Post

I've been using AI coding agents daily for months. They're incredible at generating code. They're NOT that incredible at distributed systems. Same mistakes every time: no resilience, no observability, hardcoded everything. Services that compile and fall apart in production.

maven.comGenerate, Break, Fix: Distributed Systems in the AI Era by Marcin Grzejszczak on MavenDiagnose, fix, and prevent the distributed systems failures that every AI coding agent creates - in any language.

16/2/26, 4:35 pm0 reposts1 like

Marcin Grzejszczak

@toomuchcoding.com

16 Feb

I spent ~10 years building Spring Cloud Sleuth (now Micrometer Tracing) and Spring Cloud Contract. The tools AI agents keep getting wrong. I've seen these failure patterns from both sides.

Marcin Grzejszczak

@toomuchcoding.com

16 Feb

So I'm building a workshop: "Generate, Break, Fix: Distributed Systems in the AI Era" → You bring your laptop, your AI tool, Docker → I bring broken infrastructure - injecting latency, flaky service discovery etc. → You generate with AI, I break it, you fix it Any language. No slides.

Marcin Grzejszczak

@toomuchcoding.com

16 Feb

First cohort coming soon. Join the waitlist: maven.com/toomuchcoding/generate-break-fix-distributed-systems-in-the-ai-era What's the worst thing AI has generated for you in a backend system?