Qdrant

Building a Financial Agentic RAG Pipeline - What It Really Takes If you’re designing RAG systems for real-world, high-stakes domains like finance, basic retrieval + generation isn’t enough. That’s exactly the topic of Building a Financial Agentic RAG Pipeline by our Qdrant Star Tarun R Jain. In the article, Tarun walks through a modern pipeline that blends: - Optimized semantic retrieval using vector search - Agentic orchestration with LangGraph - Observability and experimentation with Weave And, Qdrant as the backbone vector store for efficient similarity search and context retrieval that powers agent decision-making. Here’s how Qdrant fits in: - Stores high-quality semantic embeddings of financial text - Enables lightning-fast similarity search for relevant data during reasoning - Supports hybrid and scalable retrieval required by agent-driven workflows This architecture is a great example of how retrieval (Qdrant) + orchestration (LangGraph) + observability (Weave) come together to build production-grade RAG systems that go beyond static Q&A and support multi-step logic and decision-making. 👉 Read the full report here: https://lnkd.in/gqEjuy-8 If you’re building RAG for complex use cases like finance, legal, or research workflows, this is a must-read. #AI #RAG #AgenticAI #VectorDatabases #Qdrant #GenAI #AIEngineering

🚨 Feb 25 | Live Demo: Audit-Ready AI in Finance Hosted by Qdrant × cognee If you’re building AI in a regulated environment, this session shows how to turn “cool RAG demos” into production-grade, explainable systems that risk, compliance, and regulators can trust. 👉 Register: https://luma.com/xvmyvj6j Live Use Case: Explainable Finance Q&A → Audit Trail Example prompts: Why are we recommending Portfolio A over B? What drove this risk flag? See the system generate: ✔ Answer ✔ Citations + snippets ✔ Provenance path ✔ Logged retrieval plan (audit-ready) We’ll cover: • When GraphRAG actually improves reliability (and when plain retrieval is enough) • Question → graph neighborhood → docs → answer → provenance • Production memory architecture (user/tenant boundaries, filters, scoped corpora) • What “audit-ready” outputs really require (evidence coverage, repeatable retrieval, evaluation signals) Plus: a real production case study (Xaver + Qdrant) on scaling compliant financial consultations with constrained retrieval and auditable outputs. For finance, AI/ML, platform, security & compliance leaders building AI in regulated environments. #AI #GraphRAG #FinTech #Compliance #RAG #MachineLearning

Instead of cutting documents into random, linear chunks of text (which often separates a heading from its explanation), POMA-AI reconstructs the document's internal hierarchy so that an AI can understand the relationship between different sections. We'll explore how POMA AI parses complex hierarchies into "Chunksets" and how Qdrant provides the high-dimensional space to analyze how these pieces connect. This isn't just about answering questions; it's about contextualizing the architecture of your information and finding patterns that humans miss.

Bazaarvoice is turning billions of product reviews into real-time shopping help and brand insights, and they needed retrieval that could scale without turning into an ops project. After moving from Postgres + pgvector to Qdrant, the team reported: - ~100x smaller vector footprint (about a 99% reduction) - Two AI products shipped on the same vector layer: Shopping Assistant (live) + AI Insights (pre-release) One line that stuck: “We’re shipping the AI products we actually bought this for now. And we’re doing it from a single place.” - Lou Kratz, Senior Principal Engineer, Bazaarvoice Full case study: https://lnkd.in/gBfeM52W

Optimizing Semantic Caching with Qdrant in Production In this detailed write-up, Kashyab Murali shares how he optimized a semantic cache backed by Qdrant to improve latency and efficiency in LLM-powered systems. Instead of relying on exact-match caching, the system uses Qdrant as a vector store to: - Store embeddings of previous queries and responses - Perform fast similarity search to detect semantically equivalent queries - Decide cache hits based on vector distance thresholds, not string matching The article dives into: - Why semantic caching breaks down without proper similarity tuning - How Qdrant’s vector search enables high-precision cache lookups - Trade-offs between recall, accuracy, and response freshness - Real production lessons from iterating on cache thresholds and retrieval logic This approach highlights how Qdrant can act as a semantic memory layer, making LLM applications faster, cheaper, and more scalable. 📖 Read the full article here: 🔗 https://lnkd.in/gEgsh74Q We love seeing the community build and openly share real-world optimizations using Qdrant. #Qdrant #SemanticCaching #VectorSearch #RAG #LLM #GenAI #ProductionAI #AIEngineering

We just ordered wireless Reachy Mini by Hugging Face 🤗 for embedding Qdrant.to/EDGE in its head 🧠 to enable it with information retrieval and context management skills. Physical-world agents (aka robots) are coming, and they will need to learn to search for missing information and remember conversations. Qdrant Edge is #opensource and already available to developers. Docs: https://lnkd.in/dhhjDCfz Code examples: https://lnkd.in/d6H97cwy #AI Glasses demo: https://lnkd.in/dMAFMrz3 I'm continuing to practise my non-AI design skills. 👻 Stay tuned. 😎

𝐐𝐝𝐫𝐚𝐧𝐭 𝐌𝐨𝐧𝐭𝐡𝐥𝐲 𝐎𝐟𝐟𝐢𝐜𝐞 𝐇𝐨𝐮𝐫𝐬 𝐢𝐬 𝐛𝐚𝐜𝐤 👨🏼🚀 📅 19th February 2026 🕔 17:00 CEST / 08:00 PDT 📍 Qdrant Discord Join us for a casual community hang-out to share what you’re building, ask questions, and connect with the Qdrant team & fellow devs. Special Guest: hafedh hichri from Chonkie What is Chonkie? Chonkie is a fast, developer-friendly tool for chunking and preparing data for AI apps, especially RAG pipelines. It helps break large documents into clean, efficient chunks so they can be embedded, indexed, and retrieved more accurately with vector search engines like Qdrant. 👉 Discord event: https://lnkd.in/gqBJREiU See you there! 👋

Jury Reveal — Multi-Agent Intelligence Systems (MAS) Track | Convolve 4.0 We are pleased to welcome an accomplished jury panel for the Multi-Agent Intelligence Systems (MAS) Track powered by Qdrant at Convolve 4.0 — A Pan-IIT AI/ML Hackathon. Manas Chopra: DevRel @ Qdrant | Co-Founder, Geek Room — one of India’s largest developer communities. Akshay Kumar Sharma: DevRel @ OLake | PA @ SWOC | Founding Member, Geek Room. With strong experience across developer relations, open ecosystems, and emerging AI infrastructure, they will be evaluating finalist solutions focused on collaborative agents, orchestration systems, and scalable multi-agent workflows. We look forward to their insights at the Convolve 4.0 Finale. #Convolve4 #AI #MultiAgentSystems #DeveloperEcosystem #Hackathon #IITG #ArtificialIntelligence

We have a really cool #VectorSpaceTalk coming up next week with POMA AI. Founder Dr. Alexander Kihm will be joining us to talk chunking strategies, especially with novel hierarchical chunking approaches. While chunking is not core to Qdrant usage, it's something relevant to many devs using Qdrant in their workflows. Dr. Kihm has prepared a colab notebook, some benchmarks, and great discussion topics. Be sure to join us at the link below. https://lnkd.in/gNWB_jkz

⏰ Happening Today | In a Few Hours Hallucinations are still one of the hardest problems to solve in production RAG systems - and today, we’re diving deep into how to tackle them the right way. Today at 8 AM PST / 5 PM CET, we’re hosting a live session on the Qdrant Discord with Kameshwara Pavan Kumar Mantha on: Hallucination Mitigation in RAG using LLM Steering What we’ll cover: - How Qdrant-powered retrieval improves factual grounding - Using LLM steering vectors to control model behavior - Designing robust RAG architectures for production - An end-to-end walkthrough: User query → Qdrant retriever → augmented context → LLM If you’re building RAG systems in production or struggling with LLM accuracy and hallucinations, this session is highly practical and hands-on. 👉 Join us live on Discord (starting in a few hours): https://lnkd.in/gRzwjbFH #Qdrant #RAG #LLM #HallucinationMitigation #GenAI #VectorDatabases #AIEngineering #LLMSteering