Weaviate

Drag. Drop. Search. Done. 𝗣𝗗𝗙 𝗶𝗺𝗽𝗼𝗿𝘁 is now available directly through the Collections Tool in the Weaviate Cloud Console ✨ Here's what you can do: 📄 𝗗𝗿𝗮𝗴 & 𝗗𝗿𝗼𝗽 𝗣𝗗𝗙𝘀 Upload PDF files (up to 30MB) directly into the console. This automatically creates a new collection powered by 𝗺𝘂𝗹𝘁𝗶𝟮𝘃𝗲𝗰-𝘄𝗲𝗮𝘃𝗶𝗮𝘁𝗲 with ModernVBERT/colmodernvbert - giving you multi-vector embeddings that preserve visual document structure and elements, without needing OCR or chunking. 🔍 𝗜𝗻𝘀𝘁𝗮𝗻𝘁 𝗦𝗲𝗮𝗿𝗰𝗵𝗮𝗯𝗶𝗹𝗶𝘁𝘆 Your files are immediately queryable through our 𝗤𝘂𝗲𝗿𝘆 𝗔𝗴𝗲𝗻𝘁. No preprocessing pipeline. No complex chunking setup. Just upload and start asking questions. Perfect for: • Quick document analysis • Testing retrieval strategies on your own PDFs • Prototyping RAG applications • Exploring how multi-vector embeddings work in practice Try it out in your Weaviate Cloud Console! 💙 Sign in here: https://lnkd.in/dRfuSJpt

Legal RAG systems typically take 3-6 months to build. We did it in 36 hours. Then we made it possible in 𝗮 𝘀𝗶𝗻𝗴𝗹𝗲 𝗽𝗿𝗼𝗺𝗽𝘁. When our finance team asked us to help navigate internal contracts, we used Weaviate's 𝗤𝘂𝗲𝗿𝘆 𝗔𝗴𝗲𝗻𝘁 to turn raw legal documents into a fully functional assistant in just 𝘢 𝘥𝘢𝘺 𝘢𝘯𝘥 𝘢 𝘩𝘢𝘭𝘧. Building a traditional legal research tool typically takes 𝘮𝘰𝘯𝘵𝘩𝘴 of development time. Legal research is complex. You need extreme precision, absolute security, and the ability to filter by date, jurisdiction, or contract type. A naive RAG system collapses under this weight because it lacks reasoning. Ask about "2024 service agreements" and it might pull semantically similar clauses from 2022. 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝘀𝗲𝗮𝗿𝗰𝗵 changes this. The Query Agent treats your database as a set of tools rather than a static data store. It inspects your schema, constructs structured queries with the right filters, reranks results for actual relevance, and synthesizes grounded answers with citations. Here's the architecture we used: 𝗗𝗮𝘁𝗮 𝗟𝗮𝘆𝗲𝗿: PDFs embedded with ColQwen (a multivector model) and compressed with Muvera. Each page becomes a visual representation that preserves layout and tables. 𝗦𝗰𝗵𝗲𝗺𝗮: Three collections instead of one monolithic store - Commercial Agreements, Corporate & IP Agreements, and Operational Agreements. This gives the agent explicit structure to reason about. 𝗤𝘂𝗲𝗿𝘆 𝗔𝗴𝗲𝗻𝘁: The heavy lifter. It operates in Search Mode (retrieval and reranking for discovery) or Ask Mode (synthesized answers). Every response includes cited source passages to reduce hallucinations. With our new Weaviate Agent Skills, you can build this yourself with 𝗼𝗻𝗲 𝗽𝗿𝗼𝗺𝗽𝘁. Install Agent Skills: npx skills add weaviate/agent-skills Then run the prompt (available in our blog post) that tells the agent to build the full stack using the CUAD legal contract dataset, set up the three collections, configure the multivector embeddings, and create the frontend interface. The agent handles everything: downloading the dataset, embedding legal PDFs, creating the schema, and building the chat interface with source citations 🎉 Check out the blog here: https://lnkd.in/dNGYHBMH

🚀 We're growing our sales team at Weaviate (EMEA + US) We’re looking for proven AEs who drive new business and expansion while owning complex, technical sales cycles end-to-end. If this sounds like you, check out our openings below. We can't wait to connect! 📍 AE — EMEA: https://lnkd.in/gt3pBuJC 📍 AE — US East: https://lnkd.in/gRzi3xSW 📍 AE — US West: https://lnkd.in/gJbE-XKj 👉 https://lnkd.in/gdE9XE-3

Most founders dream of going viral. Aaron Edwards' viral tweet nearly killed his startup. The solution? A multi-tenancy strategy that now handles 6.1M+ queries. When Aaron launched DocsBot, a single viral tweet brought hundreds of signups overnight, his initial MVP couldn't handle the scale. Dealing with managing tens of thousands of unique, isolated customer indexes while maintaining strict data isolation - all as a solo founder without a dedicated infrastructure team. Aaron chose Weaviate because it was the only solution with an efficient tenant-based system that could scale to tens of thousands of distinct segmented indexes. As one of the first Weaviate Cloud customers, DocsBot partnered closely with the Weaviate team to architect a solution for their unique multi-tenant use case. The results speak for themselves: ✅ Answered 6.1M+ customer questions in a year ✅ Scaled to 50,000+ tenants in a single cluster ✅ Freed Aaron to focus on product innovation instead of infrastructure maintenance DocsBot's stack handles document ingestion, chunking, and embedding generation, then uses Weaviate for semantic and hybrid search to power grounded, accurate AI responses. And is now evolving from a simple answer engine to a true AI agent with capabilities like lead capture, support routing, and workflow automation - all powered by Weaviate. Read their full success story here: https://lnkd.in/dBWyQEVD

AI agents (Claude Code & Cursor) are incredible at writing code. But they keep guessing at vector DB implementations. We’ve all seen it: agents generating legacy v3 syntax, hallucinating hybrid search parameters, or failing on multivector strategies. You shouldn't have to fix the code your AI just wrote. That's why we built 𝗪𝗲𝗮𝘃𝗶𝗮𝘁𝗲 𝗔𝗴𝗲𝗻𝘁 𝗦𝗸𝗶𝗹𝗹𝘀. This repository bridges the gap between popular coding agents and Weaviate's infrastructure. Giving your coding agent the context it needs to write correct Weaviate code on the first try. The repo has two main components: 1. 𝗪𝗲𝗮𝘃𝗶𝗮𝘁𝗲 𝗦𝗸𝗶𝗹𝗹𝘀 (/skills/weaviate) - Focused scripts for specific Weaviate operations: schema inspection, data ingestion, precision search. These are the tools your agent reaches for when managing your Weaviate cluster. 2. 𝗖𝗼𝗼𝗸𝗯𝗼𝗼𝗸𝘀 (/skills/weaviate-cookbooks) - End-to-end project blueprints for building complete applications. Want a Query Agent chatbot with FastAPI and Next.js? A multivector PDF retrieval system? Various RAG pipelines? The cookbooks provide full-stack implementation guides. We’ve also included 6 specialized commands: • /weaviate:ask - Get AI-generated answers with sources using Query Agent • /weaviate:collections - List collections or inspect schemas • /weaviate:explore - Analyze collection data and property metrics • /weaviate:fetch - Retrieve objects by ID or with filters • /weaviate:query - Natural language search with Query Agent • /weaviate:search - Hybrid, semantic, or keyword search You can get started with just one line of code: npx skills add weaviate/agent-skills Set your environment variables (get a free sandbox cluster), run /weaviate:quickstart, and you're ready to build. We can’t wait to see what you’ll be building! 💚 Release blog post: https://lnkd.in/d5cW7YkH GitHub repo: https://lnkd.in/dTsBXbNT

When you move from prototype to production, getting access control right isn't optional. It’s essential. Storing embeddings of sensitive customer data, proprietary documents, or regulated information can be a challenge, but with Weaviate’s practical, layered security it becomes much easier: 🔐 Authentication - OIDC/SSO with providers like Okta and Azure AD, and runtime user management with API keys 🛡️ Authorization - Role-Based Access Control (RBAC) with custom roles and granular permissions 📋 Audit logging - Full visibility into who accessed what, and when Whether you're a solo dev getting started or a team preparing for production, the principle of least privilege is your friend. Read our new blog to learn how to implement these security measures on your Weaviate instance: https://lnkd.in/dyMj6a7H

Coding agents are only as good as the context they have. That’s why we’re releasing 𝗪𝗲𝗮𝘃𝗶𝗮𝘁𝗲 𝗔𝗴𝗲𝗻𝘁 𝗦𝗸𝗶𝗹𝗹𝘀. We're releasing 𝗪𝗲𝗮𝘃𝗶𝗮𝘁𝗲 𝗔𝗴𝗲𝗻𝘁 𝗦𝗸𝗶𝗹𝗹𝘀 - a repository that helps coding agents write better Weaviate code without hallucinating legacy syntax or guessing at parameters. Whether you're using Gemini CLI, Cursor, or Claude Code, the skills give the coding agent to access the right context to build state-of-the-art AI and agentic applications. The repo is organized into two tiers: 1️⃣ 𝗦𝗸𝗶𝗹𝗹𝘀 (/skills/weaviate): Focused scripts for schema inspection, data ingestion, and search operations. Your agent uses these when writing backend code. 2️⃣ 𝗖𝗼𝗼𝗸𝗯𝗼𝗼𝗸𝘀 (/cookbooks): Full end-to-end project blueprints (FastAPI + Next.js + Vector Database) for when you want to build entire applications. Agent Skills works with Claude Code, Cursor, GitHub Copilot, VS Code, Gemini CLI, and other tools. Get started right now with one line in your terminal: npx skills add weaviate/agent-skills Describe what you want in natural language: - "𝘊𝘳𝘦𝘢𝘵𝘦 𝘢 𝘞𝘦𝘢𝘷𝘪𝘢𝘵𝘦 𝘤𝘰𝘭𝘭𝘦𝘤𝘵𝘪𝘰𝘯 𝘧𝘰𝘳 𝘮𝘺 𝘑𝘚𝘖𝘕 𝘥𝘢𝘵𝘢 𝘤𝘢𝘭𝘭𝘦𝘥 '𝘗𝘳𝘰𝘥𝘶𝘤𝘵𝘴” - "𝘉𝘶𝘪𝘭𝘥 𝘢 𝘤𝘩𝘢𝘵𝘣𝘰𝘵 𝘶𝘴𝘪𝘯𝘨 𝘵𝘩𝘦 𝘘𝘶𝘦𝘳𝘺 𝘈𝘨𝘦𝘯𝘵” - "𝘍𝘪𝘯𝘥 𝘱𝘳𝘰𝘥𝘶𝘤𝘵𝘴 𝘴𝘪𝘮𝘪𝘭𝘢𝘳 𝘵𝘰 '𝘎𝘳𝘢𝘱𝘩𝘪𝘤 𝘵𝘦𝘦𝘴' 𝘪𝘯 𝘵𝘩𝘦 𝘗𝘳𝘰𝘥𝘶𝘤𝘵𝘴 𝘤𝘰𝘭𝘭𝘦𝘤𝘵𝘪𝘰𝘯” Check it out on GitHub: https://lnkd.in/dTsBXbNT Read the full release blog post: https://lnkd.in/dDhpxGUm

It's not you, it's your database. You two just don't speak the same language… Because you know what you want and the data's right there. But getting to it requires translating yourself into something the database understands and asking a bunch of things like: “Which collections do I need?” “Which filters won't nuke out everything?” “Do I need to manually merge results from three different places?” and so on. None of this is the actual question. It's just friction. The 𝗤𝘂𝗲𝗿𝘆 𝗔𝗴𝗲𝗻𝘁 kills the friction. You ask in plain English. The agent takes care of everything else: • Interprets your intent, not just the keywords • Identifies the right collections automatically • Constructs the correct queries and filters • Pulls from multiple data sources when needed • Validates results and retries if they don’t make sense • Maintains context, so follow-up questions just work You don’t need to memorize the schema or think in query syntax. You always stay focused on the question. It can be used in two modes depending on your use case: • 𝗔𝘀𝗸 𝗺𝗼𝗱𝗲 for getting an LLM-generated answer • 𝗦𝗲𝗮𝗿𝗰𝗵 𝗺𝗼𝗱𝗲 for returning only the raw objects Try it out directly in the Weaviate Cloud console: https://lnkd.in/d8kprCZF

The gap between a chatbot and an autonomous agent? 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴. Here's the blueprint: We've moved from tweaking prompts to orchestrating complete information architectures that power autonomous AI agents. 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 is the art of building dynamic systems that give LLMs exactly what they need, when they need it. What goes into a context-engineered system? 🧠 𝗠𝗲𝗺𝗼𝗿𝘆 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 - Short-term: Lives in the context window for current conversations - Long-term: Stored in vector databases (like Weaviate) for persistent knowledge 🔧 𝗧𝗼𝗼𝗹 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 APIs, calculators, search engines - everything your agent needs to take action. 📊 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗙𝗹𝗼𝘄 User preferences, retrieved context, conversation history, and agent reasoning - all working together. The shift? We're no longer asking "How do I write a better prompt?" We're asking "How do I provide the LLM all the context it needs to solve this task?" Our new e-book covers everything you need to know about context engineering: https://lnkd.in/dyqxstaW

The hardest part of building agentic AI isn't the AI - it's connecting all the pieces. Unless you use an integrated solution. We just built a production-ready app for our legal team to query contracts, TPAs, and NDAs using natural language. The Query Agent made this incredibly simple: ✅ Natural language queries over legal documents ✅ Autonomous decision-making (which collections to search, what filters to apply) ✅ Two query modes: search for relevant documents or ask questions directly ✅ Built-in security and compliance for sensitive legal data Because the Query Agent and Weaviate vector database are integrated, there's way less friction than connecting separate components. Our team can now instantly find specific clauses, compare contract types, or analyze agreement patterns - all through conversational queries. This isn't a demo - it's a real tool our legal council uses daily. And the speed of development shows just how powerful the Query Agent is for building production applications. Check out the demo: https://lnkd.in/dm8tbfwW Check out the query agent: https://lnkd.in/dhRmaRNn