1<!DOCTYPE html><!-- Last Published: Wed Jan 28 2026 19:57:51 GMT+0000 (Coordinated Universal Time) --> 2<html data-wf-page="66bdfeeeeb6efefc9befecc3" data-wf-site="64aceaf9cb10fc1c8e4efdbc" lang="en"> 3<head> 4 <meta charset="utf-8"> 5 <title>What is Raily AI and How we use it</title> 6 <meta content="Discover the thrill of new connections with Raily, the innovative mobile app transforming train travel! Unearth the secrets of our Terms & Conditions to unlock a world of social discovery. Make every journey memorable!" name="description"> 7 <meta content="What is Raily AI and How we use it" property="og:title"> 8 <meta content="Discover the thrill of new connections with Raily, the innovative mobile app transforming train travel! Unearth the secrets of our Terms & Conditions to unlock a world of social discovery. Make every journey memorable!" property="og:description"> 9 <meta content="What is Raily AI and How we use it" property="twitter:title"> 10 <meta content="Discover the thrill of new connections with Raily, the innovative mobile app transforming train travel! Unearth the secrets of our Terms & Conditions to unlock a world of social discovery. Make every journey memorable!" property="twitter:description"> 11 <meta property="og:type" content="website"> 12 <meta content="summary_large_image" name="twitter:card"> 13 <meta content="width=device-width, initial-scale=1" name="viewport"> 14 <link href="css/normalize.css" rel="stylesheet" type="text/css"> 15 <link href="css/webflow.css" rel="stylesheet" type="text/css"> 16 <link href="css/railybeta.webflow.css" rel="stylesheet" type="text/css"> 17
17<script type="a1766c051764a93771977d45-text/javascript">!function(o,c){var n=c.documentElement,t=" w-mod-";n.className+=t+"js",("ontouchstart"in o||o.DocumentTouch&&c instanceof DocumentTouch)&&(n.className+=t+"touch")}(window,document);</script>
17 18 <link href="images/favicon.png" rel="shortcut icon" type="image/x-icon"> 19 <link href="images/webclip.png" rel="apple-touch-icon">
vendor: 69 bytes, lines 19-20
19 20 <script async="" src="https://www.googletagmanager.com/gtag/js?id=
20G-K0DRW11MFK
vendor: 62 bytes, lines 20-21
20" type="a1766c051764a93771977d45-text/javascript"></script> 21
21<script type="a1766c051764a93771977d45-text/javascript">window.dataLayer = window.dataLayer || [];function gtag(){dataLayer.push(arguments);}gtag('set', 'developer_id.dZGVlNj', true);gtag('js', new Date());gtag('config', 'G-K0DRW11MFK');</script>
vendor: 1 bytes, line 21
21
22<script data-cfasync="false" nonce="279cd38f-1254-44c1-ac55-f0de731ba720">try{(function(w,d){!function(kP,kQ,kR,kS){if(kP.zaraz)console.error("zaraz is loaded twice");else{kP[kR]=kP[kR]||{};kP[kR].executed=[];kP.zaraz={deferred:[],listeners:[]};kP.zaraz._v="20";kP.zaraz._n="279cd38f-1254-44c1-ac55-f0de731ba720";kP.zaraz.q=[];kP.zaraz._f=function(kT){return async function(){var kU=Array.prototype.slice.call(arguments);kP.zaraz.q.push({m:kT,a:kU})}};for(const kV of["track","set","debug"])kP.zaraz[kV]=kP.zaraz._f(kV);kP.zaraz.init=()=>{var kW=kQ.getElementsByTagName(kS)[0],kX=kQ.createElement(kS),kY=kQ.getElementsByTagName("title")[0];kY&&(kP[kR].t=kQ.getElementsByTagName("title")[0].text);kP[kR].x=Math.random();kP[kR].w=kP.screen.width;kP[kR].h=kP.screen.height;kP[kR].j=kP.innerHeight;kP[kR].e=kP.innerWidth;kP[kR].l=kP.location.href;kP[kR].r=kQ.referrer;kP[kR].k=kP.screen.colorDepth;kP[kR].n=kQ.characterSet;kP[kR].o=(new Date).getTimezoneOffset();if(kP.dataLayer)for(const kZ of Object.entries(Object.entries(dataLayer).reduce((k$,la)=>({...k$[1],...la[1]}),{})))zaraz.set(kZ[0],kZ[1],{scope:"page"});kP[kR].q=[];for(;kP.zaraz.q.length;){const lb=kP.zaraz.q.shift();kP[kR].q.push(lb)}kX.defer=!0;for(const lc of[localStorage,sessionStorage])Object.keys(lc||{}).filter(le=>le.startsWith("_zaraz_")).forEach(ld=>{try{kP[kR]["z_"+ld.slice(7)]=JSON.parse(lc.getItem(ld))}catch{kP[kR]["z_"+ld.slice(7)]=lc.getItem(ld)}});kX.referrerPolicy="origin";kX.src="/cdn-cgi/zaraz/s.js?z="+btoa(encodeURIComponent(JSON.stringify(kP[kR])));kW.parentNode.insertBefore(kX,kW)};["complete","interactive"].includes(kQ.readyState)?zaraz.init():kP.addEventListener("DOMContentLoaded",zaraz.init)}}(w,d,"zarazData","script");window.zaraz._p=async kH=>new Promise(kI=>{if(kH){kH.e&&kH.e.forEach(kJ=>{try{const kK=d.querySelector("script[nonce]"),kL=kK?.nonce||kK?.getAttribute("nonce"),kM=d.createElement("script");kL&&(kM.nonce=kL);kM.innerHTML=kJ;kM.onload=()=>{d.head.removeChild(kM)};d.head.appendChild(kM)}catch(kN){console.error(`Error executing script: ${kJ}\n`,kN)}});Promise.allSettled((kH.f||[]).map(kO=>fetch(kO[0],kO[1])))}kI()});zaraz._p({"e":["(function(w,d){})(window,document)"]});})(window,document)}catch(e){throw fetch("/cdn-cgi/zaraz/t"),e;};</script>
22</head> 23<body class="body"> 24 <section class="section-mission"> 25 <div class="container"> 26 <div data-w-id="2b020496-b936-e88a-aa03-6f4b407a6a1c" style="opacity:0" class="box-text-terms"> 27 <h1 class="text-160 bracin">What is Raily AI and How we use it</h1> 28 <div class="box-text-raily"></div> 29 </div> 30 <div data-w-id="9d15e264-f047-53f4-aa14-e5e7808c61cf" style="opacity:0" class="wrapper-dropdown"> 31 <div class="rich-text-block-2 w-richtext"> 32 <p class="grin">Last Updated: 05.08.2024</p> 33 <p class="h2-doc">1. Introduction </p> 34 <p>At Raily, we harness the power of cutting-edge artificial intelligence (AI) to revolutionize the way people connect, travel, and experience the world. Our commitment to AI goes beyond mere implementation; it's about creating a seamless, intuitive, and enriching experience for our users while upholding the highest standards of privacy, security, and ethical use. We believe in transparency and responsible AI development, which is why we're sharing insights into how our AI works and the principles guiding its creation. </p> 35 <p> </p> 36 <p>The core mission of Raily AI is to connect people with 100% accuracy. We strive to find the perfect match for each user and guide them towards offline travel experiences in locations that are 100% suitable for them. This ambitious goal drives every aspect of our AI development and implementation. </p> 37 <p> </p> 38 <p>Some of the features described here are currently operational, while others are in various stages of development. We are committed to continuous improvement and innovation in our AI capabilities. </p> 39 <p> </p> 40 <p> </p> 41 <p class="h2-doc">2. Raily AI Features  </p> 42 <p class="h3-doc">2.1. Deep Social Matchmaking </p> 43 <p>Our Deep Social Matchmaking system is a multi-faceted AI-driven approach to connecting users: </p> 44 <p>1. Comprehensive Social Analysis: We use YOLO-World and CLIP-based models to analyze social media information, including posts, photos, and likes from platforms like Facebook and Instagram. </p> 45 <p>2. We incorporate data from user interactions within the Raily app, enriching user profiles with real-time behavior and preferences. </p> 46 <p>3. Data Vectorization & Anonymization: All collected data is converted into anonymized vector representations and stored in our secure Vector Database (Milvus), ensuring user privacy while enabling powerful matchmaking capabilities. </p> 47 <p>4. We employ advanced algorithms to find the best matches for users based on their vectorized data, considering both social media information and travel context. </p> 48 <p class="paragraph-doc">5. Our system includes a feedback loop that constantly updates and refines user profiles and matchmaking algorithms based on new data and user feedback. </p> 49 <p> </p> 50 <p class="h3-doc">2.2. Health-Integrated AI Concierge </p> 51 <p>Our AI Concierge system takes personalization to the next level by incorporating health data into our recommendations: </p> 52 <p>1. Health Data Integration: We securely integrate with platforms like Apple Health, Google Fit, and Samsung Health to access comprehensive health metrics. </p> 53 <p>2. Health data is converted into high-dimensional vectors and associated with anonymous user IDs, ensuring privacy and compliance with regulations like GDPR and HIPAA. </p> 54 <p>3. Based on the analysis of vectorized health data, we generate tailored suggestions for restaurants, cafes, fitness centers, and wellness activities. For example, we might recommend heart-healthy restaurants for users with high blood pressure or protein-rich options for highly active users. </p> 55 <p>4. For group travel, our AI considers the collective health needs and preferences of all members, ensuring that recommendations cater to everyone's requirements. </p> 56 <p class="paragraph-doc">5. Our system continuously updates recommendations based on real-time changes in users' health data and provides notifications about new or updated suggestions. </p> 57 <p> </p> 58 <p class="h3-doc">2.3. "Meets on the Move" Matchmaking System </p> 59 <p>This unique feature is designed specifically for travelers, enabling them to connect with others on the same routes: </p> 60 <p>1. Real-Time Location Verification: We use geolocation data, Bluetooth communication, and integration with transportation schedules to verify users' presence on specified routes. </p> 61 <p>2. Our AI considers a wide range of factors, including user interests, visual preferences, and travel patterns, to suggest highly compatible connections. </p> 62 <p>3. Privacy-Focused Design: All matching is done using anonymized vector representations, ensuring user privacy throughout the process. </p> 63 <p class="paragraph-doc">4. Multi-Modal Travel Support: The system works across various modes of transport, including trains, airplanes, cruise ships, and buses. </p> 64 <p> </p> 65 <p class="h3-doc">2.4. Digital Twin Identification </p> 66 <p>Our Digital Twin functionality uses AI to find users with remarkably similar characteristics: </p> 67 <p> </p> 68 <p>1. We consider a wide range of vectorized data points, including user interests, visual characteristics, personality traits, and travel patterns. </p> 69 <p>
692. Visual Characteristics Recognition: Our image recognition technology analyzes and vectorizes visual features such as hair color, eye color, facial structure, and clothing style. </p> 70 <p>3. The system examines and prioritizes matches based on similar travel histories and future travel plans. </p> 71 <p class="paragraph-doc">4. All data is converted into vector representations, ensuring that no original images or sensitive data are stored. </p> 72 <p> </p> 73 <p class="h3-doc">2.5. Dislike-Based Connections (patent pending, USPTO Nr. 63/651,023) </p> 74 <p>Unique to Raily, this feature connects users based on shared dislikes: </p> 75 <p>1. Automated Object Recognition: Using the YOLO deep learning model, we process visual data from device cameras or uploaded images to identify objects that users dislike. </p> 76 <p>2. Recognized objects are converted into vector representations, preserving privacy and forming the basis for our matching algorithm. </p> 77 <p>3. Users can easily express their dislikes through a straightforward interaction model, focusing solely on identifying disliked objects. </p> 78 <p>4. Our algorithm connects users with similar aversions, using similarity metrics such as Euclidean distance to compare vector representations. </p> 79 <p class="paragraph-doc">5. Users can refine object recognition results, contributing to the continuous improvement of our system's accuracy. </p> 80 <p> </p> 81 <p class="h3-doc">2.6. Raily+MERLIN AI Analysis System </p> 82 <p>Raily+MERLIN system offers advanced analytical capabilities based on user photos and data. This system can provide insights into various aspects of personality and behavior, including: </p> 83 <p> </p> 84 <p>1. Personality Type Analysis: Using established frameworks like Socionics and MBTI (Myers-Briggs Type Indicator) to provide insights into personality traits and potential compatibilities. </p> 85 <p>2. Behavioral Analysis: Assessing various behavioral traits such as curiosity, emotionality, and social interaction styles. </p> 86 <p>3. Cognitive Analysis: Evaluating traits like analytical and abstract thinking capabilities. </p> 87 <p> </p> 88 <p>This system is designed to enhance user matching and provide deeper insights for more meaningful connections, always prioritizing user privacy and ethical use of data. </p> 89 <p> </p> 90 <p class="h2-doc">3. Raily Architecture and Technology </p> 91 <p>Our AI system is built on a robust, multi-layered architecture designed for scalability, security, and performance: </p> 92 <p> </p> 93 <p>1. Application Layer: Includes our iOS and Android apps, web interface, and integrations with wearable devices and XR/VR platforms. </p> 94 <p>2. Services Layer: Houses our core AI services, including the Deep Social Matcher, AI Concierge, and various APIs for features like music matching and trip planning. </p> 95 <p>3. Platform Layer: Provides the foundation for our AI operations, including data management, model training and deployment, and analytics. </p> 96 <p>4. Infrastructure Layer: Ensures the reliability and security of our platform through cloud services, databases, and security measures. </p> 97 <p> </p> 98 <p class="paragraph-doc">This architecture allows us to deliver advanced AI capabilities while maintaining high standards of data privacy and system performance. </p> 99 <p> </p> 100 <p class="h3-doc">3.2. Core AI Technologies </p> 101 <p>At the heart of Raily's AI capabilities are several key technologies: </p> 102 <p> </p> 103 <p>1. Advanced Matching Algorithms: We use state-of-the-art algorithms to match users based on a wide range of factors, including interests, travel patterns, and visual preferences. </p> 104 <p>2. YOLO and CLIP Models: These deep learning models allow us to analyze visual data and bridge the gap between image and text understanding, enhancing our matching and recommendation capabilities. </p> 105 <p>3. Embeddings and Vector Representations: We convert various types of user data into high-dimensional vector representations, allowing for efficient processing and comparison of vast amounts of information. </p> 106 <p> </p> 107 <p>4. Similarity Search Algorithms: These algorithms quickly identify patterns and similarities across millions of data points, powering our matchmaking and recommendation systems. </p> 108 <p> </p> 109 <p>All these technologies work together to provide personalized, accurate, and privacy-preserving services to our users. </p> 110 <p> </p> 111 <p> </p> 112 <p class="h2-doc">4. User Privacy and Responsible AI </p> 113 <p>At Raily, protecting user privacy is paramount. Our approach to data handling and AI development is guided by the following principles: </p> 114 <p> </p> 115 <p>1. Data Minimization and Vectorization: We collect only necessary data and convert it into abstract vector representations. This allows us to provide personalized services while preserving user privacy. </p> 116 <p>2. Anonymization and Encryption: All user data is anonymized and encrypted using state-of-the-art techniques, including AES-256 encryption for storage and TLS for transmission. </p> 117 <p>3. Consent and Control: Users have full control over their data and can choose what information to share. We obtain explicit consent for data collection and use. </p> 118 <p>4. Compliance and Audits: Our practices comply with global data protection regulations. We conduct regular security audits and implement strict access controls. </p> 119 <p>5. Transparency: We are committed to clear communication about our data practices. Users can access or delete their data at any time. </p> 120 <p>6. Ethical AI Development: Our AI models are developed with careful consideration of potential biases and ethical impl
120ications. </p> 121 <p>7. Continuous Improvement: We regularly update our models and practices to adapt to new privacy challenges and technological advancements. </p> 122 <p> </p> 123 <p>By adhering to these principles, we ensure that our AI serves users' needs while respecting their privacy and upholding ethical standards. </p> 124 <p> </p> 125 <p class="h2-doc">5. Continuous Learning and Improvement </p> 126 <p>Our AI systems are designed to continuously learn and improve, ensuring that we deliver the best possible experience to our users: </p> 127 <p> </p> 128 <p>1. Feedback Loops: User interactions and feedback are constantly incorporated to refine our algorithms and improve accuracy. This helps us tailor our services to real user needs and preferences. </p> 129 <p>2. Regular Model Updates: We regularly update our AI models with new data to adapt to changing trends and user behaviors. This ensures our AI stays current and relevant. </p> 130 <p>3. Reinforcement Learning: Our systems adjust their recognition and matching algorithms based on collective user feedback, implementing a form of reinforcement learning from human feedback (RLHF). </p> 131 <p>4. Post-Training Process: After initial training, our models undergo a rigorous post-training process that includes: </p> 132 <ul role="list"> 133 <li>Supervised Fine-Tuning (SFT): We use high-quality, curated datasets to fine-tune our models for specific tasks. </li> 134 <li>Reinforcement Learning from Human Feedback (RLHF): We incorporate human preferences to further refine our models' outputs. </li> 135 <li>Continuous Evaluation: We regularly evaluate our models' performance using both automated metrics and human assessment. </li> 136 </ul> 137 <p>5. Expanding Knowledge Base: The number of concepts and scenarios our AI can understand and respond to is constantly growing. We are committed to expanding our AI's capabilities to offer increasingly accurate matches and more tailored recommendations. </p> 138 <p>6. Ethical Considerations: As we improve our AI, we continuously assess and mitigate potential ethical risks, ensuring our technology remains beneficial and respectful to all users. </p> 139 <p> </p> 140 <p>This commitment to continuous improvement allows us to offer a service that evolves with our users' needs while maintaining the highest standards of performance and ethical operation. </p> 141 <p> </p> 142 <p class="h2-doc">6. Control and Transparency </p> 143 <p>We believe in empowering our users with control over their data and transparency in how our AI operates: </p> 144 <p> </p> 145 <p>1. Data Control: Users can view, edit, or delete their data at any time through their account settings. </p> 146 <p>2. AI Interaction Visibility: When interacting with our AI features, users are clearly informed that they are engaging with AI-powered tools. </p> 147 <p>3. Feedback Mechanisms: We provide easy ways for users to provide feedback on AI interactions, helping us improve and address any concerns. </p> 148 <p>4. Clear AI Policies: Our AI usage policies are written in clear, understandable language and are easily accessible to all users. </p> 149 <p>5. Regular Updates: We keep our users informed about significant changes or improvements to our AI systems through our website and app notifications. </p> 150 <p> </p> 151 <p>By maintaining this level of transparency and user control, we aim to build trust and ensure that our AI serves the best interests of our users at all times. </p> 152 <p> </p> 153 <p class="h2-doc">Conclusion </p> 154 <p>AI is at the core of Raily's mission to revolutionize travel and social connections. By leveraging cutting-edge technologies like (but not limited) YOLO, CLIP, and advanced vector representations, we're able to offer uniquely personalized and privacy-focused experiences. From health-integrated recommendations to dislike-based connections, our AI-powered features are designed to make travel more enjoyable, social, and tailored to individual needs. </p> 155 <p> </p> 156 <p>Our ultimate goal is to achieve 100% accuracy in connecting people and guiding them to their perfect travel experiences. This ambitious mission drives our continuous innovation and improvement in AI technologies. </p> 157 <p> </p> 158 <p>As we continue to expand our AI capabilities, we remain committed to responsible AI use, prioritizing user privacy, data security, and ethical considerations in all our developments. We're excited about the future of AI in Raily and the endless possibilities it brings to enhance your travel experiences and connections. </p> 159 <p> </p> 160 <p> </p> 161 </div> 162 </div> 163 </div> 164 <a href="/#ai" class="link-back w-inline-block"><img src="images/Frame-1948757746.svg" loading="lazy" width="22" alt="" class="imag-back"></a> 165 </section> 166
166<script src="https://d3e54v103j8qbb.cloudfront.net/js/jquery-3.5.1.min.dc5e7f18c8.js?site=64aceaf9cb10fc1c8e4efdbc" type="a1766c051764a93771977d45-text/javascript" integrity="sha256-9/aliU8dGd2tb6OSsuzixeV4y/faTqgFtohetphbbj0=" crossorigin="anonymous"></script>
166 167
167<script src="js/webflow.js" type="a1766c051764a93771977d45-text/javascript"></script>
167 168
168<script type="a1766c051764a93771977d45-text/javascript"> 169$('[href="#to-footer"]').click(() => { 170 const isRedirectFromHome = sessionStorage.getItem('isFooterPopupOpened'); 171 if (isRedirectFromHome) { 172 history.back(); 173 } else { 174 document.location = document.location.origin + '#footer'; 175 } 176}) 177</script>
vendor: 143 bytes, lines 177-178
177 178<script src="/cdn-cgi/scripts/7d0fa10a/cloudflare-static/rocket-loader.min.js" data-cf-settings="a1766c051764a93771977d45-|49" defer></script>
178<script type="module" src="https://static.cloudflareinsights.com/beacon.min.js/v31edd6df95cf4e85bb4c19e7a9bdbcba1788362987495" integrity="sha512-iIg7k2xntmwu6/uSb5tpc/hySgZc4eoL31yB29W6tJFo2akwjPWcEqnCEdJvGexCL0KEQwVYv5BlowfhVz26hg==" data-cf-beacon='{"version":"2024.11.0","token":"7038f9abd12140aca0385cad04b55ebb","r":1,"spa":2}' crossorigin="anonymous"></script>
178 179</body> 180</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.