Oh no! Your software is unreachable, buried beneath a number of connection hops. How do you pinpoint the damaged hyperlink? How do you generate an alert or bug report from Prometheus when sure circumstances are met?
You should roll out a brand new model of your software. How do you execute a progressive rollout utilizing Argo Rollouts? How do you safely allow zero belief community safety when your software scales past a single cluster or cloud?
With so many tasks within the cloud native ecosystem, how do you determine which of them are proper in your wants and layer them along with correct configuration administration?
Sound acquainted? We hear these questions on a regular basis from platform and DevOps engineers working with Cloud Native Computing Basis (CNCF) tasks like Kubernetes, Envoy, Istio, Prometheus, and Argo.
So why not construct AI brokers to deal with widespread challenges and assist engineers and clients? Why not create a catalog of AI brokers for the cloud native ecosystem, enabling everybody to run, construct and share AI-driven options?
What’s kagent?
You’ve in all probability interacted with ChatGPT, an AI chatbot that gives pure language responses. Agentic AI, nonetheless, goes past easy chat interactions. It makes use of superior reasoning and iterative planning to autonomously clear up advanced, non-deterministic multistep issues, turning insights into actions that improve productiveness.
What if we utilized agentic AI to the day-to-day operational challenges confronted by cloud native engineers? That’s the place kagent is available in.
Kagent is an open supply programming framework designed for DevOps and platform engineers to run AI brokers in Kubernetes. It helps engineers construct highly effective inner platforms by tackling cloud native duties similar to configuration, troubleshooting, advanced deployment eventualities, observability pipelines and dashboards, and safely enabling community safety (like mTLS, authentication/authorization modifications, and many others). Recognizing that totally different customers have totally different challenges, we designed kagent to be extensible—permitting DevOps engineers, platform groups, and gear builders to create and share their very own AI brokers.
Constructed on Microsoft’s AutoGen open supply framework, kagent gives a strong basis for AI-driven options in cloud native environments.
Structure of kagent
Kagent is constructed on three key layers: Instruments, Brokers, and the Framework.
- Instruments: Pre-defined capabilities that AI brokers can use, similar to sending emails, looking out a database, displaying pod logs, or calling exterior APIs. Customers can combine Mannequin Context Protocol (MCP) servers as instruments or construct customized instruments.
- Brokers: Autonomous methods able to planning, executing duties, analyzing outcomes, and constantly enhancing outcomes. Brokers make the most of a number of instruments to perform aims.
- Framework: A easy interface to run brokers by way of UI or declaratively. The framework is absolutely extensible, permitting customers to develop new instruments or brokers, or prolong the framework itself.
Primarily based on our expertise, we’ve constructed instruments to work together with Kubernetes, Prometheus, Istio, and Argo, together with AI brokers to autonomously clear up a few of the commonest cloud native issues.
Be part of the kagent Neighborhood
When you’re a platform or DevOps engineer, and leveraging AI brokers to unravel cloud native operation challenges excites you, or for those who’re a developer or CNCF challenge maintainer taken with constructing AI brokers to counterpoint our ecosystem, we’d like to collaborate!
Take a look at our web site and GitHub repository. Observe our getting began information to experiment with AI brokers in your Kubernetes cluster, and contribute your personal brokers and instruments to increase the cloud native AI ecosystem. Be part of the dialogue within the #kagent channel on CNCF Slack—we’d love to listen to from you!
We may also be at KubeCon + CloudNativeCon Europe in London from April 1-4 on the Solo.io sales space if you’re taken with discussing the challenge additional.
Let’s construct the way forward for AI brokers to unravel cloud native operation challenges, collectively.