Blog
Updates, guides and deep dives on running LLM agents in Kubernetes.
- Deep Dive
Pipelines are straight lines. Your agents need to think in trees.
kubeswarm's search mode lets agent teams explore, score, prune, and converge - turning multi-hypothesis problems into structured tree searches on Kubernetes.
Your agent just called the same tool 20 times
kubeswarm's tool result cache cuts 80% of redundant MCP calls in reasoning loops - zero config changes to your tools, works with any provider.
- Tutorial
Automated incident response with AI agents on Kubernetes
Three AI agents investigate a PagerDuty alert, diagnose the root cause from Grafana logs, and post findings to Slack - all on a local cluster.
- Tutorial
Run AI agents on Kubernetes: a support triage pipeline with budget guardrails
Build a multi-agent support ticket triage pipeline on Kubernetes using kubeswarm - with policy enforcement, token budgets, and model restrictions. Runs locally with any LLM.