uChat

Overview Welcome to uChat, an innovative platform developed for the FetchAI Hackathon. uChat allows users to create custom chatbots tailored to their specific needs. Leveraging advanced language learning models and Retrieval-Augmented Generation (RAG), uChat bots can be configured with a variety of traits and knowledge derived from uploaded documents.

The platform is built using Next.js for the frontend, PostgreSQL for database management, and hosted via Vercel. Our backend is powered by Python and Flask, running on Render and designed to integrate with Google Cloud for file storage and ChromaDB for context management.

Key Features • User Authentication: Users can sign up and log in to manage their chatbots. • Custom Chatbots: Create and configure chatbots with: • Name: Give your chatbot a unique identity. • Description: Define the purpose and goals of the chatbot. • Role: Assign a specific role to the chatbot, such as roleplaying as a particular character or entity. • PDF Knowledge Upload: Users can upload PDFs to equip their chatbot with specific knowledge from the document. The bot can then respond based on the content of the uploaded PDF. • Context Extraction: Using Retrieval-Augmented Generation (RAG) with ChromaDB, chatbots can extract and utilize relevant context from the uploaded PDFs during conversations. • Dynamic Prompt Generation: The platform uses a uAgent called uPrompt to dynamically generate prompts by combining user-provided parameters like context extraction, role, and chat history.

Technical Details Frontend • Framework: Next.js • Hosting: Vercel • Database: PostgreSQL

Backend • Framework: Python with Flask, uAgents • Containerization: Docker • Hosting: Render (Free Plan) • Storage: Google Cloud Storage • Context Management: ChromaDB

Built With

• chromadb
• docker
• flask
• google-cloud
• llm
• nextjs
• postgresql
• python
• rag
• render
• typescript
• uagents
• vercel