# Personalized Semantic Search > Source: https://www.recombee.com/personalized-semantic-search > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Personalized # Semantic Search That Understands What Users Mean [Start Free ->](https://admin.recombee.com/sign-up) [Request Demo](https://www.recombee.com/request-demo) Proven Results ## Measurable Growth Across Industries * [![Apify](https://www.recombee.com/img/logos/apify.svg)Case Study ->+15%AI tool executions with personalized search](https://www.recombee.com/case-studies/apify) * [![Crexi](https://www.recombee.com/img/testimonials/crexi-logo.svg)Case Study ->+10%Buy actions from personalized search results](https://www.recombee.com/case-studies/crexi) * [![Sleepiest](https://www.recombee.com/img/logos/sleepiest.svg)Case Study ->+42%User retention from personalized search in the AI Sleep Guide](https://www.recombee.com/case-studies/sleepiest) Live Demo ## Search by Meaning, Not Just Keywords A live search over a demo movie catalog. Try queries such as “movies to watch with my 6yo”, “lonely astronaut stranded in space”, “mind-bending movies about dreams”, “movies about racing”. Key Features ## Search That Learns From Every User ### Semantic Understanding Understand the intent and context behind every query. An LLM helps users find what they’re looking for, even when their search terms don’t exactly match the content, such as “feel-good movies for a rainy evening” or “warm jackets for extreme cold”. Deliver intent-driven, context-aware search results even when search terms don’t directly match the content. Recombee’s search combines **full-text query** matching with machine learning on user interaction data. For enhanced accuracy, our Premium feature, **Semantic Search**, uses a large language model (LLM) to interpret the intent and context of user queries. This approach goes beyond keyword matching by capturing the deeper semantic meaning of each query, allowing the system to retrieve results that align with the user’s true intentions. With Semantic Search, you can: * **Understand What Users Mean:** results match the intent behind a query, not just the words in it * **Answer Natural Questions:** “movies about racing” or “eco-friendly jackets for extreme cold” return precisely relevant results * **Find Content Others Miss:** users get accurate, context-aware results even when their terms don’t directly match the content in your database ### Personalized Results Go beyond query relevance to reflect individual preferences. Machine learning uses each user’s interaction history and interests to personalize search rankings, so two people searching for the same thing see results tailored to them. Recombee’s search combines **full-text query** matching with machine learning on **user interaction** data to deliver precise, highly personalized results. Personalized search goes beyond just parsing search queries. It also takes into account a user’s interaction history on the site, fostering a more tailored and intuitive browsing experience. With Personalized Results, you can: * **Show Each User What Fits:** two people searching for the same thing each see first what matches their taste * **Grow Conversions and Sales:** aligning results with each user’s behavior and preferences lifts customer satisfaction, conversion rates and overall sales * **Narrow Results, Stay Personal:** filter by brand, size or price while the matches stay ranked for each user * **Forgive Typos:** even if users misspell a name, like a director’s, the right results still appear Personalized search delivered: * **+27% click-through rate** at [Cooklist](https://www.recombee.com/case-studies/cooklist) * **+15% actor runs** from existing users at [Apify](https://www.recombee.com/case-studies/apify) * **+10% buy actions** at [Crexi](https://www.recombee.com/case-studies/crexi) Explore Search Recipes For: [E-Commerce ->](https://docs.recombee.com/recipes/e-commerce/search/quick-search) [Video ->](https://docs.recombee.com/recipes/video/search-movies-and-series) [News ->](https://docs.recombee.com/recipes/news/search-articles) ### Real-Time Adaptability Keep search results fresh within milliseconds. React to new user interactions and catalog changes in real time, without waiting for scheduled model retraining. Thanks to our innovative incremental training of the models, **new interactions and new content** are taken into consideration within milliseconds. Results always reflect what users do and what is available right now, not what was true at yesterday’s training run. With Real-Time Updates, you can: * **React to Every Interaction:** what users view, click or buy shapes their next results * **Show What’s Available Now:** new items and catalog changes show up in results right away * **Surface the Latest Stories:** when readers search for “elections” or “football”, they get recent, currently popular articles that fit their reading habits ### Full Control & Flexibility Configure every search box as a Scenario directly in the Admin UI, without code changes. Choose the search Logic, apply ReQL filters and boosters to align results with business goals, and let the AI ReQL Assistant turn plain-language rules into working queries. Take charge with Scenario Settings, offering a suite of customization options to align search with your **product vision**. Each search box, referred to as a **Scenario**, can be uniquely tailored to meet diverse needs, such as a quick search in the header or a full search results page. All these configurations can be easily managed through our intuitive Admin UI, making customization a breeze. Recombee provides the essential tools to shape every search Scenario: * **Logic:** the ensemble of machine learning models that finds and ranks the results of the Scenario * **Filters:** decide which content can show up, e.g. articles from the last 7 days, promoted deals, apartments within 10 miles, or kid-friendly content for minor users * **Boosters:** bias results toward your business goals, e.g. newly added content, editors’ picks or higher-end alternatives for upselling * **Constraints:** manage diversity, e.g. no more than two items from each category or 50% from a single brand * **A/B Testing:** validate every change on real user behavior, with clear reports on improvement, statistical significance and the probability each Variant will win A **library of predefined Filters and Boosters** covers the most common needs, and anything beyond it can be described as a custom rule, written in plain language and turned into ReQL by the AI Assistant. ### Scalability Handle massive search volumes without compromising speed. Horizontally scalable infrastructure supports over 70k recommendations per second and catalogs of tens of millions of items, with globally distributed data centers ensuring low latency. Recombee operates on several hundred high-performance servers, hosting thousands of applications that interact within a micro-service architecture. This setup includes multiple layers of storage and distributed processing components, all equipped with self-healing and self-recovery capabilities. These features ensure that Recombee maintains high system availability and reliability. Users depending on our recommendations or search results are spread across the globe. Recombee operates its clusters in several geographic regions, so you can **choose the region** that is closest to your users or servers in terms of **latency**: US West Coast, Canadian East Coast, Europe (Germany) and Australia (Sydney). We build our components using a variety of industry-proven technologies. This ensures reliability, efficiency, and scalability in our systems, meeting the high standards required by our clients across diverse sectors. Because Recombee is built as a real-time, high-performance, distributed system, you can: * **Handle High Volume:** the same infrastructure serves over 70k recommendations per second, so search keeps up even on your busiest days * **Adapt Instantly:** real-time model updates and incremental training keep up with emerging trends and catalog changes * **Stay Online:** fault tolerance and high service availability keep search running * **Keep Data Safe:** robust security and data protection at all levels Today that means **over a billion recommendations served every day** and catalogs of **tens of millions of items**. [9GAG](https://www.recombee.com/case-studies/9gag) personalizes the experience for daily traffic of **hundreds of millions of users**, and at Konga.com, an e-commerce site with millions of users, Recombee **drives 24% of revenue**. ## The Building Blocks of Your Search ### Semantic Search Understand the intent and context behind every query. An LLM helps users find what they’re looking for, even when their search terms don’t exactly match the content. Deliver intent-driven, context-aware search results even when search terms don’t directly match the content. Recombee’s search combines **full-text query** matching with machine learning on user interaction data. For enhanced accuracy, our Premium feature, **Semantic Search**, uses a large language model (LLM) to interpret the intent and context of user queries. This approach goes beyond keyword matching by capturing the deeper semantic meaning of each query, allowing the system to retrieve results that align with the user’s true intentions. With Semantic Search, you can: * **Understand What Users Mean:** results match the intent behind a query, not just the words in it * **Answer Natural Questions:** “movies about racing” or “eco-friendly jackets for extreme cold” return precisely relevant results * **Find Content Others Miss:** users get accurate, context-aware results even when their terms don’t directly match the content in your database ### Quick Search Help users find what they’re looking for faster with suggestions as they type. Surface relevant items based on their query so far, personalized to each user’s interests. Quick Search enables your users to swiftly and effectively search for products that align with their preferences and activity on your platform. It runs **as the user is typing**, returning the most viable suggestions for the query so far. With Quick Search, you can: * **Suggest As Users Type:** fresh, relevant suggestions with every keystroke * **Personalize From the First Letter:** suggestions reflect each user’s preferences and activity * **Get Users There Faster:** what they are looking for is just a few keystrokes away, with no endless scrolling * **Go Live Fast With the Quick Search Widget:** configure it visually in the Admin UI, arrange its sections and match its look to your brand, then paste its embed snippet into your site [Quick Search Widget Docs ->](https://docs.recombee.com/no-code-widgets#html-widget-full-text-search) ### Faceted Search Make it easy to narrow down results by attributes such as brand, size, price, or category, while keeping results personalized to each user. Faceted search is a widely used feature that enables users to **filter products based on various attributes**. It is essential for sites with a large inventory, letting customers quickly find products that match their interests. With Faceted Search, you can: * **Filter by Any Attribute:** brand, size, price or any other property of your items * **Keep Every Result Personal:** the matches are still ranked by each user’s interactions * **Help Shoppers Find It Fast:** even across the largest catalogs [Faceted Search Recipe ->](https://docs.recombee.com/recipes/e-commerce/search/faceted-search) ### Search Item Segments Go beyond individual items and help users discover entire groups of related content or products, from categories and brands to genres and artists. Search can return **whole groups** instead of single items: a search for “iPhone” can return the “cell phones” category. These Item Segments are personalized by the user’s past interactions too. With Search Item Segments, you can: * **Find Whole Categories:** lead users to the category or brand that holds what they want * **Search Genres and Artists:** let users find the genre or artist, not just a single title or song * **Set It Up Without Code:** choose the Segmentation in the Scenario settings of the Admin UI [Search Item Segments Docs ->](https://docs.recombee.com/api/search#search-item-segments) ### Personalized Ranking Put the most relevant results first for every user. Use individual interaction history and preferences to personalize rankings, so the same query can deliver different results to different people. The results are based on the query and on the user’s **past interactions**, such as purchases or ratings: items more suitable for the user are preferred. With Personalized Ranking, you can: * **Put the Best Match First:** the most relevant item for each user tops the list * **Search Your Whole Catalog:** every text and set attribute of your items is indexed * **Learn From Every Search:** clicks on results tell Recombee which searches worked, so results keep improving [Search Items Docs ->](https://docs.recombee.com/api/search#search-items) ### Search Synonyms Help users find what they’re looking for, whatever terms they use. Define synonyms in the Admin UI or through the API to connect different search terms to the same results. When a term is used in a query, **its synonym is used for the search too**: “science fiction” for “sci-fi”. With Search Synonyms, you can: * **Match Every Way of Asking:** different words for the same thing find the same items * **Control the Direction:** a synonym works both ways, or one way only * **Manage It Without Developers:** add synonyms in the Admin UI, or through the API [Synonyms Docs ->](https://docs.recombee.com/api/search#add-search-synonym) ### A/B Testing Find out which search strategies work best before rolling them out to everyone. Test different search variants and configurations to compare their impact on click-through rates, conversions, and other key metrics. A/B Testing helps you validate changes to your search strategy using **real user behavior**. Run an experiment within any search Scenario and automatically split traffic between the current configuration, the Control, and one or more Variants. With A/B Testing, you can: * **Test Every Change:** compare different Logics and their settings, Filters, Boosters, and Constraints on real users * **Measure What Matters:** track the impact on Click-Through Rate, Conversion Rate, Revenue, or any custom metric defined in [Insights](https://www.recombee.com/real-time-analytics-insights) * **Know the Winner:** clear reports show the improvement over the Control, statistical significance, and the probability that each Variant will outperform the others * **Roll Out With Confidence:** ship optimizations backed by real user behavior, not guesswork ### ReQL Control exactly what appears in search results with flexible filters and boosters. Define business rules, from hiding out-of-stock items to promoting new arrivals, and let the AI ReQL Assistant turn plain English into working queries. With Filters and Boosters, you can hide what is out of stock, promote new arrivals, and highlight specific brands, editor-picked products, local deals, and more, helping every search box return **results that fit your business**. While our predefined rules cover a wide range of scenarios, you might find a unique requirement that calls for a custom filtering or boosting rule. No problem, **Recombee’s ReQL language** is built to handle those situations with ease. It is a powerful query language specifically designed for expressing Filters and Boosters based on the attributes of both items (content or products) and users. With ReQL, you can: * **Express Any Business Rule:** craft intricate rules with functions for geo-filtering, processing user interactions, and more * **Preview Before You Publish:** the Admin UI’s ReQL editor and validator show the outcome of your Filters and Boosters * **Reuse Rules Everywhere:** every rule joins your Filters and Boosters library, ready for any Scenario, with parameters like recency or distance that anyone can adjust * **Write Rules in Plain English:** the **AI Assistant** turns natural-language requests into working ReQL, and explains existing rules in plain terms For example, type **“Show only products in stock and boost new arrivals”** and the **AI Assistant** writes the rule for you. Or make sure users shopping the kids’ section only find age-appropriate products. [ReQL Docs ->](https://docs.recombee.com/reql) ## A Few Steps to Personalization ## Ship Personalized Search With or Without Code ### AI Agents Ready If you’re using an AI agent or assistant to integrate Recombee into your application, the documentation includes [an integration guide](https://docs.recombee.com/ai_assistant_guide) written specifically for AI assistants. Use the ready-made prompt below as a starting point. #### Initial Prompt Integrate Recombee (a recommendation engine) into my application. Before writing any code, fetch and read the integration guide at https://docs.recombee.com/ai_assistant_guide.md. It's written specifically for AI assistants and covers authentication, catalog sync, interactions, recommendations, and common pitfalls. Use it as your primary reference. First, explore the codebase to understand the stack, data models, and integration points. Then propose a plan before implementing anything. Do not decide on your own where recommendations should appear. Ask me which use cases to implement — homepage recommendations, related items on a product page, personalized search, email suggestions, and so on. The right surfaces depend on my product and priorities, so treat that as a shared decision even when you're otherwise working autonomously. After we agree on the use cases, implement them end to end, and flag any assumptions you have to make along the way. * [No-Code WidgetSet up a Quick Search widget in the Admin UI and add it to your site with an embed code.->](https://docs.recombee.com/no-code-widgets#html-widget-full-text-search) * [Widget SDKsPre-built, customizable components for vanilla JavaScript and React.->](https://docs.recombee.com/widget-sdks#quick-search) * [API SDKClient-side and server-side clients for multiple programming languages.->](https://docs.recombee.com/api_clients) ## Personalization Powerhouse works best as a whole. [![](https://www.recombee.com/img/homepage/recommendations.png)Real-TimeRecommendations->](https://www.recombee.com/real-time-recommendations) [![](https://www.recombee.com/img/homepage/analytics.png)Real-TimeAnalytics & Insights->](https://www.recombee.com/real-time-analytics-insights)