ai-quality-sentinel

Roadmap: Building the POC (Phase 1)

Setup & Design (Day 1-2)

• Persona Definition: Create the profile of the “Detailed Reviewer” (the AI voice).
• Test Dataset: Select 5 anonymized real examples of good/bad tickets and matching code.
• Model Choice: Configure access to the model via AI/works™ (or GPT-4/Claude for initial testing).

Development (Day 3-5)

• Extraction Module: Script to read ticket text (Markdown/JSON).
• Code Analysis Module: Script to process .diff or .py/.js files.
• Prompt Engineering: Create the system prompt that instructs the AI to be a rigorous quality auditor.
• Output Validation: Create a response template (e.g. Quality Traffic Light - Green/Yellow/Red).

Demonstration (Day 6-7)

• Video Recording: Demo comparing a quick human review with the AI’s deeper analysis.
• Results Documentation: Table of “Before vs After” using the tool.