This project is a lightweight AI-powered system designed to transform raw meeting notes into high-quality, structured requirements.
Unlike basic AI generators, this system focuses on:
- Requirement clarity
- Validation
- Risk identification
- Test readiness
In real-world projects, requirement quality is a major bottleneck:
- Ambiguous user stories
- Missing edge cases
- Undetected dependencies
- Rework during development
This leads to delays, misalignment, and inefficiencies.
This project introduces a Requirement Intelligence Layer that:
✔ Converts meeting notes → structured user stories
✔ Identifies gaps and ambiguities
✔ Detects risks and dependencies
✔ Generates test cases
✔ Provides estimation with reasoning
- 🧾 User Story Generation
- ✅ Validation Engine (ambiguity & completeness check)
⚠️ Risk & Dependency Detection- 🧪 Test Case Generation
- 📊 Estimation with reasoning
Meeting Notes ↓ LLM (Local - Ollama / Phi3) ↓ Structured Output ↓ (User Story + Validation + Risks + Tests + Estimation)
- Python
- Ollama (Local LLM)
- Phi3 Model (lightweight, runs on 8GB RAM)
ollama run phi3
pip install ollama python-dotenv
python ai_ba_copilot.py
User should login using email and OTP. OTP expires in 5 minutes. System should handle invalid OTP. Integration with SMS service required.
- User Story
- Validation Issues
- Risks & Dependencies
- Test Cases
- Estimation
This is not just a generator.
It acts as a thinking layer between:
Raw Input → Execution (Devs / AI Agents like Devin)
It ensures: 👉 Better requirement quality before execution
- Streamlit UI (web interface)
- RAG (context-aware generation using BRDs / past stories)
- Jira / Confluence integration
- Multi-agent workflow
Aditya
AI + Product Enthusiast | Building AI-powered BA tools
Improving requirement quality = improving delivery outcomes
This project aims to bridge the gap between:
- Business understanding
- Technical execution