AI Assistant Added to Deckhouse Products: Incident Analysis and Cluster Configuration in Dialogue Mode

Deckhouse has introduced an AI assistant in two products: Deckhouse Kubernetes Platform and Deckhouse Observability Platform. The tool is embedded in the graphical interface and lets engineers interact with infrastructure through a chat, retrieving system state data, analyzing incidents, and performing typical operations with user confirmation. In case of failures, the assistant helps identify the probable root cause faster and suggests fix options. The assistant lowers the entry threshold for working with Deckhouse platforms and reduces routine operations, enabling faster digital product launch and better resource utilization. In the Kubernetes Platform, the assistant manages containers, assesses cluster state, and can execute tasks described in plain language after user approval. It also supports virtualization scenarios by preparing VM configurations and deploying or migrating virtual machines via Model Context Protocol (MCP). For monitoring and observability, the assistant provides summaries and links service behavior to infrastructure state, reducing troubleshooting time. The assistant acts within the existing security model, performing actions under user rights (RBAC) and requiring explicit human confirmation for changes. It operates on OpenAI-compatible models, including those in isolated environments, suitable for security-sensitive sectors. Alexey Tishonko, director of the product line at Flant, stated the assistant is an experimental feature and will later evolve to diagnose incidents, include a built-in LLM for offline use, and become a full mentor for users.