Add "Running Agents on Kubernetes with Agent Sandbox" blogpost#54927
Add "Running Agents on Kubernetes with Agent Sandbox" blogpost#54927k8s-ci-robot merged 5 commits intokubernetes:mainfrom
Conversation
✅ Pull request preview available for checkingBuilt without sensitive environment variables
To edit notification comments on pull requests, go to your Netlify project configuration. |
|
@bexxmodd thanks for the feedback! Would you take another look? |
|
@lmktfy @nate-double-u this is ready for doc review. PTAL |
lmktfy
left a comment
There was a problem hiding this comment.
Thanks for the PR. Here's a review on the content.
530b09d to
bcaddf2
Compare
|
@lmktfy thanks for the review! I've addressed all the feedback. PTAL |
lmktfy
left a comment
There was a problem hiding this comment.
/lgtm
/approve
/hold
Remaining feedback can be addressed post-merge.
Ok to uphold at will.
|
|
||
| The landscape of artificial intelligence is undergoing a massive architectural shift. In the early days of generative AI, interacting with a model was often treated as a transient, stateless function call: a request that spun up, executed for perhaps 50 milliseconds, and terminated. | ||
|
|
||
| Today, we are witnessing AI v2 eating AI v1. We are moving from short-lived, isolated tasks to deploying multiple, coordinated AI agents that run constantly. These autonomous agents need to maintain context, use external tools, write and execute code, and communicate with one another over extended periods. |
There was a problem hiding this comment.
| Today, we are witnessing AI v2 eating AI v1. We are moving from short-lived, isolated tasks to deploying multiple, coordinated AI agents that run constantly. These autonomous agents need to maintain context, use external tools, write and execute code, and communicate with one another over extended periods. | |
| Today, the world is witnessing AI v2 eating AI v1. The ecosystem is moving from short-lived, isolated tasks to deploying multiple, coordinated AI agents that run constantly. These autonomous agents need to maintain context, use external tools, write and execute code, and communicate with one another over extended periods. |
(nit)
Nice to fix this use of "we" but not essential.
|
|
||
| Whether it’s a 50-millisecond stateless task, or a multi-week, mostly-idle collaborative process, extending Kubernetes with primitives designed specifically for isolated stateful singletons allows us to leverage all the robust benefits of the cloud-native ecosystem. | ||
|
|
||
| The Agent Sandbox project is open source and community-driven. If you are building AI platforms, developing agentic frameworks, or are interested in Kubernetes extensibility, we invite you to get involved: |
|
|
||
| To bridge this gap, SIG Apps is developing [agent-sandbox](https://github.com/kubernetes-sigs/agent-sandbox). The project introduces a declarative, standardized API specifically tailored for singleton, stateful workloads like AI agent runtimes. | ||
|
|
||
| At its core, the project introduces the `Sandbox` CRD. It acts as a lightweight, single-container environment built entirely on Kubernetes primitives, offering: |
There was a problem hiding this comment.
| At its core, the project introduces the `Sandbox` CRD. It acts as a lightweight, single-container environment built entirely on Kubernetes primitives, offering: | |
| At its core, the project introduces the Sandbox CRD. It acts as a lightweight, single-container environment built entirely on Kubernetes primitives, offering: |
We can delete the backticks after publication, but before is even better (we don't use them around API types).
|
|
||
| Because the AI space is moving incredibly quickly, we built an Extensions API layer that enables even faster iteration and development. | ||
|
|
||
| Starting a new pod adds about a second of overhead. That's perfectly fine when deploying a new version of a microservice, but when an agent is invoked after being idle, a one-second cold start breaks the continuity of the interaction. It forces the user or the orchestrating service to wait for the environment to provision before the model can even begin to think or act. `SandboxWarmPool` solves this by maintaining a pool of pre-provisioned Sandbox pods, effectively eliminating cold starts. Users or orchestration services can simply issue a `SandboxClaim` against a `SandboxTemplate`, and the controller immediately hands over a pre-warmed, fully isolated environment to the agent. |
There was a problem hiding this comment.
If these are API types, drop the backticks. Especially for formal launches and similar announcements, following the style guide aims to help people build a good metal model of all our jargon.
On the other hand, field names (eg generateName) are written with the same case that they have within a manifest, and do go inside backticks.
|
LGTM label has been added. DetailsGit tree hash: a036d7f05eedb13781dc0d56b805e5cb521f0df2 |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: bexxmodd, lmktfy The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
/unhold |
|
Addressing nits in #54962 |
Partnering with @justinsb