Building the Brain: Designing the Senior AI Persona (Part 2)

Part 2 of a series on building an AI-powered performance analysis system. The focus is on designing a 'Senior AI Persona' through structured prompt engineering to get expert-level root cause analysis from GPT-4o. A Python script loads a pre-processed JSON summary of 14,000 stress test samples and sends a detailed prompt that assigns the AI a 20-year performance engineering background with specific analytical tasks. The resulting output identifies a cascading failure pattern, links a 34.18% error rate to an 89.9-second latency spike, and suggests connection pool exhaustion as the likely root cause along with a circuit breaker as an immediate fix.