The challenge
This global media and entertainment leader runs a multi-cloud Kubernetes estate spanning EKS, AKS, and GKE. Workloads are spun up at a pace and scale that outran the Platform Engineering team’s ability to see what was happening, let alone whether it was sized correctly. Clusters were standing up without clear visibility. Containers were running without anyone able to confirm whether they matched the workloads on them. Workloads that had completed their job sometimes stayed on long after they were needed.
The team needed a way to put optimization intelligence directly in the hands of the engineers running the workloads, in the tools and workflows those engineers already used. Pulling answers from dashboards was not going to scale.
The solution
The customer built a custom AI agent interface and connected it to Kubex MCP. The same agent talks to ServiceNow, Jira, and Amazon Cost Explorer through their MCP servers, letting engineers ask one question and pull context from across the estate.
With the agent in place, the team’s interaction with Kubex shifted from dashboards to conversation. Engineers ask questions in natural language and get back tabular answers, recommended remediation paths, and the launch-template steps to execute them. The agent flags clusters that are already near-optimal alongside the ones with the largest savings opportunity, so teams focus their work where the impact is greatest.
The results
- Conversational, natural language access to Kubernetes optimization data across a multi-cloud estate
- Top-10 ASGs by cost avoidance pulled in roughly ten seconds during a working demo
- Actionable, step-by-step remediation, including instance modernization paths and the launch-template steps to apply them
- Cross-source enrichment built on Kubex MCP plus ServiceNow, Jira, and Amazon Cost Explorer MCPs
- Manual lookups, URL-chasing, and cross-system data pulls eliminated for engineering teams
- Built on standard MCP connections, no integration heroics required
In their words
“It’s nuts. It’ll give me tables for all of them. It’s fantastic.” Infrastructure Platform Engineering, global media and entertainment leader
