated environments - **Session consistency** — a single-writer architecture prevents conflicting writes in distributed workflows - **Connection recovery** — clients can reconnect and pick up where they left off - **Trajectory branching** — checkpoints let you fork execution and test alternate paths, useful for debugging and evaluation Agent Executor is framework-agnostic and works with LangChain/LangGraph, Google's ADK, and the Agent2Agent (A2A) protocol. It supports on-prem and cloud deployments and integrates with Google's Antigravity platform. The open-source move is being compared to Google's Kubernetes strategy: release the runtime, drive adoption of the managed cloud service. That's a reasonable read. Analysts note that governance, accountability, and compliance challenges aren't solved by runtime infrastructure alone — those problems sit above the execution layer. Agent Executor is available now in preview on GitHub. ## GKE Agent Sandbox GKE Agent Sandbox is now generally available after a preview at KubeCon NA in November 2025. In under five months, the number of sandboxes running on it grew 16x. Early customers include LangChain and Lovable. The sandbox uses gVisor-based kernel isolation to keep agent workloads secure and separated. The performance numbers are notable: warm pools support 300 sandbox allocations per second at under 200ms latency, and pod snapshots let idle workloads suspend and resume without a cold start. There's also a standby capacity buffer to keep costs manageable. ## Agent Substrate Agent Substrate is a new open-source Kubernetes extension aimed at workloads that push past what the standard Kubernetes control plane can handle — think hundreds of millions of agents making sub-second tool calls. Rather than replacing Kubernetes, it adds a minimal control plane layer on top to optimize scheduling, latency, and density for agentic workloads. It's available in preview on GitHub alongside Agent Executor. Together, these three projects sketch out Google's infrastructure stack for AI agents: Agent Executor handles reliable execution semantics, GKE Agent Sandbox provides the secure runtime environment, and Agent Substrate handles the scheduling and density problems that emerge at genuine scale. ## Similar posts on daily.dev - [Google Announces GKE Agent Sandbox and Hypercluster at Next '26, Positioning Kubernetes as AI Agent](https://daily.dev/posts/google-announces-gke-agent-sandbox-and-hypercluster-at-next-26-positioning-kubernetes-as-ai-agent-cfmrtqf9a) · InfoQ · 0 upvotes · 0 comments - [Kubernetes won the container decade. Google’s Agent Substrate wants the next one.](https://daily.dev/posts/kubernetes-won-the-container-decade-google-s-agent-substrate-wants-the-next-one--htft1i3hx) · The New Stack · 3 upvotes · 0 comments --- Tags: [#kubernetes](https://daily.dev/tags/kubernetes), [#ai-agents](https://daily.dev/tags/ai-agents), [#distributed-systems](https://daily.dev/tags/distributed-systems), [#gke](https://daily.dev/tags/gke) [View this post on daily.dev](https://daily.dev/posts/google-releases-gke-agent-sandbox-and-agent-substrate-for-running-ai-agents-at-scale-8my5a98qr) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Google releases GKE Agent Sandbox and Agent Substrate for running AI agents at scale","url":"https://daily.dev/posts/google-releases-gke-agent-sandbox-and-agent-substrate-for-running-ai-agents-at-scale-8my5a98qr","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/google-releases-gke-agent-sandbox-and-agent-substrate-for-running-ai-agents-at-scale-8my5a98qr"},"datePublished":"2026-05-20T16:01:50.596Z","dateModified":"2026-05-25T18:30:25.006Z","description":"Google has open-sourced Agent Executor, a distributed runtime for production AI agents that provides durable execution, secure sandboxing, session consistency,...","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/google-releases-gke-agent-sandbox-and-agent-substrate-for-running-ai-agents-at-scale-8my5a98qr","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"kubernetes,ai-agents,distributed-systems,gke","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Google releases GKE Agent Sandbox and Agent Substrate for running AI agents at scale"}]} ```