What it does
By using advanced semantic search algorithms, the system retrieves and analyzes data from diverse sources, including player stats, match histories, and esports-related content.
The assistant processes input prompts and performs semantic searches across a custom-built repository of VALORANT esports data stored in a vector database. This ensures that the system retrieves the most relevant and accurate information to answer specific questions about VALORANT players, their performances, teams, and tournaments.
The platform is designed to handle multiple formats of data, such as images, YouTube videos, and PDF documents. For videos, the system integrates with the YouTube API to fetch relevant metadata, which is then summarized by the LLM to provide concise insights into esports events or player highlights. The system also handles image-based content, offering the ability to extract and summarize data from esports-related infographics or images.
Additionally, the platform supports image generation through the Stable Diffusion algorithm, enabling users to request custom images such as VALORANT player representations or team visuals based on textual prompts. All generated content is stored securely in Pinata (IPFS) for easy access and reference.
The platform maintains a seamless chat history, allowing users to build on previous conversations and enabling smooth, ongoing interactions with the digital assistant. This continuity ensures that users can ask follow-up questions or request updates on ongoing tournaments without needing to repeat context, enhancing the overall user experience.
How we built it
NextJS Langchain LangGraph Huggingface API Pinecone Pinata Firebase
Built With
- langchain
- llm
- nextjs
- pinata
- pinecone
Log in or sign up for Devpost to join the conversation.