AGENTIC AI

This is an Agentic AI application built on the Spring Boot framework that automates animal adoption and care processes. This prototype validates the core architecture for a conversational assistant that can understand user intent, retrieve contextual information, and perform real-world actions

🌟 Features

• Intelligent Routing: A router-agent architecture that dynamically delegates user queries to specialized AI agents based on intent.
• Retrieval Augmented Generation (RAG): Integrates with a PostgreSQL Vector database to provide accurate, data-driven responses using contextual information.
• Automated Tool Calling: Enables the AI to perform business logic (e.g., booking appointments) by calling a custom tool.
• Conversational Memory: Maintains per-user chat history to support coherent, multi-turn interactions.
• Modular & Scalable: Designed with a clean, loosely coupled architecture to easily integrate new features and tools.

🚀 Technologies Used

• Backend Framework: Spring Boot 3.x
• AI Framework: Spring AI
• Database: PostgreSQL with the PGVector extension
• LLM Provider: OpenAI
• Build Tool: Maven

Getting Started

This guide will help you get a local copy of the project up and running for development and testing.

1. Prerequisites

• Java Development Kit (JDK) 17 or higher
• Maven 3.6.3 or higher
• Docker (for running the PostgreSQL database with PGVector)

2. Setup

• Clone the repo

  git clone https://github.com/wizzy-abhishek/MCP

• Configure Environment Variables:
  Create an application.properties file in src/main/resources and add your configurations.

  # Your Database connection
  spring.datasource.url=jdbc:postgresql://localhost:5432/your_db_name
  spring.datasource.username=your_username
  spring.datasource.password=your_password
  
  # Your LLM API key
  spring.ai.openai.api-key=your_openai_api_key
  
  # RAG Vector Store configuration
  spring.ai.vectorstore.pgvector.embedding-dimension=1536
  

3. Run the Database:

• You can use Docker to run a PostgreSQL instance with the PGVector extension.

  docker run --name pgvector -e POSTGRES_PASSWORD=your_password -p 5432:5432 -d ankane/pgvector

  Then, create the your_db_name database.

4. Running the Application

• You can run the application directly from your IDE or using Maven:

  ./mvnw spring-boot:run

  The application will start on port 8080.

💡 Usage

Once the application is running, you can test the AI assistant by making a GET request to the following endpoint:

    GET http://localhost:8080/inquire?user=john_doe&question=Do you have a golden retriever?

Example Queries:

    GET http://localhost:8080/inquire?user=john_doe&question=Tell me about a dog that is good with kids.
    GET http://localhost:8080/inquire?user=john_doe&question=Book an appointment for my dog Max on August 20, 2025.