Based on presentations from customers at Perform 2026, here are five real-world lessons for operationalizing agentic AI in dev workflows.

Dynatrace's platform is a central hub for IT professionals and DevOps teams, offering insights into application performance monitoring, cloud observability, and digital experience management. Through articles, case studies, and industry reports, Dynatrace offers insights into monitoring distributed systems, detecting performance bottlenecks, and optimizing application performance. Readers can learn about end-to-end visibility, AI-driven analytics, and automation solutions to deliver superior digital experiences for users.

Dynatrace

Real-world lessons from enterprise teams (TELUS, Lockheed Martin, Autodesk, Macquarie Group, United Wholesale Mortgage) on integrating AI agents into developer and operations workflows. Key patterns include: giving agents real-time telemetry access via MCP, embedding AI into existing developer tools like VS Code and Slack, standardizing observability with OpenTelemetry and enriching it with domain context, treating the AI stack itself as a production system requiring governance and monitoring, and closing the loop from insight to automated action. Companies that contextualized their MCP configurations saw success rates jump from 60% to 95%.

Five real-world lessons for building developer workflows in the agentic era

1. Give agents real-time access to telemetry data

3. Standardize telemetry and enrich it with domain context

4. Observe and govern the AI stack itself