Use cases

DoorDash is a leading local commerce and delivery platform that connects consumers with food, groceries, and retail partners while helping businesses grow their online reach, enabling fast, same-day delivery across tens of thousands of merchants in over 40 countries.

Exa is an AI-powered web search engine and API built for modern AI applications, providing real-time semantic search and high-quality web data retrieval to help developers and AI agents access relevant information beyond traditional keyword search.

Airtable is a cloud-based platform that combines the simplicity of spreadsheets with the power of a database, enabling teams to build custom workflows, manage data, and collaborate across product, operations, and business processes without heavy engineering effort.

Roblox's platform team utilizes Milvus to support a variety of internal use cases, with the most significant being avatar search, leveraging Milvus' vector search capabilities to efficiently match users with avatar options, enhancing the customization and user experience on the platform.

NVIDIA adopts Milvus to power large-scale multimodal data search for autonomous driving, enabling engineers to run semantic queries over tens of billions of vectorized sensor data points. By replacing FAISS with Milvus, NVIDIA’s AV data platform improves scalability, operational reliability, and indexing efficiency for mining rare driving scenarios and validating real-world model behavior.

Reddit is a large online community platform where millions of users gather in topic-based forums to share content, ask questions, and discuss interests in real time across thousands of communities.

Filevine utilizes Milvus to enable semantic search across large volumes of unstructured legal data, allowing legal teams to quickly surface relevant case documents, communications, and insights. By integrating Milvus’ vector search capabilities into its legal work platform, Filevine improves information retrieval accuracy and reduces the time spent navigating complex case materials.

Read AI uses Milvus as the core of its semantic search infrastructure to retrieve relevant information from large-scale, unstructured communication data, including meetings, chats, emails, and internal knowledge bases. With Milvus’ high-performance vector search and multi-tenant scalability, Read AI delivers low-latency, accurate retrieval across billions of records for millions of active users.

OpenEvidence is a medical AI platform that helps clinicians quickly find evidence-based answers by searching and summarizing trusted clinical literature, enabling faster, more informed decision-making at the point of care.

Walmart employs Milvus across multiple use cases, with its platform team leveraging the vector database to support various internal applications, including enhancing product search capabilities by efficiently matching consumer queries with relevant product listings through advanced vector search technology.

Doximity is a professional networking and communication platform for clinicians, helping doctors securely connect, collaborate, and access medical news, tools, and career opportunities across the U.S. healthcare system.

Poshmark employs Milvus for product recommendation, utilizing vector search to analyze user behavior and preferences, efficiently matching buyers with relevant fashion items, thereby personalizing and enhancing the shopping experience.

Notta uses Milvus to power semantic search and RAG over large-scale meeting transcripts, enabling fast, context-aware retrieval across millions of conversations. With Milvus, Notta reduced search latency by 10× and delivers real-time meeting intelligence that turns discussions into actionable knowledge.

Plaud is an AI company focused on turning voice into structured, actionable knowledge. It combines AI-powered hardware and software to record, transcribe, and summarize conversations automatically. Plaud helps professionals capture ideas effortlessly and turn everyday conversations into lasting value.

Bosch uses Towhee and Milvus to enhance its manufacturing process by detecting defects in images, leveraging the combined power of Towhee's data processing and Milvus' vector search capabilities for improved quality control.

IBM is a global technology and consulting company delivering hybrid cloud, AI, automation, and enterprise software solutions to help organizations modernize, secure, and grow their businesses across industries.

Salesforce's platform team uses Milvus to support a wide range of internal use cases, serving 100+ tenants with diverse applications and varying service levels, leveraging Milvus' vector search technology to enhance functionality and performance across their extensive ecosystem.

Cisco is a global technology company that designs and delivers networking hardware, software, and secure cloud-enabled solutions to help organizations connect, protect, and automate digital operations across hybrid and multicloud environments.

eBay is a global e-commerce platform that connects millions of buyers and sellers through online marketplaces and apps, enabling auctions and fixed-price sales across diverse product categories in 190+ markets worldwide.

Fanatics is a global digital sports platform and online retailer offering officially licensed team apparel, fan gear, trading cards, collectibles, and more, while also expanding into sports betting and gaming to serve fans worldwide.

Fiverr is a global online marketplace that connects businesses and individuals with skilled freelancers for digital services across design, writing, tech, marketing, and more, making it easy to find and hire talent worldwide.

March Networks is a global provider of intelligent video surveillance and AI-powered video analytics solutions, helping enterprises improve security, reduce loss, and gain operational insights from video data.

Compass utilizes Milvus for custom search by vectorizing floorplans, enabling the search for homes beyond traditional keyword matches and enhancing the property discovery process with advanced spatial and feature-based queries.

New Relic is an AI-powered observability platform that helps engineers monitor, debug, and improve applications, infrastructure, and digital experiences in real time by unifying metrics, logs, traces, and telemetry across the full stack.

Accenture is a global professional services and technology consulting company that delivers strategy, digital, cloud, and AI solutions to help organizations transform and grow in a changing world.

Deloitte is a global professional services network delivering audit, consulting, tax, and advisory solutions to help organizations tackle complex challenges, drive growth, and build trust across industries worldwide.

Shell employs Milvus for multiple applications, primarily focusing on document search within a Retrieval-Augmented Generation (RAG) context, utilizing Milvus' vector search technology to enhance the accessibility and retrieval of documents in their vast corporate knowledge base.

Shopee employs Milvus to enhance its real-time search capabilities, particularly in video recall, copyright matching, and video deduplication systems, leveraging Milvus' ability to efficiently handle billions of vectors and scale with data volume, thereby improving user experience and maintaining content integrity.

Grab employs Milvus for its food and restaurant recommendation system, utilizing vector search to analyze customer preferences and match them with relevant dining options, enhancing the user experience through personalized suggestions.

ByteDance is a global internet technology company that builds AI-driven platforms for content creation, discovery, and social interaction, inspiring creativity and enriching life with products like TikTok and personalized media services.

NAVER deploys Milvus to support large-scale semantic and multimodal retrieval across search, content, and recommendation services, enabling natural language queries and improving relevance beyond keyword-based search.

Line utilizes Milvus to power its user-generated content (UGC) recommendation engine, enabling personalized news and music suggestions based on users' existing preferences through advanced vector search technology.

SoftBank Robotics Group is a global robotics company that develops and integrates AI-enabled robots and automation solutions to help businesses and people improve efficiency, safety, and daily operations across industries.

Rexera uses Milvus on Zilliz Cloud to power large-scale AI agents for real estate closings, enabling fast, accurate hybrid search across millions of documents and supporting real-time decision-making for over 10,000 automated tasks per day.

Omers' Data Science/Data Engineering team is utilizing Milvus to create a semantic search solution for financial documents, harnessing Milvus' vector database capabilities to efficiently index and query complex financial data for improved insight and retrieval.

IKEA utilizes Milvus for product recommendation, leveraging vector search technology to analyze customer preferences and match them with the most relevant products, enhancing the shopping experience through personalized suggestions.

Zomato utilizes Milvus for its restaurant search functionality, leveraging vector search to analyze and match user queries with the most relevant dining options, enhancing the discovery process for users looking for specific restaurant experiences.

Beatoven.ai uses Milvus on Zilliz Cloud to power its music similarity search and retrieval workflows, enabling accurate, scalable matching of moods, genres, and prompts to generate personalized, royalty-free soundtracks for millions of creators.

Bigo leverages Milvus for its efficient video deduplication system, enabling rapid similarity searches and high recall rates for its vast video content, significantly improving query throughput and operational efficiency by using Milvus' capabilities to handle and index large-scale vector data.

JERA is Japan’s largest power generation and energy company, operating the full supply chain from fuel procurement and LNG to power generation and wholesale energy solutions. Its services support stable, low-carbon energy supply and infrastructure domestically and globally across thermal, gas, and renewable projects.

Otter.ai leverages Milvus for Retrieval-Augmented Generation (RAG) to enhance Q&A functionality by efficiently sourcing and referencing relevant information from meeting transcripts, improving access to key insights and answers directly from recorded discussions.

Palo Alto Networks utilizes Milvus for fraud detection, employing its vector search capabilities to analyze and identify patterns indicative of fraudulent activity, thereby enhancing the security and integrity of their networks and services.

Chegg employs Milvus for storing document chunks and embeddings to enhance document search capabilities, powering the homework help chat feature on chegg.com, thereby streamlining access to educational resources and assistance.

AT&T leverages Milvus for document search, utilizing its vector search capabilities to fetch the most relevant semantic results for customers, significantly enhancing the accuracy and efficiency of information retrieval.

Tokopedia upgraded its product search and ranking by adopting Milvus for vector similarity search, enhancing semantic understanding and efficiency over Elasticsearch, resulting in a smarter, stable, and reliable Ads service with significantly improved click-through and conversion rates.

Mozat leverages Milvus for Stylepedia's image search system, utilizing its capability for real-time, large-scale vector similarity searches across billions of datasets for garment detection, feature extraction, and refined post-processing, enhancing user experiences with functions like searching for similar clothing items, outfit suggestions, and personalized fashion recommendations.

SmartNews leveraged Milvus for its high-throughput, low-latency vector similarity search capabilities to efficiently match users with relevant ads in real-time, significantly enhancing ad recommendation performance and scalability.

Farfetch employs Milvus as a vector database to support iFetch, its conversational AI for personalized shopping, by enabling real-time, accurate product recommendations through machine learning-generated product embeddings.

HumanSignal, formerly Heartex, is a leading provider of AI data labeling solutions and is known for its open-source platform, Label Studio. It integrates Milvus into its Label Studio platform to enhance data discovery and labeling processes. Milvus's support for various indexing algorithms improves semantic search efficiency, enabling users to identify relevant data subsets swiftly. Deployed on AWS using Elastic Kubernetes Service (EKS), this integration ensures scalable and reliable data management, significantly boosting model accuracy and development speed.

ZipRecruiter leverages Milvus to enhance the recruitment process by embedding candidates and job opportunities in the same vector space, enabling efficient and accurate matching between job seekers and relevant positions.

Shutterstock enhances its "Search by image" feature by leveraging Milvus, enabling reverse image search capabilities that efficiently match user-uploaded images with similar content in their vast digital asset library.

Zigram employs Milvus for fraud detection by comparing real-time transactions against a database of known fraudulent activities, utilizing vector search technology to quickly and accurately identify potential fraud, enhancing security and trust in their operations.

VerSe employs Milvus to enhance its recommendation engine for short videos by utilizing vector search to align viewer preferences with relevant video content, thus improving user engagement through personalized content suggestions.

VIPShop utilizes Milvus for its recommendation system, transforming product features and user behaviors into vector embeddings for fast, accurate similarity searches, outperforming traditional solutions with features like distributed deployment and multi-language SDKs.

PayPal utilizes Milvus for its impressive performance and scalability in handling large-scale data and AI use cases, starting with a recommender system and expanding to a multilingual customer service chatbot

Airbnb utilizes Milvus for its text-to-image search feature, enabling users to find rental listings by describing their desired features, such as "modernist building near the forest," through efficient vector search capabilities that match descriptive text with relevant property images.

Credal AI deployed Milvus on Amazon EKS to power semantic search in GenAI-driven workflows, enabling scalable, real-time processing of large datasets across diverse hosting environments, including cloud and on-prem setups.

Landing AI, founded by Andrew Ng, is a pioneer in Visual AI solutions, empowering industries to harness the full potential of visual data. By leveraging Milvus for image-based object detection, Landing AI enhances quality control in semiconductor manufacturing, specifically detecting disinfection in chip production processes. Using domain-specific Large Vision Models (LVMs) and Large Multimodal Models (LMMs), the platform ensures high accuracy and consistency in AI-driven inspections, transforming proof-of-concept AI projects into production-ready solutions.

TrendMicro adopted Milvus as a scalable, flexible vector search engine that overcame the limitations of MySQL and Faiss, enhancing APK security analysis with its advanced integration capabilities, intuitive API, and robust monitoring features, improving query performance and system scalability.

Sohu uses Milvus for semantic searches, optimizing news recommendations and classification to boost user experience and content relevance.

Deepset incorporates Milvus into their Haystack framework to boost semantic search capabilities. Milvus enhances data indexing and similarity searches, supporting dynamic data management and integrating with multiple Approximate Nearest Neighbours libraries.

The European Patent Office utilizes Milvus for patent search, leveraging its vector search capabilities to efficiently match queries with relevant patents, enhancing the precision and speed of patent retrieval and analysis.