This project simulates a real-world IT development team using a multi-agent AI system. It automates key stages of the software development lifecycle — from requirement analysis to code generation and testing.
The system is built using agent orchestration frameworks like CrewAI and LangChain, with local LLM integration via Ollama.
- 📄 Business Analyst Agent → Generates user stories from high-level requirements
- 🧩 Design Agent → Produces system design based on user stories
- 💻 Developer Agent → Generates code from design and requirements
- 🧪 Testing Agent → Creates and executes test cases
- 🔄 Sequential Agent Pipeline (BA → Design → Dev → Test)
- 🖥️ Basic UI built with Streamlit
User Input
↓
Business Analyst Agent
↓
Design Agent
↓
Developer Agent
↓
Testing Agent
↓
Final Output
👉 Note: Currently, agents execute sequentially, not in parallel.
- Python
- Streamlit
- CrewAI
- LangChain
- Ollama (Local LLM)
- ChromaDB / Pinecone (for vector storage)
git clone https://github.com/kb-111/College-project
cd College-project
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
streamlit run app.py- High latency due to local LLM (Ollama)
- Difficulty in integrating real-time chatbot functionality
- Managing coordination between multiple agents
- Limited access to paid APIs (e.g., OpenAI)
- Chatbot interface is not fully implemented
- Agents do not run in parallel (sequential execution only)
- Performance is slower due to local model usage
- Add fully functional chatbot interface
- Enable parallel execution of agents
- Integrate faster cloud-based LLM APIs
- Improve UI/UX for better interaction
- Designed and implemented the core multi-agent workflow
- Integrated CrewAI and LangChain for agent orchestration
- Built Streamlit interface for interaction
- Handled LLM integration using Ollama
- Team of 4 (6th Semester Project)
This project was developed as a learning-focused implementation of multi-agent AI systems and is not production-ready.