From Code to Agents: Building Production MCP Servers on Azure Functions with .NET

Speakers

• Jonah Andersson

About this session

Every enterprise team building with AI agents hits the same wall: your agents are smart, but they’re isolated. They can’t access your internal tools, your DevOps pipelines, your ticketing systems, or your customer data — not without fragile, custom integrations that break every time something changes. The Model Context Protocol (MCP) solves this by giving agents a universal way to discover and call tools. But the question developers keep asking is: how do I take MCP from a local prototype to a production-grade, scalable service my entire organization can rely on? This question hits differently depending on who you are. Enterprise developers are blocked by security and compliance requirements and need to prove ROI to leadership. AI developers are drowning in agent orchestration complexity while trying to balance cost, latency, and quality. Elite developers need scale and reliability at the frontier and can’t afford slow innovation cycles — time is their scarcest resource. This session answers that question with a live, end-to-end build. We’ll construct production MCP Servers using Azure Functions and .NET — turning serverless functions into discoverable, secure tool endpoints that any AI agent can call. Attendees will see two real-world scenarios come to life: • DevOps Automation Agents: A multi-agent system that monitors deployments, triages incidents in Azure DevOps, runs diagnostics via MCP tool calls, and auto-creates remediation PRs — all orchestrated by agents that discover capabilities at runtime. • Multi-Agent Customer Service: A customer-facing system where specialized agents each expose their domain tools as MCP servers, and a routing agent dynamically selects the right specialist based on context.