The VisionQA Application leverages LLM models to automate BDD test case generation and execution. It also provides project summaries and context-aware chat support to simplify the process of writing and executing BDD tests.
📹 Video Demo: https://drive.google.com/file/d/1aYgrzA8HfzB8GE7Ex6YE-cmqOHhkR4eL/view?usp=sharing
AI Agent Demo: https://drive.google.com/file/d/1w-KR0klxA5wT5HJPcOfuat1NQxZbnW1Q/view?usp=sharing
Our AI agent combines techniques from Adaptive RAG, Corrective RAG, and Self-RAG to enhance retrieval accuracy and reliability. It dynamically routes queries, falls back to web search when needed, and self-corrects responses to reduce hallucinations and improve answer quality.The agent also stores history in state using checkpointer memory for context retention.
- User Onboarding: Upload Confluence and JIRA URLs to extract data.
- Project Summary: Generates a comprehensive summary from contextual data.
- Context-Aware Chat: Provides project-related answers and insights.
- BDD Test Case Generation: Creates test cases based on contextual data.
- Syntax Validation: Validates BDD syntax with advanced LLM models.
- Step Definition Generation: Generates precise and executable step definitions.
- Test Execution: Runs tests and generates reports using Allure.
- Reporting: Provides a shareable URL for the generated reports.
cd code/src/backend
pip install -r requirements.txt
uvicorn main:app --reloadcd code/src/frontend/visionqa
npm install
npm startcd code/src/model/
pip install -r requirements.txt
cd agent/
langgraph dev- Frontend: React, Tailwind CSS
- Backend: FastAPI
- Database: Pinecone Vector DB
- LLM: Langchain, LLaMA, deepseek-r1-distill-llama-70b, deepseek-r1-distill-qwen-32b, Chat Groq API
- Sentence Transformer: sentence-transformers/all-mpnet-base-v2
- Image Processing: Salesforce BLIP, OCR
- Testing and Reporting: Behave, Allure