Tikeetron AI (Python FastAPI)

This is the AI backend for Tikeetron, an AI-first event marketplace app that uses FastAPI for managing AI functionalities such as event recommendations, ticketing assistance, and real-time user interaction. The AI system is powered by Llama3, utilizing Retrieval-Augmented Generation (RAG) with MongoDB vectors for data retrieval. The backend also interacts with the blockchain and serves as the core engine for the app's AI assistant.

Features

• AI-Powered Event Recommendations: Provide personalized event recommendations to users based on historical data and preferences.
• Real-Time Ticketing Assistance: Users can ask the AI about their tickets, event details, and manage ticket transfers.
• Integration with MongoDB: Store event and ticket information as vectors for efficient retrieval and ranking of results.
• FastAPI Framework: Lightweight and fast API framework to handle AI interactions.

Core Dependencies

The Tikeetron AI backend relies on the following core dependencies for efficient functionality:

1. LangChain

LangChain is used to build and manage complex language model pipelines. It helps us handle natural language understanding (NLU), event recommendations, and interactions with the AI assistant.

• Use Case: Managing the conversation flow and decision-making for AI-powered features in the app.

2. LangChain-GROQ

This extends LangChain's capabilities by adding integration with GROQ, a high-performance hardware and software stack for accelerated AI workloads. It enhances the speed and efficiency of running language models like Llama3.

• Use Case: Enhancing the inference speed and performance of our AI models when processing complex queries from users.

3. PyMongo

PyMongo allows us to interact with MongoDB, which stores all the event and ticket metadata as vectorized information for retrieval-augmented generation (RAG).

• Use Case: Storing and retrieving event data efficiently, enabling fast AI query results based on user input.

4. LangGraph

LangGraph is used for building conversational agents and knowledge graphs. This allows us to structure and link event data, ticket information, and user interaction histories for more context-aware AI responses.

• Use Case: Building a knowledge graph of event data and user preferences for improved AI-powered event recommendations.

Installation

To set up the Tikeetron AI backend, follow these steps:

1. Clone the Repository

   git clone https://github.com/TikeeTron/tikeetron_ai.git
   cd tikeetron-ai-backend

2. Install Dependencies
   Install the required Python packages listed in requirements.txt:

3. Set Up MongoDB

   • Ensure that MongoDB is up and running. The AI system interacts with MongoDB to retrieve and store vectorized data.
   • Create a MongoDB database for event and ticket storage.

4. Run the FastAPI Server
   Start the FastAPI server to handle AI interactions:

   uvicorn main:app --reload

5. Environment Variables
   Make sure to configure the example environment variables

AI Architecture

The Tikeetron AI backend uses a combination of:

• Llama3 for language understanding.
• Retrieval-Augmented Generation (RAG) for retrieving context-relevant information from MongoDB vector data.
• FastAPI to serve AI interactions at scale.
• LangChain and LangGraph for building conversation flows and linking knowledge across event datasets.

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

For more information about the AI assistant and its integration with the Tikeetron mobile app, refer to our White Paper.