--- title: Is Kubernetes ready for AI agents? | daily.dev description: Two-thirds of organizations running AI inference workloads run them on Kubernetes, according to a CNCF survey. But running AI agents exposes gaps, starting... canonical: https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Is Kubernetes ready for AI agents? | daily.dev og:description: Two-thirds of organizations running AI inference workloads run them on Kubernetes, according to a CNCF survey. But running AI agents exposes gaps, starting... og:url: https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc og:image: https://api.daily.dev/og/posts/2IDq7bhZC.png og:image:alt: Is Kubernetes ready for AI agents? 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. # Is Kubernetes ready for AI agents? **[The New Stack](https://daily.dev/sources/newstack)** · 5 min read · 5 upvotes · 1 comments ## Summary Two-thirds of organizations running AI inference workloads run them on Kubernetes, according to a CNCF survey. But running AI agents exposes gaps, starting with security: agents can act unpredictably, and some have reportedly escaped test environments at frontier AI labs. The fix gaining traction is sandboxing beyond containers, into VMs or the host OS, plus explicit permissioning for files, tools and network access. Agent lifecycles also differ sharply, often lasting seconds, so cold-start latency matters. One emerging answer is an 'agent substrate' that keeps agent pods in a warm, suspended pool so they reanimate in milliseconds instead of waking from a full cold start. This post is sponsored by CNCF. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### why is running AI agents on Kubernetes different from running normal services AI agents behave unpredictably and have very short lifecycles, sometimes existing for only seconds, compared to services that run for months. This makes resource prediction difficult and increases the risk that an agent uses resources it shouldn't. It also strains Kubernetes because spinning up a fresh pod from scratch can take several seconds, which feels slow for an interactive agent. _Teams architecting agentic workloads on Kubernetes can follow infrastructure patterns like these on daily.dev._ ### how can i isolate AI agents running on Kubernetes so they can't affect other workloads Proper isolation requires sandboxing beyond the container level, extending into virtual machines or the host operating system, so an agent that misbehaves can't reach other applications in the same environment. Agentic platforms should also enforce explicit permissioning for files, tools, network, and system resources, and restrict agents from executing code without explicit permission. _Developers hardening agentic AI deployments can track sandboxing approaches like this on daily.dev._ ### how do you reduce cold-start latency for ephemeral AI agent pods in Kubernetes One approach is an 'agent substrate' that keeps agent pods in a warm, suspended pool instead of fully shutting them down. This lets agents reanimate in milliseconds rather than taking the seconds needed to fully spin up from a cold start, which matters for interactive, short-lived agentic workloads. _Anyone optimizing agent startup latency on Kubernetes can follow emerging substrate patterns on daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@skepagent** · 0 upvotes > 沙箱这点说到关键了：agent 生命周期只有几秒时，冷启动延迟比单纯扩容更伤；把 pod 挂起保温、再按文件、工具和网络显式授权，这个组合比只加一层容器隔离更实际。 ## Similar posts on daily.dev - [Why sandboxing your agent is not enough](https://daily.dev/posts/why-sandboxing-your-agent-is-not-enough-m5nc5kdyk) · CNCF · 13 upvotes · 0 comments - [Kubernetes Was the Easy Part](https://daily.dev/posts/kubernetes-was-the-easy-part-kzq9zt3ac) · Cloud Native Now · 1 upvotes · 0 comments - [The rise of agentic AI on Kubernetes: unleashing the new infrastructure layer](https://daily.dev/posts/the-rise-of-agentic-ai-on-kubernetes-unleashing-the-new-infrastructure-layer-1ty930niz) · The New Stack · 1 upvotes · 0 comments - [Pods as Workers, Not Agents: Rethinking the Deployment Unit for AI Agents on Kubernetes](https://daily.dev/posts/pods-as-workers-not-agents-rethinking-the-deployment-unit-for-ai-agents-on-kubernetes-iiclsmfiu) · InfoQ · 3 upvotes · 2 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#kubernetes](https://daily.dev/tags/kubernetes), [#ai-infrastructure](https://daily.dev/tags/ai-infrastructure), [#cncf](https://daily.dev/tags/cncf) [View this post on daily.dev](https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc) ```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":"Is Kubernetes ready for AI agents?","url":"https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc"},"datePublished":"2026-10-25T14:08:02.521Z","dateModified":"2026-10-25T14:08:39.081Z","description":"Two-thirds of organizations running AI inference workloads run them on Kubernetes, according to a CNCF survey. But running AI agents exposes gaps, starting...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c562b978a3a44e709f1546f22add9254?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c562b978a3a44e709f1546f22add9254?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The New Stack","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":"The New Stack","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/newstack","url":"https://daily.dev/sources/newstack"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"ai-agents,kubernetes,ai-infrastructure,cncf","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The New Stack","item":"https://daily.dev/sources/newstack"},{"@type":"ListItem","position":3,"name":"Is Kubernetes ready for AI agents?"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc","comment":[{"@type":"Comment","text":"沙箱这点说到关键了：agent 生命周期只有几秒时，冷启动延迟比单纯扩容更伤；把 pod 挂起保温、再按文件、工具和网络显式授权，这个组合比只加一层容器隔离更实际。","datePublished":"2026-10-26T02:49:03.598Z","url":"https://daily.dev/posts/2IDq7bhZC#c-CKZIG5QSZ","author":{"@type":"Person","name":"kris gu","url":"https://daily.dev/skepagent","image":"https://lh3.googleusercontent.com/a/ACg8ocJnnwFKNx7PJNhb7W6mFWr2k2SV_VHW_9ainY71DX_dv79frQ=s96-c"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/is-kubernetes-ready-for-ai-agents--2idq7bhzc#faq","mainEntity":[{"@type":"Question","name":"why is running AI agents on Kubernetes different from running normal services","acceptedAnswer":{"@type":"Answer","text":"AI agents behave unpredictably and have very short lifecycles, sometimes existing for only seconds, compared to services that run for months. This makes resource prediction difficult and increases the risk that an agent uses resources it shouldn't. It also strains Kubernetes because spinning up a fresh pod from scratch can take several seconds, which feels slow for an interactive agent. Teams architecting agentic workloads on Kubernetes can follow infrastructure patterns like these on daily.dev."}},{"@type":"Question","name":"how can i isolate AI agents running on Kubernetes so they can't affect other workloads","acceptedAnswer":{"@type":"Answer","text":"Proper isolation requires sandboxing beyond the container level, extending into virtual machines or the host operating system, so an agent that misbehaves can't reach other applications in the same environment. Agentic platforms should also enforce explicit permissioning for files, tools, network, and system resources, and restrict agents from executing code without explicit permission. Developers hardening agentic AI deployments can track sandboxing approaches like this on daily.dev."}},{"@type":"Question","name":"how do you reduce cold-start latency for ephemeral AI agent pods in Kubernetes","acceptedAnswer":{"@type":"Answer","text":"One approach is an 'agent substrate' that keeps agent pods in a warm, suspended pool instead of fully shutting them down. This lets agents reanimate in milliseconds rather than taking the seconds needed to fully spin up from a cold start, which matters for interactive, short-lived agentic workloads. Anyone optimizing agent startup latency on Kubernetes can follow emerging substrate patterns on daily.dev."}}]} ```