Summary
- Observed potential performance patterns in the ad service and wanted to understand what had been causing issues over recent weeks
- Traditional investigation would require manually analyzing weeks of logs, correlating code changes with performance metrics, and synthesizing scattered observations into actionable insights
- Resolve AI performed comprehensive system analysis, reconstructing the complete story of recent performance issues and their root causes
- Generated detailed recommendations spanning code review processes, testing protocols, monitoring improvements, and developer education to prevent future issues
“I had a hunch that our ad service might have performance issues based on some patterns I'd been noticing. Instead of digging through weeks of logs and code changes myself, I asked Resolve AI to do a comprehensive analysis. It didn't just confirm my suspicions \- it gave me a complete post-mortem-style analysis of what had been happening, specific root causes, and detailed recommendations across our entire engineering process. What would have been weeks of investigation became a systematic improvement plan in minutes.”
What was the investigation?
Spiros had been noticing concerning patterns in the ad service performance and wanted to understand what had been causing these issues over recent weeks. Rather than having a crisis to respond to, he wanted to proactively investigate and improve the system's reliability.
This type of proactive investigation typically requires extensive manual work: analyzing weeks of performance data, correlating deployment timelines with system behavior, reviewing recent code changes for potential issues, and synthesizing scattered observations into coherent insights. For a busy engineering team, this comprehensive analysis could easily consume weeks of effort across multiple engineers.
How did Resolve AI provide systematic analysis?
*Resolve AI performed comprehensive system analysis, examining weeks of performance data and code changes to understand systemic patterns.*
Resolve AI approached this as a comprehensive system health investigation. Rather than waiting for specific alerts or incidents, it systematically analyzed recent system behavior, examining deployment histories, code changes, and performance patterns to understand what had been affecting the ad service.
The analysis revealed that a code change from June 17th had introduced a blocking SQL query (select count(1) from product;) that executed on every ad service request. This seemingly innocent addition was opening new JDBC database connections each time, creating performance degradation that had been gradually affecting system reliability.
*Root cause analysis connected code changes with performance impact across multiple weeks of system behavior.*
Resolve AI reconstructed the complete story of how this code change had been affecting the system. The blocking database query wasn't causing dramatic failures, but it was creating consistent performance degradation that manifested as elevated latency, increased error rates in frontend operations like /api/checkout, and general system stress under load.
The analysis provided the kind of comprehensive understanding that typically emerges only during major incident post-mortems - but applied proactively to understand and improve ongoing system behavior.
*Systematic recommendations across multiple engineering domains to improve overall system reliability.*
Beyond identifying the specific technical issue, Resolve AI generated comprehensive recommendations for systemic improvement. Rather than just fixing the immediate problem, it provided guidance spanning multiple organizational levels: code review processes requiring database impact assessment, mandatory performance testing for query-heavy changes, enhanced monitoring for service-level latency patterns, and developer education on proper database connection management.
The recommendations addressed the complete engineering lifecycle: preventing similar issues during development (PR templates, code review guidelines), detecting them earlier (monitoring improvements), and building organizational knowledge (developer training on N+1 queries and connection pooling).
What was the impact?
- Transformed observation into systematic understanding by converting vague performance concerns into precise root cause analysis and comprehensive system health assessment
- Accelerated proactive investigation from weeks of manual analysis across logs, code, and metrics to minutes of comprehensive synthesis
- Generated organizational improvement plan with specific recommendations spanning development processes, testing protocols, and monitoring capabilities
- Enabled immediate technical remediation through integration with Cursor while simultaneously building process improvements to prevent future issues
- Built proactive engineering culture by demonstrating how systematic analysis can transform operational hunches into concrete system reliability improvements
