How AI Speeds Up Trace Analysis in Observe

OpenTelemetry trace data, often used by engineers for debugging complex applications, can be challenging to interpret due to the sheer volume of spans involved in complex operations. To address this, Observe developed the o11y AI trace analysis tool, a feature integrated into the Trace Explorer, which leverages large language models to streamline the debugging process. This tool, born from an internal hackathon, offers insights such as critical path analysis and service relations by analyzing trace data and identifying bottlenecks and common patterns, such as N+1 query issues. The "Ask o11y" button allows users to easily access AI-driven insights and jump to specific spans for detailed review. Feedback from customers has been positive, as the tool enhances existing workflows by reducing the time engineers spend scanning spans and enabling quicker identification of issues, thereby decreasing Mean Time to Recovery (MTTR). The successful implementation and positive reception of this tool underscore its utility in managing large, complex traces more efficiently during incidents.