> Markdown version of [/events/world-congress-2026-europe/sessions/1364-when-agents-meet](https://www.wearedevelopers.com/events/world-congress-2026-europe/sessions/1364-when-agents-meet). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # When Agents Meet Legacy: Never Change a Running System - **Date:** Friday, Jul 10, 2026 - **Time:** 15:00–15:30 (30 min) - **Room:** Stage 7 - **Event:** World Congress 2026 Europe ## Recording [Watch recording](https://www.wearedevelopers.com/videos/100320-when-agents-meet-legacy-never-change-a-running-system) ## Description I've spent a decade debugging undocumented C code, hand-written C++ Makefiles, and grown Java monoliths, knowing exactly which files will break production. Left "allowed to fail" CI/CD pipelines because fixing root causes is harder than ignoring them. I was curious: Can AI Agents help with modernization, and where would they still fail? We'll live-code: untangling complex architectures in C and C++, and if Rust is the answer to everything, breaking Java monoliths to Kubernetes, where agents propose but humans validate, and fixing CI/CD burning budgets when culture resists tooling. The Tanuki IoT Platform environment presents more challenges when agents meet COBOL. What I've learned: agents can't replace judgment on "should we change this?" You'll leave knowing where they help triage, where they fail on business context and experience, and which human strategies, like debugging, you still need. Agents are powerful guides, but the messy decisions are still yours. ## Speaker ### [Michael Friedrich](https://www.wearedevelopers.com/@michael-friedrich) Principal Developer Advocate at GitLab ## Related talks at this congress - [Build Agents Once, Run Them Anywhere: AI Systems That Survive Change](https://www.wearedevelopers.com/events/world-congress-2026-europe/sessions/1471-build-agents-once) — Isha Salania - [What Production Knows: Closing the Loop Between AI Agents and the Systems They Build](https://www.wearedevelopers.com/events/world-congress-2026-europe/sessions/1470-what-production) — May Walter - [Teaching GitHub Copilot COBOL: A Practical Guide to Agentic AI Legacy Modernization](https://www.wearedevelopers.com/events/world-congress-2026-europe/sessions/1242-teaching-github) — Julia Kordick, Laetitia Maar - [Building a Multi-Agent Orchestration Engine That Actually Follows the Rules](https://www.wearedevelopers.com/events/world-congress-2026-europe/sessions/1174-building-a-multi) — Hussein Jundi, Torsten Stiller