Imagine a world where diagnosing complex data issues is as simple as pushing a button. That's the promise of ARCAS, Microsoft's automated root cause analysis system. In the sprawling landscape of cloud-based data systems, where countless interconnected components can cause havoc, pinpointing the source of a problem can be like finding a needle in a haystack. Traditional methods rely on skilled engineers painstakingly combing through logs and metrics, a process that's both time-consuming and prone to human error. ARCAS changes the game by automating this intricate detective work. ARCAS uses a unique approach: it leverages a custom-built Domain Specific Language (DSL) that allows experts to encode their knowledge into automated troubleshooting guides (Auto-TSGs). These guides act like digital detectives, tirelessly investigating potential problems using product telemetry. The beauty of the DSL is its simplicity; even engineers unfamiliar with the system can quickly learn to create and modify these Auto-TSGs. Each guide is self-contained and can operate independently, yet they can also collaborate and share context, creating a network of intelligent investigators working in concert. But what happens when multiple Auto-TSGs identify potential problems? This is where ARCAS's secret weapon comes into play: a Large Language Model (LLM). This LLM acts as a judge, prioritizing the findings based on their relevance to the overall problem and the product's architecture. It even generates concise summaries, helping engineers quickly grasp the situation and take appropriate action. Unlike traditional monitoring platforms like Datadog and New Relic, which primarily focus on alerting and require manual intervention, ARCAS takes the next step by automating mitigation. It can trigger actions like rebooting processes, applying feature flags, or even canceling problematic operations, all without human intervention. Already deployed within Azure Synapse Analytics and Microsoft Fabric Synapse Data Warehouse, ARCAS has proven its worth, significantly reducing the time to resolve complex issues and freeing up valuable engineering cycles. While challenges remain in streamlining onboarding new products and refining the LLM's judgment capabilities, ARCAS represents a significant leap forward in automating the often-chaotic world of data systems troubleshooting. It's a testament to the power of automation and AI, offering a glimpse into a future where diagnosing complex technical issues becomes increasingly effortless.