👩🏻‍🍳 A collection of example notebooks using Haystack

GPT chat with your docs!

Fully working applications that demonstrate how to use Haystack to implement various use cases

AI agent grounded with search and extract tools to reduce hallucination.

An example of a RAG backend plus UI

YouTube Video Summarization App built using open source LLM and Framework like Llama 2, Haystack, Whisper, and Streamlit. This app smoothly runs on CPU as Llama 2 model is in GGUF format loaded through Llama.cpp.

Simple Haystack in-memory document store alternative that performs incremental indexing and supports SentencePiece tokenizer.

GitHub Issue Resolver Agent with Anthropic Claude 3.7 Sonnet and Haystack

Different retrieval techniques implemented in Haystack - presented at PyCon Lithuania 2025 🇱🇹

👾 A Template for Haystack Apps with Streamlit

The official integration for Apify and Haystack 2.0

Used cognitivecomputations/dolphin-2.9-llama3-8b Open source model deployed on Serverless GPUs using Runpod

WIP - An intelligent assistant that automatically routes and processes user queries across a collection of BBC News articles, with access to different tools and capabilities.

A Chatbot used to fetch articles from PubMed based on user queries and displays the queries and responses in a user friendly interface built using Streamlit. It fetches the Title, Abstract and Keywords based on user's input and then stores the Question Answer pair in Firebase.

e2e Retrieval Augment Generation (RAG) social app, that lets users collaborate(query and index docs, reddit searches/comments and wiki articles) on public workspaces or private ones. Has multiple connectors for reddit, wikipedia, pdfs etc.. Deployed on AWS

Playing around with Retrieval Augmented Generation.

Generative QA pipeline with RAG and Gemini-Pro model

RAGlery is a project dedicated to testing, comparing, and documenting various Retrieval-Augmented Generation (RAG) frameworks. The primary goal is to provide insights into the implementation, performance, and use cases of different RAG systems.

This project automates the extraction and cleaning of stock concall transcripts from PDFs, preparing them for analysis. It leverages a RAG pipeline with Haystack and GPT models to summarize key points, compare concalls, and uncover trading opportunities.