Qdrant

🔍 Did you know you can create Qdrant collections with just sparse vectors? No need to declare dense vectors when you don’t need them! Learn more: https://buff.ly/3Ou1YYk

🤖 Build an Agentic RAG System with Qdrant, CrewAI & Anthropic Claude! Discover how to create an intelligent meeting assistant that combines vector search with CrewAI agents. This step-by-step guide covers: ✅ Vector Search Integration: Store and retrieve meeting data efficiently with Qdrant. ✅ AI Agent Framework: Use CrewAI agents to analyze and summarize transcripts. ✅ Interactive Interface: Build a chat UI with Streamlit for user interaction. By combining Qdrant’s vector search with modular AI agents, you’ll transform meeting data into real value. Check out the tutorial and start building today! Check out the tutorial: https://buff.ly/41cFMJP Github repo: https://buff.ly/3Z5BO2V

⚖️ #VectorWeekly 𝐨𝐧 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐅𝐫𝐚𝐦𝐞𝐰𝐨𝐫𝐤𝐬 Agents might have caught your attention recently. So, we tested and compared four popular agentic frameworks (#LangGraph, #CrewAI, #Autogen, OpenAI #Swarm): Choose your fighter! 👉 Check out the code examples for using these frameworks with Qdrant + theory and practical insights: https://lnkd.in/dyNSfChK P.S. Should you always use agents? Not necessarily. Agents rely heavily on LLMs, making multiple calls to them, which can sometimes be too slow or expensive.

Contrary to popular belief, the latest AI tools are stymieing misinformation rather than facilitating it. The Center for Autonomy at The University of Texas at Austin recently published their research into misinformation detection where they improved the detection of misinformation by 26% over previous state of the art methods. How did they do it? CrediRAG - it leverages AskNews and Qdrant for fast and accurate retrieval of carefully indexed news in combination with a graph attention network. 1. Input a Reddit Post: We take any Reddit post and pass it through AskNews. 2. Retrieve Relevant News: AskNews scours its expansive Qdrant database for enriched news articles that match the post’s content and conform to time/reporting voice filters. 3. Compare and Classify: The retrieved articles are compared against the original post to classify it as factual or misinformative. Doing this type of backtesting and real-time analysis *at scale* across millions of news articles demands high-throughput and precision from Qdrant + AskNews. This would have been unimaginable just 24 months ago, but now it all sits behind a python SDK in the cloud. Links to the original CrediRAG publication and blog in the comments below 👇 https://lnkd.in/gnnNeHuT #fakenews #misinformation #qdrant #asknews #rag #llm #ai4news #transparency First author: Ashwin Ram

🛠️ RAG Systems: Build It, Then Break It (For Science) Building a RAG system is just the beginning. The real challenge? Making sure it performs consistently under pressure. Missed retrievals? Hallucinated answers? Poorly optimized pipelines? These are the bottlenecks that turn promising systems into frustrating ones. In our latest blog, we break down what it takes to go from “it works” to “it works well.” ✅ Spot retrieval blind spots using precision metrics and relevance testing. ✅ Fine-tune embedding strategies to ensure accurate context is passed to your LLM. ✅ Measure and reduce hallucination rates, so your LLM generates grounded, factual responses. We’re talking real frameworks (Ragas, Quotient AI, Arize AI Phoenix), specific fixes for underperforming pipelines, and concrete metrics like NDCG and Recall to measure success. 📖 Learn how to evaluate your RAG system: https://lnkd.in/ez2DEEhj

I received this gorgeous T-shirt from Qdrant ! Qdrant vector database is still to me one of the best options for semantic similarity and other use cases, with countless possibilities (did you know you could make it play a game?) and easy implementation. Looking forward to the upcoming updates! the T-shirt not only has an amazing design, but the quality is also top-notch! Thanks again Qdrant !

This demo is powered by Qdrant for multimodal embedding storage, Sendbird Ai Chatbot No Code Widget and Anthropic Claude 3.5 Sonnet. You can quickly explore and prototype multimodal recommendation use cases. Disclaimer. We don't store image embeddings in our native vector DB yet so I had to improvise with Qdrant Cloud which is free to try out! #anthropic #sendbird #qdrant #vectordatabase #chatbot

🚀 𝐆𝐞𝐧𝐨𝐦𝐢𝐜𝐬 + 𝐕𝐞𝐜𝐭𝐨𝐫 𝐃𝐁𝐬 = 𝐢𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐨𝐧 #Gaia by Tatta Bio is a protein context-aware search platform combining #gLM2 (a genomic language model) and Qdrant. It doesn't just compare sequences or structures — it looks at the "bigger picture" of where proteins are located in the genome and how they interact. ✅ Context-aware searches reveal functionally related genes 𝐦𝐢𝐬𝐬𝐞𝐝 by older protein search methods; ✅ Gaia has already 𝐝𝐢𝐬𝐜𝐨𝐯𝐞𝐫𝐞𝐝 new protein functions; ✅ It can search in real-time across 85M protein clusters in just 0.2 seconds per query, 𝐨𝐫𝐝𝐞𝐫𝐬 𝐨𝐟 𝐦𝐚𝐠𝐧𝐢𝐭𝐮𝐝𝐞 𝐟𝐚𝐬𝐭𝐞𝐫 than existing tools. Truly, innovation happens at the boundary of disciplines! 👉 Check it out https://lnkd.in/gs6T9MJQ

🤔 What is Agentic RAG? Vanilla RAG works well for simple, linear tasks: retrieve, respond, repeat. But when queries are complex or require iterative reasoning, it’s not enough. Agentic RAG integrates agents with multiple knowledge sources. It adapts workflows to real-time needs and these agents can make intelligent decisions about when and how to search your data. They refine, iterate, and self-correct. ❗ Choosing the proper agentic framework is important! Each option we compare - LangGraph, CrewAI, AutoGen, and OpenAI Swarm - has unique strengths for different use cases. Learn how they integrate with Qdrant and pick the best fit for your tech stack and requirements. 🚀 Read the full article by Kacper Łukawski: https://lnkd.in/dyNSfChK

Do you think 𝘝𝘦𝘤𝘵𝘰𝘳 𝘚𝘦𝘢𝘳𝘤𝘩 is only about RAG? Not at all. Meet Gaia, context-aware protein search across genomic datasets powered by Qdrant. Gaia, developed by Tatta Bio, is an advanced protein search platform incorporating genomic context alongside sequence and structure data, enhancing protein function discovery in microbial systems. Powered by the gLM2 genomic language model, it searches 85 million protein clusters from 131,744 genomes with unmatched speed and accuracy. "𝘎𝘢𝘪𝘢 𝘱𝘳𝘰𝘷𝘪𝘥𝘦𝘴 𝘵𝘩𝘦 𝘢𝘣𝘪𝘭𝘪𝘵𝘺 𝘵𝘰 𝘪𝘥𝘦𝘯𝘵𝘪𝘧𝘺 𝘳𝘦𝘭𝘢𝘵𝘦𝘥 𝘱𝘳𝘰𝘵𝘦𝘪𝘯𝘴 𝘸𝘪𝘵𝘩 𝘤𝘰𝘯𝘴𝘦𝘳𝘷𝘦𝘥 𝘨𝘦𝘯𝘰𝘮𝘪𝘤 𝘤𝘰𝘯𝘵𝘦𝘹𝘵𝘴 𝘸𝘪𝘵𝘩 𝘴𝘦𝘢𝘳𝘤𝘩 𝘴𝘱𝘦𝘦𝘥 𝘰𝘳𝘥𝘦𝘳𝘴 𝘰𝘧 𝘮𝘢𝘨𝘯𝘪𝘵𝘶𝘥𝘦 𝘧𝘢𝘴𝘵𝘦𝘳 𝘵𝘩𝘢𝘯 𝘦𝘹𝘪𝘴𝘵𝘪𝘯𝘨 𝘮𝘦𝘵𝘩𝘰𝘥𝘴. 𝘛𝘩𝘪𝘴 𝘤𝘢𝘱𝘢𝘣𝘪𝘭𝘪𝘵𝘺 𝘦𝘯𝘢𝘣𝘭𝘦𝘴 𝘳𝘦𝘴𝘦𝘢𝘳𝘤𝘩𝘦𝘳𝘴 𝘵𝘰 𝘢𝘥𝘥𝘳𝘦𝘴𝘴 𝘤𝘰𝘮𝘱𝘭𝘦𝘹 𝘲𝘶𝘦𝘴𝘵𝘪𝘰𝘯𝘴 𝘪𝘯 𝘤𝘰𝘮𝘱𝘢𝘳𝘢𝘵𝘪𝘷𝘦 𝘨𝘦𝘯𝘰𝘮𝘪𝘤𝘴, 𝘴𝘶𝘤𝘩 𝘢𝘴 𝘪𝘥𝘦𝘯𝘵𝘪𝘧𝘺𝘪𝘯𝘨 𝘤𝘰𝘯𝘴𝘦𝘳𝘷𝘦𝘥 𝘨𝘦𝘯𝘦 𝘤𝘭𝘶𝘴𝘵𝘦𝘳𝘴 𝘢𝘤𝘳𝘰𝘴𝘴 𝘥𝘪𝘴𝘵𝘢𝘯𝘵𝘭𝘺 𝘳𝘦𝘭𝘢𝘵𝘦𝘥 𝘰𝘳𝘨𝘢𝘯𝘪𝘴𝘮𝘴 𝘢𝘯𝘥 𝘦𝘭𝘶𝘤𝘪𝘥𝘢𝘵𝘪𝘯𝘨 𝘵𝘩𝘦 𝘧𝘶𝘯𝘤𝘵𝘪𝘰𝘯 𝘰𝘧 𝘩𝘺𝘱𝘰𝘵𝘩𝘦𝘵𝘪𝘤𝘢𝘭 𝘱𝘳𝘰𝘵𝘦𝘪𝘯𝘴 𝘣𝘢𝘴𝘦𝘥 𝘰𝘯 𝘵𝘩𝘦𝘪𝘳 𝘨𝘦𝘯𝘰𝘮𝘪𝘤 𝘯𝘦𝘪𝘨𝘩𝘣𝘰𝘳𝘩𝘰𝘰𝘥𝘴." Official announcement: https://lnkd.in/dteBsXaP #vectorsearch #science #vectordatabase #qdrant