A Fork of Rikkahub with an overhauled UI and feature additions

A Streamlit app for chatting with the content of uploaded files (PDF, DOCX, images) using AI, featuring conversational memory, document summarization, retrieval-augmented generation, and powered by GPT-4o-mini, LangChain, and LangGraph.

LLM-Powered AI Mentor System delivers focused technical mentorship across Python, SQL, ML, DL, GenAI domains using Streamlit + LangChain. Features strict module restriction, conversation memory, and TXT chat export. Production-ready with virtual environment setup, professional UI, and scalable architecture for structured learning.

Stateful Telegram AI Assistant powered by Gemini 2.5 Flash & LangChain. Securely manages Google Calendar (CRUD) via custom tools. Features SQL-backed conversational memory for dynamic chats, zero-downtime error handling, strict access control, and cloud-ready architecture.

Self-organizing agent memory. No embeddings, no vector DB.

The repository accompanies the SSPM research preprint and includes a Google Colab–ready notebook for experimental validation and visualization.

A context-aware Retrieval-Augmented Generation (RAG) chatbot built with LangChain v1.0+, OpenAI, FAISS, and Streamlit that delivers accurate, grounded answers with conversational memory.

Advanced AI Assistant with conversational memory, agent capabilities, MCP , knowledge bases, multi-user support, and RAG (knowledge graph + vector search). Includes prompt management and other AI services.

Flowise chatflow implementing a tool-using LLM agent with OpenAI embeddings (3072-dim), Pinecone similarity search, Jina reranking, Tavily API search, document loaders, text splitting, and conversational memory.

ShoppingGPT is a production-ready, agentic AI chatbot for e-commerce. It routes every conversation through a semantic classifier, dispatches to specialized LangChain tools, and streams back intelligent responses — all in under a second.

LLM-Talk enables natural voice conversations with language models using hotword activation.

Persistent per-project memory for Claude Code via MCP server (4 tools) + 4-layer L0/L1/L2/L3 extraction pipeline. English-localized coder-focused fork of Tencent/TencentDB-Agent-Memory with vector recall, Voyage embeddings, and SessionStart auto-injection.