This thesis presents the design, implementation, and evaluation of a multi-agent conversational assistant for incident management and system monitoring. The system integrates reasoning agents with enterprise services through the Model Context Protocol (MCP). The proposed architecture aims to automate monitoring, documentation, and operational workflows in complex cloud environments while maintaining security, observability, and scalability. The work explores agent orchestration, LLM interfaces, modular backend design, and evaluates system performance in real-world consulting scenarios. The thesis outlines the design principles, architecture, and decision-making logic of the agent, and presents the implementation process. It concludes with a performance evaluation based on predefined KPIs, measuring the solution’s impact on operational efficiency, resolution time, and response quality.