Improve one existing AI prompt or structured-output flow that is producing inconsistent, incomplete, or hard-to-parse results.

I will review the current prompt, expected inputs and outputs, failure examples, and the surrounding validation logic. The work may include clearer instructions, output-schema improvements, representative test cases, and lightweight parsing or retry guidance.

This service is for one narrowly scoped flow. It does not include building a complete chatbot, RAG system, agent platform, model training, production deployment, or guarantees about model behavior.