SureIn

surein

How to Prompt Engineer Phi-3-mini: A Practical Guide

Hoe om 'n ingenieur Phi-3-mini te spoor: 'n Praktiese gids

Πώς να καλέσετε τον μηχανικό Phi-3-mini: ένας πρακτικός οδηγός

Cómo impulsar al ingeniero Phi-3-mini: una guía práctica

כיצד להזמין מהנדס Phi-3-mini: מדריך מעשי

इंजीनियर Phi-3-mini को कैसे प्रोत्साहित करें: एक व्यावहारिक गाइड

Hogyan kell a mérnök Phi-3-mini: egy gyakorlati útmutató

エンジニアPhi-3-miniをプロンプトする方法:実践ガイド

Ahoana no ahazoan'ny injeniera Phi-3-mini: Manoro hevitra mahomby

Како да се поттикне инженер Phi-3-мини: Практичен водич

Инженер Phi-3-mini-ийг хэрхэн дэмжих вэ: практик гарын авлага

மெனுவில் தோன்றும், Staff details ஐ Click செய்யவும்.

Paano Mag-prompt Engineer Phi-3-mini: Isang Praktikal na Guide

The discourse around building accessible websites tends to focus on physical disabilities - visual impairments (alt tags, screen reader compatibility), deafness (captions), color blindness (nothing should be signified by changes in color alone). However, mental health conditions such as PTSD, ADHD, autism, and depression are also considered disabilities under the American with Disabilities Act (ADA), and can affect how users navigate and interact with the web. The accessibility needs of people with mental health related conditions are often overlooked in mainstream accessibility conversations, in spite of the fact that mental health-related disabilities affect a large number of Americans. According to the World Health Organization, 1 in 8 people live with a mental disorder, which is nearly 1 billion people worldwide.  In the US, it is estimated that there are: According to the World Health Organization 13 million people with PTSD (US Department of Veteran Affairs)
21 million adults and 3.7 million teens with Depression (Mental Health America)
15.5 million adults and 7.1 million children with ADHD (CDC - adults, children)
5.5 million adults with autism (CDC)
40 million people with dyslexia (Discovery ABA)
of people over 65, 10% have dementia and 22% have mild cognitive impairment (Columbia University) 13 million people with PTSD (US Department of Veteran Affairs) US Department of Veteran Affairs 21 million adults and 3.7 million teens with Depression (Mental Health America) Mental Health America 15.5 million adults and 7.1 million children with ADHD (CDC - adults, children) adults children 5.5 million adults with autism (CDC) CDC 40 million people with dyslexia (Discovery ABA) Discovery ABA of people over 65, 10% have dementia and 22% have mild cognitive impairment (Columbia University) Columbia University In this blog, we’ll explore common barriers on web pages faced by users with mental health challenges and highlight design strategies that foster a safer, more accessible digital experience. What Makes the Web Difficult for Users with Mental Health Conditions? Every user is different, but some challenges are common within and even across conditions. These include: Susceptibility to Overwhelm/Cognitive Overload Many, if not all, mental health conditions make a person more easily overwhelmed. Experiencing overwhelm can cause someone to have impaired comprehension, make more mistakes, take impulsive actions they later regret, forget what they were doing, or need to leave the website or pause other responsibilities in order to regain emotional regulation. Common web designs that can cause overwhelm include cluttered layouts, inconsistent navigation, long and overly complex forms, and long blocks of text without paragraph breaks, images, or white space to break up the text. Sensory Issues and Overstimulation When a user is overstimulated, they are more easily overwhelmed. Additionally, overstimulation from sensory issues can be viscerally painful, and trigger migraines or panic attacks. Overstimulation can be caused by cluttered designs, and especially animations, video, audio, and flashing lights. Sensory issues are most common in people with autism, though it can affect people with other conditions as well. Susceptibility to Dysphoria Dysphoria is a state of feeling deeply unhappy or uneasy, often to the point of disruption, where the person can’t function to their full potential. Many mental health conditions, in particular depression, bipolar disorder, anxiety disorders, and schizophrenia, can make individuals more prone to experiencing dysphoria. Web content that triggers dysphoria can have disabling effects for hours or even days, and in extreme cases, can trigger suicidal thoughts. Susceptibility to Flashbacks to Traumatic Events and Heightened Startle Responses Similar to susceptibility to dysphoria, people with PTSD can have flashbacks to traumatic events triggered by content on the web. Flashbacks can cause debilitating symptoms, such as anxiety, headaches, other physical pain, intrusive painful memories, fatigue, inability to concentrate, irritability, and suicidal thoughts, and it can take days or even weeks for flashback related symptoms to fully resolve. People with PTSD also have a heightened startle response, so a surprising action or sound - such as modals, pop ups, animations, etc - can trigger a disruptive stress response. Memory Issues Users with memory issues can have trouble accessing services if forms or other flows are long or complex, if there are long load times, if the page changes during flows without clear summaries of where the user is in the process, or if forms only use placeholders to tell the user what the input is asking for (placeholders disappear once a user starts typing). Impaired working memory affects users with age related memory loss, and can also affect users with other conditions like PTSD. Intense Distractibility A user with intense distractibility will not be able to fully engage with content on pages that has distractions - they will end up distracted and unable to access what they are looking for, and can end up distracted to the point that it disrupts their other plans and responsibilities for the day. Intense distractibility is the hallmark of ADHD, though other mental health conditions can also involve decreased concentration and focus. Distractions can include things that actively draw the users attention away from the information they need (audio, video, animations that cannot be turned off, complex layouts), and it can also include designs that make it difficult to stay focused on the content (ie, the delivery is boring or overwhelming, such has overly verbose or academic writing style, or long blocks of text without breaks such as white space or images). Best Practices for Making Web Pages Accessible to Users with Mental Health Conditions Many practices will support users across a variety of conditions. Avoiding cognitive overload, overwhelm, and confusion is one of the most fundamental aspects of building websites that are accessible to people with mental health issues, as all mental health conditions can increase a person’s susceptibility to overwhelm. This means designing simple, clear, and easily understood online experiences. This includes: Clear content with simple words, short blocks of text that are supported by images and broken up by white space
Use literal language - no jargon, acronyms, slang, or word play (this also makes content to more accessible to users across the globe, who aren’t familiar with local slang and idioms or may struggle to understand word play)
Designs that make errors less likely (like spell check, auto suggestions, and making sure your forms support autofill from saved browser data, allowing users to see and check their answers before submitting)
Clear and specific instructions (for instance, if your forms require a specific format for the information, stating that - such as “Date of birth (MM/DD/YYYY):”
Clear and specific error messages (ie, “Phone number is not in a valid format, the format should be (xxx) xxx-xxxx” vs “Error in form”)
Designs that make correcting errors as easy as possible (ie, auto-focus the input in the form which has invalid data, offer “go back” and “undo” options)
Intuitive navigation that is consistent across all pages and includes breadcrumbs
Simple and straightforward layouts Including progress bars and lengths on videos and audios so users have a clear sense of how much more they have to do
Ensure that all links and buttons are visibly clickable - the cursor changes to a pointer when they hover over the element
Avoid timeouts on forms
Use positive reinforcement and success messages to communicate clearly to the user that they have successfully completed their action Offering help via a variety of formats. For instance, some users with anxiety disorders are afraid of phone calls, while other users with cognitive impairments and learning disabilities find phone calls reduce their anxiety, and users with dyslexia may struggle with written support chats. Clear content with simple words, short blocks of text that are supported by images and broken up by white space Use literal language - no jargon, acronyms, slang, or word play (this also makes content to more accessible to users across the globe, who aren’t familiar with local slang and idioms or may struggle to understand word play) Designs that make errors less likely (like spell check, auto suggestions, and making sure your forms support autofill from saved browser data, allowing users to see and check their answers before submitting) Clear and specific instructions (for instance, if your forms require a specific format for the information, stating that - such as “Date of birth (MM/DD/YYYY):” Clear and specific error messages (ie, “Phone number is not in a valid format, the format should be (xxx) xxx-xxxx” vs “Error in form”) Designs that make correcting errors as easy as possible (ie, auto-focus the input in the form which has invalid data, offer “go back” and “undo” options) Intuitive navigation that is consistent across all pages and includes breadcrumbs breadcrumbs Simple and straightforward layouts Including progress bars and lengths on videos and audios so users have a clear sense of how much more they have to do Ensure that all links and buttons are visibly clickable - the cursor changes to a pointer when they hover over the element Avoid timeouts on forms Use positive reinforcement and success messages to communicate clearly to the user that they have successfully completed their action Offering help via a variety of formats. For instance, some users with anxiety disorders are afraid of phone calls, while other users with cognitive impairments and learning disabilities find phone calls reduce their anxiety, and users with dyslexia may struggle with written support chats. Avoiding overstimulating or distracting content is also a strategy that supports users with a variety of conditions, such as users with ADHD, autism, and sensory issues. This includes: No content that involves sound should play by default
All content that involves sound or motion, such as animation, audio, video, and image carousels, should be able to be paused, stopped, or hidden
Only using pop-ups when they provide expected information (for instance, a confirmation message that the task is successfully completed)
Avoiding designs with automatic changes on the screen No content that involves sound should play by default All content that involves sound or motion, such as animation, audio, video, and image carousels, should be able to be paused, stopped, or hidden Only using pop-ups when they provide expected information (for instance, a confirmation message that the task is successfully completed) Avoiding designs with automatic changes on the screen Making sure your content is respectful to people with mental health related disabilities is another important aspect of building an accessible page. This means avoiding any language that uses mental health related words (or disability-related terms of any kind) as derogatory terms, such as insane, crazy, psycho, psychotic, schizo. Condition-Specific Accessibility Strategies: Memory-Related issues In order to provide an accessible digital experience for users with impaired memories, a website should reduce the ways users need to rely on memory. This can be done by: Displaying the search term a user searched outside of the search bar, so it’s clear what the results refer to, even if the user has typed something else in the search input
Including breadcrumb navigation so the user knows where they are
Including summaries of what is happening on the page a user is on, especially if they are in a multi-page flow
Having reasonable password creation requirements Displaying the search term a user searched outside of the search bar, so it’s clear what the results refer to, even if the user has typed something else in the search input Including breadcrumb navigation so the user knows where they are breadcrumb navigation Including summaries of what is happening on the page a user is on, especially if they are in a multi-page flow Having reasonable password creation requirements PTSD - Trigger warnings and content previews for sensitive content The most important aspect of providing an accessible experience for people with PTSD is requiring content warnings or previews. “Trigger warnings” have received ridicule as “coddling” people, with the idea that they are about individuals avoiding confrontation or being challenged. This is a common misunderstanding of what “trigger warnings” are, however. The purpose of trigger warnings is to help individuals with PTSD avoid having flashbacks - and a flashback is much more than being upset, it triggers a deep internal perception of immediate, life-threatening danger, which will disrupt the person’s functioning for hours, even as long as days or weeks. Providing trigger warnings allows users to choose to engage with the content, and avoid content that may cause PTSD-related symptoms. ADHD To create an accessible digital experience for users with ADHD, the web page should be designed to reduce the likelihood of distractions - this means both avoiding distracting features, as well as presenting content in a way that does not require a great deal of focus or concentration to engage with. As previously mentioned, this includes: Avoiding any sounds, videos, or animations that play by default
Allowing users to pause, stop, and hide any animations, videos, or audios
Avoiding long, unbroken blocks of text - text should be broken up with white space, images, or other content
Avoiding cluttered and complex layouts Avoiding any sounds, videos, or animations that play by default Allowing users to pause, stop, and hide any animations, videos, or audios Avoiding long, unbroken blocks of text - text should be broken up with white space, images, or other content Avoiding cluttered and complex layouts Other ways to support users with ADHD include: Present information in streamlined formats - include only key info above the fold, and provide more thorough details and information below the fold or in expandable components
Keep carousels or other rotating banners on slow timers
Keep the number of options small - users with ADHD can experience decision paralysis.
When possible, provide only one call to action, and make sure that call to action is clear and prominently displayed.
Use minimalist designs with plenty of white space (Apple is a good example of this kind of design) Present information in streamlined formats - include only key info above the fold, and provide more thorough details and information below the fold or in expandable components Keep carousels or other rotating banners on slow timers Keep the number of options small - users with ADHD can experience decision paralysis. When possible, provide only one call to action, and make sure that call to action is clear and prominently displayed. Use minimalist designs with plenty of white space (Apple is a good example of this kind of design) Customizable User Experiences Some websites have specific profiles users can select to tailor their web experience to their needs. For instance, accessiBe is an agency that implements accessibility options on client web pages, and their offering includes an ADHD Friendly Profile which boasts “more focus and fewer distractions”, a Seizure Safe Profile, a Cognitive Disability Profile, and others. However, some reviews have stated that accessiBe’s profiles aren’t compatible with screen readers, and their ADHD-specific design was distracting to some users. accessiBe Another example of customizable experiences are sites that allow users to set color themes. Users with visual impairments will likely prefer higher contrast colors and designs, whereas users with sensory issues can be overstimulated by high contrast sites and prefer lower contrast color themes. Color themes that support users with color blindness can also be offered. Conclusion Building accessible websites for people with mental health-related disabilities isn’t just about compliance—it’s about empathy, inclusion, and thoughtful design. By implementing strategies like reducing cognitive load, avoiding triggering content, providing clear navigation, and offering customization options, we can make the web a more supportive and empowering place for everyone. As designers and developers, we have the power to shape digital spaces that respect mental well-being and embrace neurodiversity. Let’s build with care, and create experiences where no one is left behind. https://pmc.ncbi.nlm.nih.gov/articles/PMC8869505/
https://www.w3.org/WAI/people-use-web/abilities-barriers/cognitive/
https://www.verndale.com/insights/accessibility/website-design-trends-for-dyslexia-and-adhd
https://www.inclusionhub.com/articles/improving-wcag-for-mental-health
https://cacm.acm.org/research/considering-trauma-in-accessible-design-for-adults-with-intellectual-and-developmental-disabilities/
https://www.wcag.com/blog/digital-accessibility-and-neurodiversity/
https://edoc.ub.uni-muenchen.de/26029/7/Bernard_Renaldo.pdf
https://medium.com/@madisondunaetz/trauma-informed-ux-design-1621957dd097 https://pmc.ncbi.nlm.nih.gov/articles/PMC8869505/ https://pmc.ncbi.nlm.nih.gov/articles/PMC8869505/ https://www.w3.org/WAI/people-use-web/abilities-barriers/cognitive/ https://www.w3.org/WAI/people-use-web/abilities-barriers/cognitive/ https://www.verndale.com/insights/accessibility/website-design-trends-for-dyslexia-and-adhd https://www.verndale.com/insights/accessibility/website-design-trends-for-dyslexia-and-adhd https://www.inclusionhub.com/articles/improving-wcag-for-mental-health https://www.inclusionhub.com/articles/improving-wcag-for-mental-health https://cacm.acm.org/research/considering-trauma-in-accessible-design-for-adults-with-intellectual-and-developmental-disabilities/ https://cacm.acm.org/research/considering-trauma-in-accessible-design-for-adults-with-intellectual-and-developmental-disabilities/ https://www.wcag.com/blog/digital-accessibility-and-neurodiversity/ https://www.wcag.com/blog/digital-accessibility-and-neurodiversity/ https://edoc.ub.uni-muenchen.de/26029/7/Bernard_Renaldo.pdf https://edoc.ub.uni-muenchen.de/26029/7/Bernard_Renaldo.pdf https://medium.com/@madisondunaetz/trauma-informed-ux-design-1621957dd097 https://medium.com/@madisondunaetz/trauma-informed-ux-design-1621957dd097

A Developer’s Guide to Accessibility For Mental Health and Cognitive Disabilities

This One Spark SQL Trick Will Instantly Upgrade Your Data Analysis Game

Let me start with a confession. In one of my previous projects, I used to think project chaos was just part of the job. You know the vibe — people stepping on each other's toes, endless Slack threads trying to figure out who was supposed to do what, decisions made in meetings that no one remembered a week later. At some point, you just accept it as the cost of doing cross-functional work, right? Wrong. That's just lazy leadership. The more projects I led — across teams big and small, from product rewrites to data platform migrations — the more obvious it became: projects don't fall apart because of bad tech. They fall apart because of bad clarity. And I don't mean "bad Jira grooming" or "a lack of alignment on deliverables" (although, yeah, those suck too). I mean fundamental confusion about who owns what. Who makes the call? Who's doing the actual work? Who gets a say — and who just needs to be updated once it's done? If you've ever ended a week thinking, "Wait, I thought you were doing that", you've already felt the consequences of ignoring this. This brings me to the most boringly named, wildly underrated tool I know: the RACI Matrix. What the Heck is a RACI Matrix? Technically, RACI stands for Responsible, Accountable, Consulted, Informed. You've probably seen that somewhere before — maybe in a boring slide deck, maybe in a textbook you ignored. Responsible Accountable Consulted Informed It's like a contract that the project team makes with itself to ensure that every task has its rightful owner and everyone knows their part in the grand scheme of things. But here's what it means in real life. real life Imagine your project is a band. The Responsible folks are the ones actually playing the instruments. They're the ones up late, fixing bugs, writing queries, designing mockups — whatever it takes. Without them, there's no music. The Responsible folks are the ones actually playing the instruments. They're the ones up late, fixing bugs, writing queries, designing mockups — whatever it takes. Without them, there's no music. Responsible The Accountable one is the producer. They're not playing every instrument, but they're making sure the record actually ships. They own the outcome. They take the blame if it sucks — or the credit if it slaps. The Accountable one is the producer. They're not playing every instrument, but they're making sure the record actually ships. They own the outcome. They take the blame if it sucks — or the credit if it slaps. Accountable The Consulted are like the studio musicians or trusted friends who give you honest feedback during the recording. You want their input. You value their ears. But they're not in the band. The Consulted are like the studio musicians or trusted friends who give you honest feedback during the recording. You want their input. You value their ears. But they're not in the band. Consulted And the Informed? That's your audience, your label, maybe your investors. They need to know what's going on, but they're not getting involved in the creative process. And the Informed? That's your audience, your label, maybe your investors. They need to know what's going on, but they're not getting involved in the creative process. Informed And yes, one person or group can take on multiple roles within the project. The RACI matrix is typically represented as a table with tasks or outcomes listed in the left column and individual people or groups involved in the process listed on top. The intersection of each task and person or group indicates their corresponding RACI role. But Isn't This Just More Process? Yeah, I've heard this one before. Usually from senior ICs or managers who are allergic to anything that looks like structure. "Do we really need a matrix? Can't we just talk it through?" Sure. In a perfect world. On a team of three. For a one-week sprint. But real projects are messy. Real teams are overloaded. And real work happens asynchronously, across time zones, with shifting priorities and surprise stakeholder escalations. "Just talk it through" doesn't cut it when Bob is OOO, Sarah's on three projects, and Marketing is asking for a timeline update you didn't even know existed. The beauty of RACI is that it makes who does what explicit — before shit hits the fan. Not through another tool. Not through process theater. Just through one honest conversation about roles. And yeah, sometimes that conversation is uncomfortable. It surfaces power dynamics. It forces decisions. It makes people ask, "Wait, why am I even on this project?" Good. Because nothing kills a project faster than pretending everyone's equally involved when they're not. RACI doesn't create complexity — it exposes the complexity that's already there. You're Not Needed Here One of the underrated powers of RACI is that it gives people permission not to care about things that aren't their job. And I mean that in the best possible way. We're so used to over-collaboration — to inviting everyone into every decision — that we forget: clarity isn't exclusion. It's respect. Telling someone "you're Informed, not Consulted" isn't shutting them out. It's protecting their time. It's like saying, "We'll let you know when there's something worth reacting to — but you don't need to be in this meeting." Making a RACI Matrix That Doesn't Suck So, how do you actually build one that doesn't feel like a waste of everyone's time? Start by dumping out all the stuff your team is supposed to do. Not just the big-ticket items like "launch feature X", but the actual messy middle: writing specs, reviewing designs, testing, RFCs, whatever. Then, write down the name of every person who touches the project — even if it's just once. Yes, that one stakeholder who only shows up at the end with Opinions counts. Now comes the good part: assigning who's doing what. Not in vague terms like "we'll figure it out" or "team effort." You're putting names in boxes. Responsible is the person doing the work. Keyboard. Fingers. Doing the thing.
Accountable is the one on the hook if it doesn't happen. Usually, the person who gets yelled at.
Consulted is the person who needs to weigh in before anything goes out. Think feedback, not sign-off.
Informed is the person who just needs a heads-up. No input is required. Responsible is the person doing the work. Keyboard. Fingers. Doing the thing. Responsible Accountable is the one on the hook if it doesn't happen. Usually, the person who gets yelled at. Accountable Consulted is the person who needs to weigh in before anything goes out. Think feedback, not sign-off. Consulted Informed is the person who just needs a heads-up. No input is required. Informed The conversations you'll have while building the RACI are more important than the actual matrix itself. Once it's done? Share it. Talk it through. Make sure everyone's actually okay with their roles — not just nodding because they want the meeting to end. RACI Matrix Checklist Vertical Check (Roles): Too Much "R"? Ensure one person isn't overloaded with "Responsible" tasks. Distribute the workload evenly to avoid bottlenecks.
No Empty Cells: Evaluate if each stakeholder needs to be involved in so many activities.
Too Many "A"s? Reduce the number of decision-makers and accountability holders if possible.
No "R" or "A"? Assess the necessity of the role and consider redistributing responsibilities.
Double A/R or A/I? Stick to one "Accountable" role per stage. Too Much "R"? Ensure one person isn't overloaded with "Responsible" tasks. Distribute the workload evenly to avoid bottlenecks. Too Much "R"? No Empty Cells: Evaluate if each stakeholder needs to be involved in so many activities. No Empty Cells: Too Many "A"s? Reduce the number of decision-makers and accountability holders if possible. Too Many "A"s? No "R" or "A"? Assess the necessity of the role and consider redistributing responsibilities. No "R" or "A"? Double A/R or A/I? Stick to one "Accountable" role per stage. Double A/R or A/I? Horizontal Check (Project Stages): No "R"? Assign a "Responsible" stakeholder for each stage.
Too Much "R"? Divide complex stages into specific tasks to distribute responsibilities.
No "A"? Ensure there's accountability for each stage.
More Than One "A"? Stick to one "Accountable" stakeholder per stage.
All Cells Filled? Verify that all stages have stakeholder involvement.
No C/I? Ensure all relevant stakeholders are consulted and informed.
Too Much "C"? Limit excessive consultations; consider keeping stakeholders informed instead.
Too Much "I"? Evaluate if the information shared is truly beneficial or creates unnecessary meetings.
Double Positions? Avoid having one person take both the R and A roles, as it may lead to misunderstandings. No "R"? Assign a "Responsible" stakeholder for each stage. No "R"? Too Much "R"? Divide complex stages into specific tasks to distribute responsibilities. Too Much "R"? No "A"? Ensure there's accountability for each stage. No "A"? More Than One "A"? Stick to one "Accountable" stakeholder per stage. More Than One "A"? All Cells Filled? Verify that all stages have stakeholder involvement. All Cells Filled? No C/I? Ensure all relevant stakeholders are consulted and informed. No C/I? Too Much "C"? Limit excessive consultations; consider keeping stakeholders informed instead. Too Much "C"? Too Much "I"? Evaluate if the information shared is truly beneficial or creates unnecessary meetings. Too Much "I"? Double Positions? Avoid having one person take both the R and A roles, as it may lead to misunderstandings. Double Positions? Make sure all stakeholders agree to their assigned roles and document it on the project page. Wrapping It Up Of course, RACI isn't perfect. Like any tool, it can be misused. You'll see teams where everyone gets labeled "Consulted" just to keep them happy — which defeats the point. Or where people game the system to avoid being "Accountable" — because they don't want the heat. Or worse, people fill out a RACI matrix once, paste it in the project doc, and never look at it again. That's not RACI. That's checkbox theater. The matrix is only as good as your willingness to have the hard conversations behind it. And revisit them when things change. Which they will. Because projects evolve. People rotate out. Priorities shift. If your RACI doesn't shift with them, congratulations — you've got a beautiful artifact that doesn't reflect reality. So yeah, it takes work. But it's the kind of work that pays off before the deadline, not after. Thank you for reading! Thank you for reading! Curious about something or have thoughts to share? Leave your comment below! Check out my blog or follow me via LinkedIn, Substack, or Telegram. Curious about something or have thoughts to share? Leave your comment below! Check out my blog or follow me via LinkedIn, Substack, or Telegram. my blog LinkedIn Substack Telegram

Mastering Project Clarity: How The Power of the RACI Matrix Can Improve Your Teamwork

Welcome to MTP Business Boost — a series of expert publications filled with real-world strategies, transforming complex ideas into simple, actionable advice you can apply immediately (MTP stands for Marketing, Technology, and Psychology). In just 9 minutes, you’ll uncover how AI isn’t just adapting marketing communication for different generations—it’s giving businesses the power to dominate. By leveraging AI-driven personalization and precision, you can craft messages that resonate deeply, drive higher engagement, maximize conversions, and secure long-term customer loyalty. The result? Reduced costs, a stronger competitive edge, and undeniable market leadership. Personalization isn’t just a trend—it’s an expectation. A McKinsey study found that 71% of consumers expect personalized interactions, and 76% feel frustrated when companies fail to deliver them. This frustration often leads to lost sales and weakened brand loyalty. On the flip side, companies that embrace personalization can see revenue increases of 5-15% and a boost in marketing ROI by 10-30%. Without it, businesses risk inefficiencies in customer acquisition and retention, ultimately falling behind their competitors (source). source source Before you dive in, make sure you’re subscribed so you don’t miss the next expert insights packed with practical recommendations. Let’s get started! Weekend on a Yacht and Unexpected Consequences Weekend on a Yacht and Unexpected Consequences The golden hues of the Sydney sunrise painted the glass towers in warm amber as Alex leaned back in his sleek office chair, staring at the analytics report on his screen. His IT startup, a promising SaaS company specializing in cybersecurity solutions, had been steadily growing. But now, the data revealed an unexpected challenge—his customer base spanned multiple generations, each responding differently to marketing messages. What worked for Gen Z seemed irrelevant to Boomers, and Millennials had their own distinct preferences. Frustration crept in. How could he fine-tune his communication without hiring a team of specialists or overhauling his entire strategy? The budget was tight, the stakes were high, and he couldn’t afford to misstep. Just then, his phone buzzed. It was Daniel, an old friend and fellow entrepreneur, but leagues ahead in the game—seasoned, successful, and always three steps ahead. "Mate, your voice sounds tense. You need to unwind. Come spend the weekend on my yacht. A little time away will do you good." Alex hesitated. Leaving work for an entire weekend? Now? But something about Daniel’s tone made him reconsider. And so, with a mix of reluctance and curiosity, Alex accepted. The Sydney Harbour shimmered under the midday sun, a dazzling contrast to the deep blue waters. As Alex stepped onto Daniel’s sleek, state-of-the-art yacht—a floating masterpiece of luxury—he felt an unexpected sense of release. Polished teak, crisp ocean air, and the gentle lapping of waves made the city’s worries feel miles away. Over gourmet seafood and a chilled bottle of vintage Chardonnay, Alex shared his predicament. Daniel listened, nodding knowingly. "You’re making it harder than it needs to be," he finally said. "AI can handle this in minutes. It can analyze generational preferences, tailor messaging to their values and interests, and ensure each group receives communication that truly resonates." Alex raised an eyebrow. AI? Sure, his company used machine learning for cybersecurity, but for marketing? Daniel smirked, pulled out his tablet, and with a few taps, demonstrated how AI could segment audiences, adapt messaging for different generations, and align content with their unique expectations and behaviors. The next morning, back in his office, Alex hesitated for only a second before trying it himself. Within minutes, AI had analyzed his customer segments, optimized messaging for each generation, and provided actionable insights. It was effortless. It was powerful. And it worked. As he leaned back, a satisfied grin spread across his face. This wasn’t just a solution—it was a revelation. For the first time in weeks, he felt in control, ahead of the curve. And to think, it all started with a simple phone call and a weekend on a yacht. Winning Every Generation: AI-Powered Messaging Guide Let’s take an example and see how to adapt a marketing message for different generations. We’ll use a sample ad from a hypothetical SaaS company offering a sales automation product. Here’s the original marketing message: Boost your sales like never before with the ultimate SaaS solution for automating your pipeline! Say goodbye to manual follow-ups and hello to smart workflows that save you time and close deals faster. With AI-powered insights, you’ll know exactly when to reach out to prospects for maximum impact. Try it today and watch your revenue soar in just weeks. Sign up for a free 14-day trial – no credit card required! I recommend taking your own marketing message, following the steps below, and applying them right away. This way, you’ll put the knowledge into practice and see results quickly. Go to chatgpt.com Go to chatgpt.com chatgpt.com 2. Paste the following instruction for the AI and press Enter: Hello. You are a world-class expert in effective marketing communications. I need help adapting the following advertising message for optimal reception by my audience, which consists of the Baby Boomer generation. Hello. You are a world-class expert in effective marketing communications. I need help adapting the following advertising message for optimal reception by my audience, which consists of the Baby Boomer generation. Boost your sales like never before with the ultimate SaaS solution for automating your pipeline! Say goodbye to manual follow-ups and hello to smart workflows that save you time and close deals faster. With AI-powered insights, you’ll know exactly when to reach out to prospects for maximum impact. Try it today and watch your revenue soar in just weeks. Sign up for a free 14-day trial – no credit card required! Boost your sales like never before with the ultimate SaaS solution for automating your pipeline! Say goodbye to manual follow-ups and hello to smart workflows that save you time and close deals faster. With AI-powered insights, you’ll know exactly when to reach out to prospects for maximum impact. Try it today and watch your revenue soar in just weeks. Sign up for a free 14-day trial – no credit card required! Instead of “Baby Boomer,” you can swap in the generation your customers belong to: The Silent Generation (born 1928–1945)


Gen X (born 1965–1980)


Millennials, Gen Y (born 1981–1996)


Zoomers, Gen Z (born 1997–2012)


Gen Alpha (born 2013–2025) The Silent Generation (born 1928–1945) The Silent Generation (born 1928–1945) Gen X (born 1965–1980) Gen X (born 1965–1980) Millennials, Gen Y (born 1981–1996) Millennials, Gen Y (born 1981–1996) Zoomers, Gen Z (born 1997–2012) Zoomers, Gen Z (born 1997–2012) Gen Alpha (born 2013–2025) Gen Alpha (born 2013–2025) Here’s how the AI adapted the message for the Baby Boomer generation: 3. You can also ask the AI to explain why the adapted message works for that generation. Use this instruction: Write a detailed argument explaining why this advertising text is suitable for the generation of my clients. Write a detailed argument explaining why this advertising text is suitable for the generation of my clients. Here’s how the AI explained it in our example: 4. Now, let’s ask the AI to adapt the message for another generation, like Gen Y (born 1981–1996): Now, adapt this advertising message for Generation Y and explain why it is effective specifically for this generation. Now, adapt this advertising message for Generation Y and explain why it is effective specifically for this generation. The AI will also explain why the message fits Gen Y. 5. You can also have the AI adapt the message for all generations at once to compare them: Adapt this advertising message for all generations and present the results in a table format Adapt this advertising message for all generations and present the results in a table format 6. Plus, you can get the table as an Excel file: Generate this table in Excel format Generate this table in Excel format Besides Excel, the AI can provide the results in CSV or PDF formats. 7. You can also ask the AI for specific tips on communicating with your customers’ generation. In our example, try this instruction: Write the most effective recommendations for communication with the Baby Boomer generation. My product is a sales automation service Write the most effective recommendations for communication with the Baby Boomer generation. My product is a sales automation service 8. To make your communication with clients as effective as possible and avoid missteps, ask the AI: What mistakes should be avoided in communication with the Baby Boomer generation and why? What mistakes should be avoided in communication with the Baby Boomer generation and why? 9. After getting a lot of information from the AI, you can ask it to summarize everything into a handy PDF: Summarize all the information in this dialogue and present it in PDF format Summarize all the information in this dialogue and present it in PDF format Here’s what the AI-generated PDF might include: Extra Tips With AI, you can adapt almost any type of marketing communication: emails, Google Ads, social media posts, sales proposals, landing page content, and more. If you have data on your customers’ ages, you can group them by generation and make your communication even more personal and targeted. For example, if your email list has 100,000 subscribers—30,000 from Gen X and 70,000 from Gen Y—a single generic email might not work as well as two versions tailored to each generation. Now you know how to do this with AI, and it only takes a few minutes! Test the results to see the impact of adapting your marketing for specific generations. For instance, when sending an email, create two versions and check which one gets better conversions. Summary In this article, you’ve learned how to easily and quickly adapt your communication with clients based on the psychological perception of the generation they belong to. If the knowledge you gained from this article has been valuable and helpful to you, you can show your appreciation to the author by making a donation via this link or giving a like and sharing it. this link this link Would you also like to learn more about how AI can be used to grow your business and increase your market share? Feel free to leave your questions in the comments! About the Author Gene Misiev is a marketing expert with over seven years of experience in digital marketing. He thrives in highly competitive "red ocean" B2B markets. His focus is on growth marketing, market research, and strategic development for a complex SaaS product—a tender platform for government and private procurement. He has completed advanced training in Strategic Marketing and Generative AI applications in business at the Institute of Applied Technology (Australia). He has also undertaken specialized courses in leveraging ChatGPT for business and marketing automation. See You in the Next Issue Thanks for reading! Make sure you’re subscribed so you don’t miss the next publication, which will bring even more insights to help you expand your business empire with the power of marketing, technology and psychology.

Want to Boost Your Sales? AI Knows Exactly How to Speak to Every Generation

Obsidian is one of the best note-taking apps available today. It provides a powerful, Markdown-based experience with local-first storage. However, there is one problem: the official sync feature costs around $8 per month. What if I told you there’s a way to sync your notes across multiple devices completely free? In this guide, I will walk you through a method using GitHub and Git that allows you to keep your notes in sync without spending a dime. What Would Need It could feel that you need to do a lot of things, but don’t worry; in an ideal scenario, you would need about 10–15 minutes and these things: GitHub Account & Repository
GitHub Access Token
SSH key (optional)
Git
Obsidian
Git Plugin for Obsidian
iSH app for iPhone
Obsidian App for iPhone Step 1: Create a GitHub Account and Repository GitHub is a cloud-based platform primarily used for software development, but it can also be used for managing personal projects and files — including Obsidian notes. A Git repository (or repo) is a storage space where Git tracks all the changes to a set of files. It records modifications, allowing you to revert to previous versions, collaborate with others, and synchronize your files across different devices. In the context of Obsidian, a Git repository helps store and sync your notes while keeping track of all edits. Go to GitHub.com, and sign up.
Once logged in, click on the New button to create a new repository.
Give it a name (e.g., “Obsidian-Notes”).
Make sure to set the repository to Private so your notes are not publicly accessible.
Click Create Repository. Step 2: Install Git on Your Computer If you don’t have Git installed, follow these steps: Windows: Download and install Git from git-scm.com. Mac: Install Git using Homebrew with brew install git. Linux: Use sudo apt-get install git (for Debian-based systems) or sudo dnf install git (for Fedora-based systems). Once installed, open your terminal (Command Prompt, PowerShell, or macOS Terminal) and verify installation by running: git --version Basic Git Commands Here are three essential Git commands that you’ll use frequently: git status This command shows the current state of your repository. It tells you which files have been modified, added, or staged for commit.git status git pull This command fetches the latest changes from the remote repository (GitHub) and updates your local repository. git push After making changes, you need to upload them to GitHub using git push. This command sends your committed changes from your local repository to the remote repository. Step 3: Clone the GitHub Repository Now, let’s connect your local Obsidian vault to GitHub: Open your terminal, and navigate to the folder where you want to store your notes. Run the following command, replacing YOUR-REPO-URL with your GitHub repository URL: git clone YOUR-REPO-URL This will create a local folder linked to your GitHub repository. Move your Obsidian notes into this folder so they are ready for syncing. Step 3. How to Get a GitHub Classic Token GitHub has deprecated password-based authentication for Git operations. Instead, it requires you to use a Personal Access Token (PAT), which provides a more secure way to authenticate. How to Get a GitHub Classic Token Go to GitHub Developer Settings: Open GitHub Token Settings.
Click Generate new token → Select Classic. Set Expiration & Permissions: Choose an expiration date or set it to No Expiration (not recommended for security). Select the necessary scopes: repo → For accessing private repositories. Generate and Copy the Token: Click Generate token and copy it immediately.
GitHub won’t show it again after you leave the page. Use the Token in Git Authentication: When prompted for a password in Git operations, paste the token instead. Step 4: Set Up SSH for Authentication (Optional) To avoid entering your password every time you sync, you can set up SSH authentication: Generate an SSH key by running: ssh-keygen -t ed25519 -C "your-email@example.com" Copy the SSH key using: cat ~/.ssh/id_ed25519.pub Go to GitHub, navigate to Settings > SSH and GPG keys, and add the copied key. Now, your system will authenticate with GitHub automatically. Step 5: Set Up Git Plugin in Obsidian I assume you already have the Obsidian App, which is why I won’t cover the installation process for it here. I will only show you the quick steps to install the Git plugin. Obsidian has a plugin that makes Git syncing easier: Open Obsidian, and go to Settings > Community Plugins.
Search for “Git,” and install it.
Enable Auto Commit and Sync (set an interval, e.g., 5 minutes).
Enable Pull on Startup to prevent conflicts. Now, whenever you edit notes, Obsidian will sync them automatically with GitHub. Step 6: Syncing Notes on Mobile (iOS, iPhone, iPad) Syncing on mobile is slightly more complicated but still doable. Install Obsidian from the App Store. Install iSH, a terminal app that allows you to run Linux commands. iSH App


Open iSH, and install Git using: apk add git Create a folder for your obsidian notes: mkdir obsidian Run the mount command to mount the obsidian vault folder. mount -t ios . obsidian A file picker will show up. Choose the folder with your local vault. Then use the following commands: cd obsidianrm -rf .git clone YOUR-REPO-URL . Once this step is done, you will see your notes in the Obsidian application. Step 7: Install Obsidian Git Plugin on iPhone The last step in our tutorial - the Git community plugin. Open Obsidian.


Go to Settings > Community Plugins.


Tap Browse and search for Obsidian Git.


Tap Install, then Enable the plugin.


Set up an auto-commit interval (e.g., every 5 minutes).


Enable Pull on Startup to sync changes when opening Obsidian. Video Tutorial If you are struggling with the steps, I recommend you to watch my detailed video tutorial. https://youtu.be/PScdHzUiBLA?si=C5UNiV7Ou1BJhEbr&embedable=true Conclusion While it takes a bit of setup, once done, it works seamlessly. If you found this guide helpful, let me know in the comments, and feel free to ask any questions! Cheers! ;) Obsidian is one of the best note-taking apps available today. It provides a powerful, Markdown-based experience with local-first storage. However, there is one problem: the official sync feature costs around $8 per month. What if I told you there’s a way to sync your notes across multiple devices completely free? In this guide, I will walk you through a method using GitHub and Git that allows you to keep your notes in sync without spending a dime. Obsidian What Would Need It could feel that you need to do a lot of things, but don’t worry; in an ideal scenario, you would need about 10–15 minutes and these things: GitHub Account & Repository GitHub Access Token SSH key (optional) Git Obsidian Git Plugin for Obsidian iSH app for iPhone Obsidian App for iPhone GitHub Account & Repository GitHub Access Token SSH key (optional) Git Obsidian Git Plugin for Obsidian iSH app for iPhone Obsidian App for iPhone Step 1: Create a GitHub Account and Repository GitHub is a cloud-based platform primarily used for software development, but it can also be used for managing personal projects and files — including Obsidian notes. GitHub A Git repository (or repo) is a storage space where Git tracks all the changes to a set of files. It records modifications, allowing you to revert to previous versions, collaborate with others, and synchronize your files across different devices. In the context of Obsidian, a Git repository helps store and sync your notes while keeping track of all edits. Git repository Go to GitHub.com, and sign up. Once logged in, click on the New button to create a new repository. Give it a name (e.g., “Obsidian-Notes”). Make sure to set the repository to Private so your notes are not publicly accessible. Click Create Repository. Go to GitHub.com , and sign up. GitHub.com Once logged in, click on the New button to create a new repository. New Give it a name (e.g., “Obsidian-Notes”). Make sure to set the repository to Private so your notes are not publicly accessible. Private Click Create Repository . Create Repository Step 2: Install Git on Your Computer If you don’t have Git installed, follow these steps: Windows: Download and install Git from git-scm.com. Windows : Download and install Git from git-scm.com. Windows git-scm.com. Mac: Install Git using Homebrew with brew install git. Mac : Install Git using Homebrew with brew install git . Mac brew install git Linux: Use sudo apt-get install git (for Debian-based systems) or sudo dnf install git (for Fedora-based systems). Linux : Use sudo apt-get install git (for Debian-based systems) or sudo dnf install git (for Fedora-based systems). Linux sudo apt-get install git Once installed, open your terminal (Command Prompt, PowerShell, or macOS Terminal) and verify installation by running: git --version git --version Basic Git Commands Here are three essential Git commands that you’ll use frequently: git status git status This command shows the current state of your repository. It tells you which files have been modified, added, or staged for commit.git status git pull git pull This command fetches the latest changes from the remote repository (GitHub) and updates your local repository. git push git push After making changes, you need to upload them to GitHub using git push. This command sends your committed changes from your local repository to the remote repository. Step 3: Clone the GitHub Repository Now, let’s connect your local Obsidian vault to GitHub: Open your terminal, and navigate to the folder where you want to store your notes. Open your terminal, and navigate to the folder where you want to store your notes. Run the following command, replacing YOUR-REPO-URL with your GitHub repository URL: Run the following command, replacing YOUR-REPO-URL with your GitHub repository URL: git clone YOUR-REPO-URL git clone YOUR-REPO-URL This will create a local folder linked to your GitHub repository. This will create a local folder linked to your GitHub repository. Move your Obsidian notes into this folder so they are ready for syncing. Step 3. How to Get a GitHub Classic Token GitHub has deprecated password-based authentication for Git operations. Instead, it requires you to use a Personal Access Token (PAT) , which provides a more secure way to authenticate. Personal Access Token (PAT) How to Get a GitHub Classic Token Go to GitHub Developer Settings: Go to GitHub Developer Settings: Open GitHub Token Settings. Click Generate new token → Select Classic. Open GitHub Token Settings . GitHub Token Settings Click Generate new token → Select Classic . Generate new token Classic Set Expiration & Permissions: Set Expiration & Permissions: Choose an expiration date or set it to No Expiration (not recommended for security). Choose an expiration date or set it to No Expiration (not recommended for security). No Expiration Select the necessary scopes: repo → For accessing private repositories. repo → For accessing private repositories. Generate and Copy the Token: Generate and Copy the Token: Click Generate token and copy it immediately. GitHub won’t show it again after you leave the page. Click Generate token and copy it immediately . Generate token copy it immediately GitHub won’t show it again after you leave the page. Use the Token in Git Authentication: Use the Token in Git Authentication: When prompted for a password in Git operations, paste the token instead. When prompted for a password in Git operations, paste the token instead. password Step 4: Set Up SSH for Authentication (Optional) To avoid entering your password every time you sync, you can set up SSH authentication: Generate an SSH key by running: Generate an SSH key by running: ssh-keygen -t ed25519 -C "your-email@example.com" ssh-keygen -t ed25519 -C "your-email@example.com" Copy the SSH key using: Copy the SSH key using: cat ~/.ssh/id_ed25519.pub cat ~/.ssh/id_ed25519.pub Go to GitHub, navigate to Settings > SSH and GPG keys, and add the copied key. Go to GitHub, navigate to Settings > SSH and GPG keys , and add the copied key. Settings > SSH and GPG keys Now, your system will authenticate with GitHub automatically. Step 5: Set Up Git Plugin in Obsidian I assume you already have the Obsidian App, which is why I won’t cover the installation process for it here. I will only show you the quick steps to install the Git plugin. Obsidian has a plugin that makes Git syncing easier: Open Obsidian, and go to Settings > Community Plugins. Search for “Git,” and install it. Enable Auto Commit and Sync (set an interval, e.g., 5 minutes). Enable Pull on Startup to prevent conflicts. Open Obsidian, and go to Settings > Community Plugins . Settings > Community Plugins Search for “Git,” and install it. Enable Auto Commit and Sync (set an interval, e.g., 5 minutes). Auto Commit and Sync Enable Pull on Startup to prevent conflicts. Pull on Startup Now, whenever you edit notes, Obsidian will sync them automatically with GitHub. Step 6: Syncing Notes on Mobile (iOS, iPhone, iPad) Syncing on mobile is slightly more complicated but still doable. Install Obsidian from the App Store. Install Obsidian from the App Store. Obsidian Install iSH, a terminal app that allows you to run Linux commands. Install iSH , a terminal app that allows you to run Linux commands. iSH iSH App Open iSH, and install Git using: iSH App iSH App Open iSH, and install Git using: Open iSH, and install Git using: apk add git apk add git Create a folder for your obsidian notes: Create a folder for your obsidian notes: mkdir obsidian mkdir obsidian Run the mount command to mount the obsidian vault folder. Run the mount command to mount the obsidian vault folder. Run the mount command to mount the obsidian vault folder. mount -t ios . obsidian mount -t ios . obsidian A file picker will show up. Choose the folder with your local vault. A file picker will show up. Choose the folder with your local vault. Then use the following commands: Then use the following commands: cd obsidianrm -rf .git clone YOUR-REPO-URL . cd obsidianrm -rf .git clone YOUR-REPO-URL . Once this step is done, you will see your notes in the Obsidian application. Step 7: Install Obsidian Git Plugin on iPhone The last step in our tutorial - the Git community plugin. Open Obsidian. Go to Settings > Community Plugins. Tap Browse and search for Obsidian Git. Tap Install, then Enable the plugin. Set up an auto-commit interval (e.g., every 5 minutes). Enable Pull on Startup to sync changes when opening Obsidian. Open Obsidian. Open Obsidian. Go to Settings > Community Plugins. Go to Settings > Community Plugins. Tap Browse and search for Obsidian Git. Tap Browse and search for Obsidian Git. Tap Install, then Enable the plugin. Tap Install, then Enable the plugin. Set up an auto-commit interval (e.g., every 5 minutes). Set up an auto-commit interval (e.g., every 5 minutes). Enable Pull on Startup to sync changes when opening Obsidian. Enable Pull on Startup to sync changes when opening Obsidian. Video Tutorial If you are struggling with the steps, I recommend you to watch my detailed video tutorial. https://youtu.be/PScdHzUiBLA?si=C5UNiV7Ou1BJhEbr&embedable=true https://youtu.be/PScdHzUiBLA?si=C5UNiV7Ou1BJhEbr&embedable=true Conclusion While it takes a bit of setup, once done, it works seamlessly. If you found this guide helpful, let me know in the comments, and feel free to ask any questions! Cheers! ;)

How to Obsidian Sync for FREE! (No Paid Plan Required)

কিভাবে বিনামূল্যে অবসিডিয়ান সিঙ্ক করবেন! (কোনও পেইড প্ল্যান প্রয়োজন নেই)

Cómo sincronizar Obsidian GRATIS (no se requiere plan pago)

چگونه Obsidian Sync را به صورت رایگان انجام دهیم! (بدون نیاز به طرح پرداختی)

Obsidian Sync ILMAISEKSI! (Maksullista suunnitelmaa ei vaadita)

मुफ़्त में ओब्सीडियन सिंक कैसे करें! (कोई भुगतान योजना की आवश्यकता नहीं)

Obsidian Sync を無料で使う方法! (有料プランは不要)

Кантип Obsidian Sync АКЫСЫЗ! (Эч кандай акы төлөнүүчү план талап кылынбайт)

Kā sinhronizēt Obsidian BEZMAKSAS! (Nav nepieciešams apmaksāts plāns)

नि:शुल्क ओब्सिडियन सिङ्क कसरी गर्ने! (कुनै सशुल्क योजना आवश्यक छैन)

Paano Obsidian Sync nang LIBRE! (Walang Kinakailangang May Bayad na Plano)

Як БЕЗКОШТОВНО синхронізувати Obsidian! (Платний план не потрібен)

Qanday qilib Obsidian Sync-ni BEPUL qilish kerak! (Pulli reja shart emas)

In this post we'll continue working on link prediction with the Twitch dataset: we'll export the graph data from the Neptune DB cluster to an S3 bucket using the neptune-export utility provided by AWS. We'll choose the 'neptune_ml' profile when we create the data export task and the utility will create the 'training-data-configuration.json' file that we'll use later in the pipeline. The exported data will be ready for feature encoding and data processing, which is the next step required for link prediction. Read part 1 here. GRAPH DATA IN NEPTUNE DB We start with the graph data that we have in Neptune DB after uploading the lists of vertices and edges using the Neptune Bulk Loader API (as described in Part 1 of this guide). The vertices represent users. All vertices contain the same set of properties, and a single vertex looks like this: {<T.id: 1>: '153', <T.label: 4>: 'user', 'days': 1629, 'mature': True, 'views': 3615, 'partner': False} All edges have the same label ('follows'), each edge connects 2 users. A single edge looks like this: {<T.id: 1>: '0', <T.label: 4>: 'follows', <Direction.IN: 'IN'>: {<T.id: 1>: '255', <T.label: 4>: 'user'}, <Direction.OUT: 'OUT'>: {<T.id: 1>: '6194', <T.label: 4>: 'user'}} Our goal is to export the data so that it can be used in the next part of our data pipeline: preprocessing and feature encoding. RUNNING THE NEPTUNE-EXPORT UTILITY ON EC2 We'll use the neptune-export utility provided by AWS to export data from the database. To allow the utility access to the DB, we'll run it on an EC2 instance inside the VPC where the Neptune DB cluster is. The utility will get the data from the DB, save it in local storage (an EBS volume), and then it will upload the exported data to S3. Although AWS provides a Cloudformation template that deploys a private API inside your VPC to allow the export process to be started with an HTTP request, we won't focus on that this time. As our goal is to demonstrate how the data pipeline works (and not to set up an API), we'll just use the EC2 instance's console to interact with the neptune-export utility. By the way, those console commands can be automated with AWS Systems Manager Run Command and Step Functions. Let's create the EC2 instance that we'll run neptune-export on. For AMI, we choose Ubuntu 24.04 LTS. We need to make sure that the Neptune cluster is reachable from the EC2 instance, so we'll create the instance in the same VPC where the Neptune cluster is, and we'll configure the security groups to allow network traffic between the instance and the cluster. We also need to attach an EBS volume of sufficient size to contain the exported data. For the dataset that we're working on, an 8GB volume is enough. While the instance is starting, we need to create an IAM role that allows write access to the destination S3 bucket, and also some RDS actions, as it is shown in the policy below. While the first statement of the policy is mandatory, the second one is only needed if you export data from a cloned cluster. Exporting data from cloned clusters will be be discussed later in this post. {
	"Version": "2012-10-17",
	"Statement": [
		{
			"Sid": "RequiredPart",
			"Effect": "Allow",
			"Action": [
				"rds:ListTagsForResource",
				"rds:DescribeDBInstances",
				"rds:DescribeDBClusters"
			],
			"Resource": "*"
		},
		{
			"Sid": "OptionalPartOnlyRequiredForExportingFromClonedCluster",
			"Effect": "Allow",
			"Action": [
                "rds:AddTagsToResource",
                "rds:DescribeDBClusters",
                "rds:DescribeDBInstances",
                "rds:ListTagsForResource",
                "rds:DescribeDBClusterParameters",
                "rds:DescribeDBParameters",
                "rds:ModifyDBParameterGroup",
                "rds:ModifyDBClusterParameterGroup",
                "rds:RestoreDBClusterToPointInTime",
                "rds:DeleteDBInstance",
                "rds:DeleteDBClusterParameterGroup",
                "rds:DeleteDBParameterGroup",
                "rds:DeleteDBCluster",
                "rds:CreateDBInstance",
                "rds:CreateDBClusterParameterGroup",
                "rds:CreateDBParameterGroup"
            ],
			"Resource": "*"
		}
	]
} You can allow access to just the target cluster (instead of all clusters) by editing the ‘Resource’ field. The role must also have a trust policy that allows EC2 to assume the role: {
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Service": "ec2.amazonaws.com"
            },
            "Action": "sts:AssumeRole"
        }
    ]
} Once the EC2 instance and the role is ready, we'll attach the role to the instance. Next, we need to install the neptune-export utility on the instance. To do that, we’ll log into the instance and use these commands to install JDK 8 and download the utility: sudo apt update -y
sudo apt install -y openjdk-8-jdk
curl -O https://s3.amazonaws.com/aws-neptune-customer-samples/neptune-export/bin/neptune-export.jar Now that we've prepared the EC2 instance, the destination S3 bucket, and attached the IAM the role that allows write access to the destination S3 bucket to the instance, we can start exporting data. We'll use this command to initiate the process, providing the required parameters as a JSON object: java -jar /home/ubuntu/neptune-export.jar nesvc \
  --root-path /home/ubuntu/neptune-export \
  --json '{
    "command": "export-pg",
    "outputS3Path" : "s3://YOUR_TARGET_S3_BUCKET/neptune-export",
    "params": {
       "endpoint" : "YOUR_CLUSTER_ENDPOINT",
       "profile": "neptune_ml"
    }
  }' We used only the required parameters here but you can easily extend the config. You can choose what part of the graph you export using the 'filter' parameter: you can select nodes, edges, and their properties. If you're exporting data from a live database, you can use the 'cloneCluster' and 'cloneClusterReplicaCount' parameters to make the neptue-export utility take a snapshot of the database, create a new Neptune cluster from that snapshot, deploy read replicas, and use them to export the data. By doing that, you can make sure the live database is not affected by the additional load from the data exporting. The full list of parameters can be found here (https://docs.aws.amazon.com/neptune/latest/userguide/export-parameters.html). VIEWING EXPORTED DATA AND NEXT STEPS When the export process is completed, neptune-export prints some stats including the numbers of vertices and edges: Source:
  Nodes: 7126
  Edges: 70648
Export:
  Nodes: 7126
  Edges: 70648
  Properties: 28504
Details:
  Nodes: 
    user: 7126
        |_ days {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Integer:7126]}
        |_ mature {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Boolean:7126]}
        |_ views {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Integer:7126]}
        |_ partner {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Boolean:7126]}
  Edges: 
    (user)-follows-(user): 70648 And then it uploads the exported data to S3. Let's look at the files that are created in the target S3 bucket: The ‘nodes’ and ‘edges’ directories contain CSV files with the lists of nodes and edges that are similar to what we used in Part 1 when we uploaded data. For large graphs, there are multiple files, but our dataset is small and there's just one file in each directory. There's also the training-data-configuration.json file that we'll edit and use in the next step of our process. If you're doing a one time export, now it's safe to delete the EC2 instance and the EBS volume, since only the files in the target S3 bucket will be used in the next step. Otherwise, you can just stop the EC2 instance to avoid getting charged for idle time (you'll still be charged for EBS storage unless you delete it). At this point we have the graph data in S3 in the format that can be used in the next step of the process, and we're ready to do feature encoding and data processing, which will be discussed in our next post. In this post we'll continue working on link prediction with the Twitch dataset: we'll export the graph data from the Neptune DB cluster to an S3 bucket using the neptune-export utility provided by AWS. We'll choose the 'neptune_ml' profile when we create the data export task and the utility will create the 'training-data-configuration.json' file that we'll use later in the pipeline. The exported data will be ready for feature encoding and data processing, which is the next step required for link prediction. Read part 1 here. Read part 1 here . here GRAPH DATA IN NEPTUNE DB We start with the graph data that we have in Neptune DB after uploading the lists of vertices and edges using the Neptune Bulk Loader API (as described in Part 1 of this guide). The vertices represent users. All vertices contain the same set of properties, and a single vertex looks like this: {<T.id: 1>: '153', <T.label: 4>: 'user', 'days': 1629, 'mature': True, 'views': 3615, 'partner': False} {<T.id: 1>: '153', <T.label: 4>: 'user', 'days': 1629, 'mature': True, 'views': 3615, 'partner': False} All edges have the same label ('follows'), each edge connects 2 users. A single edge looks like this: {<T.id: 1>: '0', <T.label: 4>: 'follows', <Direction.IN: 'IN'>: {<T.id: 1>: '255', <T.label: 4>: 'user'}, <Direction.OUT: 'OUT'>: {<T.id: 1>: '6194', <T.label: 4>: 'user'}} {<T.id: 1>: '0', <T.label: 4>: 'follows', <Direction.IN: 'IN'>: {<T.id: 1>: '255', <T.label: 4>: 'user'}, <Direction.OUT: 'OUT'>: {<T.id: 1>: '6194', <T.label: 4>: 'user'}} Our goal is to export the data so that it can be used in the next part of our data pipeline: preprocessing and feature encoding. RUNNING THE NEPTUNE-EXPORT UTILITY ON EC2 We'll use the neptune-export utility provided by AWS to export data from the database. To allow the utility access to the DB, we'll run it on an EC2 instance inside the VPC where the Neptune DB cluster is. The utility will get the data from the DB, save it in local storage (an EBS volume), and then it will upload the exported data to S3. neptune-export Although AWS provides a Cloudformation template that deploys a private API inside your VPC to allow the export process to be started with an HTTP request, we won't focus on that this time. As our goal is to demonstrate how the data pipeline works (and not to set up an API), we'll just use the EC2 instance's console to interact with the neptune-export utility. By the way, those console commands can be automated with AWS Systems Manager Run Command and Step Functions. Let's create the EC2 instance that we'll run neptune-export on. For AMI, we choose Ubuntu 24.04 LTS. We need to make sure that the Neptune cluster is reachable from the EC2 instance, so we'll create the instance in the same VPC where the Neptune cluster is, and we'll configure the security groups to allow network traffic between the instance and the cluster. We also need to attach an EBS volume of sufficient size to contain the exported data. For the dataset that we're working on, an 8GB volume is enough. While the instance is starting, we need to create an IAM role that allows write access to the destination S3 bucket, and also some RDS actions, as it is shown in the policy below. While the first statement of the policy is mandatory, the second one is only needed if you export data from a cloned cluster. Exporting data from cloned clusters will be be discussed later in this post. {
	"Version": "2012-10-17",
	"Statement": [
		{
			"Sid": "RequiredPart",
			"Effect": "Allow",
			"Action": [
				"rds:ListTagsForResource",
				"rds:DescribeDBInstances",
				"rds:DescribeDBClusters"
			],
			"Resource": "*"
		},
		{
			"Sid": "OptionalPartOnlyRequiredForExportingFromClonedCluster",
			"Effect": "Allow",
			"Action": [
                "rds:AddTagsToResource",
                "rds:DescribeDBClusters",
                "rds:DescribeDBInstances",
                "rds:ListTagsForResource",
                "rds:DescribeDBClusterParameters",
                "rds:DescribeDBParameters",
                "rds:ModifyDBParameterGroup",
                "rds:ModifyDBClusterParameterGroup",
                "rds:RestoreDBClusterToPointInTime",
                "rds:DeleteDBInstance",
                "rds:DeleteDBClusterParameterGroup",
                "rds:DeleteDBParameterGroup",
                "rds:DeleteDBCluster",
                "rds:CreateDBInstance",
                "rds:CreateDBClusterParameterGroup",
                "rds:CreateDBParameterGroup"
            ],
			"Resource": "*"
		}
	]
} {
	"Version": "2012-10-17",
	"Statement": [
		{
			"Sid": "RequiredPart",
			"Effect": "Allow",
			"Action": [
				"rds:ListTagsForResource",
				"rds:DescribeDBInstances",
				"rds:DescribeDBClusters"
			],
			"Resource": "*"
		},
		{
			"Sid": "OptionalPartOnlyRequiredForExportingFromClonedCluster",
			"Effect": "Allow",
			"Action": [
                "rds:AddTagsToResource",
                "rds:DescribeDBClusters",
                "rds:DescribeDBInstances",
                "rds:ListTagsForResource",
                "rds:DescribeDBClusterParameters",
                "rds:DescribeDBParameters",
                "rds:ModifyDBParameterGroup",
                "rds:ModifyDBClusterParameterGroup",
                "rds:RestoreDBClusterToPointInTime",
                "rds:DeleteDBInstance",
                "rds:DeleteDBClusterParameterGroup",
                "rds:DeleteDBParameterGroup",
                "rds:DeleteDBCluster",
                "rds:CreateDBInstance",
                "rds:CreateDBClusterParameterGroup",
                "rds:CreateDBParameterGroup"
            ],
			"Resource": "*"
		}
	]
} You can allow access to just the target cluster (instead of all clusters) by editing the ‘Resource’ field. The role must also have a trust policy that allows EC2 to assume the role: {
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Service": "ec2.amazonaws.com"
            },
            "Action": "sts:AssumeRole"
        }
    ]
} {
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Service": "ec2.amazonaws.com"
            },
            "Action": "sts:AssumeRole"
        }
    ]
} Once the EC2 instance and the role is ready, we'll attach the role to the instance. Next, we need to install the neptune-export utility on the instance. To do that, we’ll log into the instance and use these commands to install JDK 8 and download the utility: sudo apt update -y
sudo apt install -y openjdk-8-jdk
curl -O https://s3.amazonaws.com/aws-neptune-customer-samples/neptune-export/bin/neptune-export.jar sudo apt update -y
sudo apt install -y openjdk-8-jdk
curl -O https://s3.amazonaws.com/aws-neptune-customer-samples/neptune-export/bin/neptune-export.jar Now that we've prepared the EC2 instance, the destination S3 bucket, and attached the IAM the role that allows write access to the destination S3 bucket to the instance, we can start exporting data. We'll use this command to initiate the process, providing the required parameters as a JSON object: java -jar /home/ubuntu/neptune-export.jar nesvc \
  --root-path /home/ubuntu/neptune-export \
  --json '{
    "command": "export-pg",
    "outputS3Path" : "s3://YOUR_TARGET_S3_BUCKET/neptune-export",
    "params": {
       "endpoint" : "YOUR_CLUSTER_ENDPOINT",
       "profile": "neptune_ml"
    }
  }' java -jar /home/ubuntu/neptune-export.jar nesvc \
  --root-path /home/ubuntu/neptune-export \
  --json '{
    "command": "export-pg",
    "outputS3Path" : "s3://YOUR_TARGET_S3_BUCKET/neptune-export",
    "params": {
       "endpoint" : "YOUR_CLUSTER_ENDPOINT",
       "profile": "neptune_ml"
    }
  }' We used only the required parameters here but you can easily extend the config. You can choose what part of the graph you export using the 'filter' parameter: you can select nodes, edges, and their properties. If you're exporting data from a live database, you can use the ' cloneCluster ' and ' cloneClusterReplicaCount ' parameters to make the neptue-export utility take a snapshot of the database, create a new Neptune cluster from that snapshot, deploy read replicas, and use them to export the data. By doing that, you can make sure the live database is not affected by the additional load from the data exporting. cloneCluster cloneClusterReplicaCount The full list of parameters can be found here ( https://docs.aws.amazon.com/neptune/latest/userguide/export-parameters.html ). https://docs.aws.amazon.com/neptune/latest/userguide/export-parameters.html VIEWING EXPORTED DATA AND NEXT STEPS When the export process is completed, neptune-export prints some stats including the numbers of vertices and edges: Source:
  Nodes: 7126
  Edges: 70648
Export:
  Nodes: 7126
  Edges: 70648
  Properties: 28504
Details:
  Nodes: 
    user: 7126
        |_ days {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Integer:7126]}
        |_ mature {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Boolean:7126]}
        |_ views {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Integer:7126]}
        |_ partner {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Boolean:7126]}
  Edges: 
    (user)-follows-(user): 70648 Source:
  Nodes: 7126
  Edges: 70648
Export:
  Nodes: 7126
  Edges: 70648
  Properties: 28504
Details:
  Nodes: 
    user: 7126
        |_ days {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Integer:7126]}
        |_ mature {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Boolean:7126]}
        |_ views {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Integer:7126]}
        |_ partner {propertyCount=7126, minCardinality=1, maxCardinality=1, recordCount=7126, dataTypeCounts=[Boolean:7126]}
  Edges: 
    (user)-follows-(user): 70648 And then it uploads the exported data to S3. Let's look at the files that are created in the target S3 bucket: The ‘nodes’ and ‘edges’ directories contain CSV files with the lists of nodes and edges that are similar to what we used in Part 1 when we uploaded data. For large graphs, there are multiple files, but our dataset is small and there's just one file in each directory. There's also the training-data-configuration.json file that we'll edit and use in the next step of our process. training-data-configuration.json If you're doing a one time export, now it's safe to delete the EC2 instance and the EBS volume, since only the files in the target S3 bucket will be used in the next step. Otherwise, you can just stop the EC2 instance to avoid getting charged for idle time (you'll still be charged for EBS storage unless you delete it). At this point we have the graph data in S3 in the format that can be used in the next step of the process, and we're ready to do feature encoding and data processing, which will be discussed in our next post. At this point we have the graph data in S3 in the format that can be used in the next step of the process, and we're ready to do feature encoding and data processing,

Before AI Predicts Your Next Friend, It Needs to Do This First

Janïr AI ukax Jutïr Amigomar yatiykasaxa, Nayraqatax Uk lurañaw wakisi

Antes de que la IA prediga quién será tu próximo amigo, primero debe hacer esto

Avant que l'IA ne prédise votre prochain ami, elle doit d'abord faire ceci

Նախքան արհեստական ինտելեկտը կանխատեսում է ձեր հաջորդ ընկերոջը, նախ պետք է դա անի

AIがあなたの次の友達を予測する前に、まずこれが必要です

Pirms mākslīgais intelekts paredz jūsu nākamo draugu, tas vispirms ir jādara

Пред вештачката интелигенција да го предвиди вашиот следен пријател, прво треба да го стори тоа

Mbere yuko AI Ihanura Inshuti Yawe Itaha, Irakeneye Kubanza Kubikora

Përpara se AI të parashikojë mikun tuaj të ardhshëm, duhet ta bëjë këtë së pari

AI indiki dostuňyzy çaklamazdan ozal muny etmeli

Loko AI Yi nga si Vula Munghana Wa Wena La landzelaka, Yi Fanele Ku Rhanga Ku Endla Leswi

اس سے پہلے کہ AI آپ کے اگلے دوست کی پیشین گوئی کرے، اسے پہلے یہ کرنے کی ضرورت ہے۔

AI Diagnosed Her With Breast Cancer - So She Decied to See How Accurate It Was

Time is the one resource that money can’t buy, which is why I strongly believe in the transformative potential of AI in healthtech. This market is undeniably one of the most challenging: data protection laws, stringent regulations, and bureaucratic hurdles create significant barriers. Yet, these challenges are precisely what make innovation here so impactful. A good friend of mine, Alex Badalyan, is working on something extraordinary—think of it as the Spotify of healthcare. While I can’t reveal too much just yet, what I can say is that this platform is already backed by some of the biggest names in the industry. Alex’s proven track record in building scalable services, combined with the expertise of Anton Shchipanov, ex-Revolut, the startup’s CTO, makes this venture one to watch. Today, I’m speaking with Anton about his journey, his decision to join this stealth-mode healthtech startup, and his insights into the market’s future. Mary Glazkova: Anton, thank you for joining me today. Let’s dive right in—why did you decide to join this new AI healthtech startup? Anton Shchipanov: Thank you, Mary. It’s a pleasure to be here. My decision to join this venture comes from both personal and professional alignment. In my career, I’ve worked on building and scaling data-driven, AI-powered systems. At one of the top music streaming platforms Yandex Music, I helped develop recommendation algorithms and real-time systems that processed vast amounts of user data. This experience taught me how to harness data to deliver personalized user experiences, a concept that translates nicely to tailored health insights. Later at Revolut, I managed adaptable systems capable of rapid customization—a vital feature for personalizing recommendations based on an individual’s unique needs. When Alex approached me with this opportunity, it felt like the perfect chance to apply these skills to a sector that needs innovation urgently. Preventive health, in particular, can shift the focus of healthcare from reactive treatments to proactive solutions. Mary: From your perspective, what’s currently happening in the healthtech market, and how is AI shaping it? Anton: The healthtech market is growthing, for sure. In 2025, the global healthtech market is projected to surpass $500 billion, driven largely by advancements in AI and data analytics. AI is revolutionizing the field by enabling precision medicine, which personalizes healthcare based on individual data such as genetics, lifestyle, and medical history. For example, wearable devices and electronic health records are being analyzed with AI to predict health risks and recommend preventive actions, improving patient outcomes while streamlining healthcare costs. However, the market faces significant challenges—data privacy concerns, fragmented systems, and the complexities of navigating regulatory frameworks. Startups that can address these issues while delivering tangible value will define the future of healthtech. Our startup focuses on integrating AI with preventive healthcare, bridging the gap between cutting-edge technology and practical, user-centric solutions. Mary: I spent a year working in London back in 2018. From your experience, how has the tech landscape changed since then? Anton: The tech landscape has evolved dramatically since 2018. AI adoption has moved from experimental pilots to becoming a core business strategy across industries. In London, the growth of healthtech and fintech has been especially pronounced. The pandemic accelerated the adoption of digital tools, particularly in areas like telemedicine and remote health monitoring. Startups are not just using AI to improve products but are also leveraging it for operational efficiency and scalability. The ecosystem is now more collaborative, with a stronger focus on sustainability and long-term impact, reflecting a broader global trend. Mary: You’ve had an impressive career spanning several industries and roles, and you’ve mentored startups. Could you share some highlights? Anton: Mentoring startups has been incredibly rewarding. Through programs like Solyanka’s Scalerator and EMERGE conferences, I’ve worked with startups such as DataHive and PayNet, helping them build scalable architectures and navigate technical challenges. Interestingly, mentoring is a two-way street. While guiding others, I’ve learned a great deal myself—especially about emerging trends and innovative approaches to problem-solving. For instance, at EMERGE, I noticed a growing interest in ESG principles and sustainable tech solutions, which aligns closely with my belief that technology must be responsible and impactful. One of the key lessons I emphasize to startups is the importance of a solid foundation. Rushing to scale without addressing the basics, like resilient architecture and robust DevOps practices, often leads to costly setbacks. Mary: Can you share a bit about your new venture? Anton: While I can’t go into detail yet, I can tell you this: our platform is designed to redefine preventive healthcare by combining AI-powered insights with a seamless user experience. We’re tackling some of the biggest challenges in the industry, and we’re doing it with a team that deeply understands both technology and user needs. The most important thing is that we’re giving users the ability to consolidate all their previously unstructured health data in one place and leverage it for the most personalized approach to preventive health. This means empowering individuals to take control of their health in a way that’s never been possible before. Mary: Fundraising in healthtech is no easy feat. How did you manage to secure funding? Anton: Fundraising is indeed challenging, especially in healthtech. But we approached it strategically. First, Alex leveraged his network to connect with investors who share our vision. Second, we conducted extensive customer interviews to validate the market need, ensuring that our MVP addressed real pain points. This gave us credibility and confidence during investor pitches. Third, we focused on building a strong team with a proven track record, which resonated well with investors. Mary: Any advice for CTOs, especially those in startups? Anton: Absolutely. My advice would be to prioritize scalability and adaptability from the outset. Build systems that can grow with your user base and evolve as the market changes. Secondly, don’t underestimate the value of collaboration—engage with other teams early to ensure alignment across product, engineering, and business goals. Lastly, stay curious. The tech landscape evolves rapidly, and staying ahead means constantly learning and experimenting. Mary: Thank you for sharing your journey and insights, Anton. I look forward to seeing the impact of your work in this exciting field. Anton: Thank you, Mary. It’s been a pleasure discussing these topics. The intersection of AI and healthtech holds immense potential, and I’m excited to be part of this journey. Time is the one resource that money can’t buy, which is why I strongly believe in the transformative potential of AI in healthtech. This market is undeniably one of the most challenging: data protection laws, stringent regulations, and bureaucratic hurdles create significant barriers. Yet, these challenges are precisely what make innovation here so impactful. A good friend of mine, Alex Badalyan, is working on something extraordinary— think of it as the Spotify of healthcare . While I can’t reveal too much just yet, what I can say is that this platform is already backed by some of the biggest names in the industry. Alex’s proven track record in building scalable services, combined with the expertise of Anton Shchipanov , ex-Revolut , the startup’s CTO , makes this venture one to watch. think of it as the Spotify of healthcare think of it as the Spotify of healthcare Anton Shchipanov ex-Revolut the startup’s CTO Today, I’m speaking with Anton about his journey, his decision to join this stealth-mode healthtech startup, and his insights into the market’s future. Mary Glazkova: Anton, thank you for joining me today. Let’s dive right in—why did you decide to join this new AI healthtech startup? Mary Glazkova: Anton Shchipanov: Thank you, Mary. It’s a pleasure to be here. My decision to join this venture comes from both personal and professional alignment. Anton Shchipanov: In my career, I’ve worked on building and scaling data-driven, AI-powered systems. At one of the top music streaming platforms Yandex Music, I helped develop recommendation algorithms and real-time systems that processed vast amounts of user data. This experience taught me how to harness data to deliver personalized user experiences, a concept that translates nicely to tailored health insights. Later at Revolut, I managed adaptable systems capable of rapid customization—a vital feature for personalizing recommendations based on an individual’s unique needs. When Alex approached me with this opportunity, it felt like the perfect chance to apply these skills to a sector that needs innovation urgently. Preventive health, in particular, can shift the focus of healthcare from reactive treatments to proactive solutions. Mary: From your perspective, what’s currently happening in the healthtech market, and how is AI shaping it? Mary: Anton: The healthtech market is growthing, for sure. In 2025, the global healthtech market is projected to surpass $500 billion, driven largely by advancements in AI and data analytics . Anton: market is projected to surpass $500 billion, driven largely by advancements in AI and data analytics AI is revolutionizing the field by enabling precision medicine, which personalizes healthcare based on individual data such as genetics, lifestyle, and medical history. For example, wearable devices and electronic health records are being analyzed with AI to predict health risks and recommend preventive actions, improving patient outcomes while streamlining healthcare costs. AI is revolutionizing the field by enabling precision medicine, which personalizes healthcare based on individual data such as genetics, lifestyle, and medical history. For example, wearable devices and electronic health records are being analyzed with AI to predict health risks and recommend preventive actions, improving patient outcomes while streamlining healthcare costs. However, the market faces significant challenges— data privacy concerns, fragmented systems, and the complexities of navigating regulatory frameworks . Startups that can address these issues while delivering tangible value will define the future of healthtech. data privacy concerns, fragmented systems, and the complexities of navigating regulatory frameworks Our startup focuses on integrating AI with preventive healthcare, bridging the gap between cutting-edge technology and practical, user-centric solutions. Mary: I spent a year working in London back in 2018. From your experience, how has the tech landscape changed since then? Mary: Anton: The tech landscape has evolved dramatically since 2018. AI adoption has moved from experimental pilots to becoming a core business strategy across industries. Anton: In London, the growth of healthtech and fintech has been especially pronounced. The pandemic accelerated the adoption of digital tools, particularly in areas like telemedicine and remote health monitoring . Startups are not just using AI to improve products but are also leveraging it for operational efficiency and scalability. The pandemic accelerated the adoption of digital tools, particularly in areas like telemedicine and remote health monitoring The ecosystem is now more collaborative, with a stronger focus on sustainability and long-term impact, reflecting a broader global trend. Mary: You’ve had an impressive career spanning several industries and roles, and you’ve mentored startups. Could you share some highlights? Mary: Anton: Mentoring startups has been incredibly rewarding. Through programs like Solyanka’s Scalerator and EMERGE conferences, I’ve worked with startups such as DataHive and PayNet, helping them build scalable architectures and navigate technical challenges. Anton: Interestingly, mentoring is a two-way street. While guiding others, I’ve learned a great deal myself—especially about emerging trends and innovative approaches to problem-solving. For instance, at EMERGE, I noticed a growing interest in ESG principles and sustainable tech solutions , which aligns closely with my belief that technology must be responsible and impactful. I noticed a growing interest in ESG principles and sustainable tech solutions One of the key lessons I emphasize to startups is the importance of a solid foundation. Rushing to scale without addressing the basics, like resilient architecture and robust DevOps practices, often leads to costly setbacks. One of the key lessons I emphasize to startups is the importance of a solid foundation. Rushing to scale without addressing the basics, like resilient architecture and robust DevOps practices, often leads to costly setbacks. Mary: Can you share a bit about your new venture? Mary: Anton : While I can’t go into detail yet, I can tell you this: our platform is designed to redefine preventive healthcare by combining AI-powered insights with a seamless user experience. We’re tackling some of the biggest challenges in the industry, and we’re doing it with a team that deeply understands both technology and user needs. Anton The most important thing is that we’re giving users the ability to consolidate all their previously unstructured health data in one place and leverage it for the most personalized approach to preventive health . This means empowering individuals to take control of their health in a way that’s never been possible before. the ability to consolidate all their previously unstructured health data in one place and leverage it for the most personalized approach to preventive health Mary: Fundraising in healthtech is no easy feat. How did you manage to secure funding? Mary: Fundraising in healthtech Anton: Fundraising is indeed challenging, especially in healthtech. But we approached it strategically. First, Alex leveraged his network to connect with investors who share our vision. Second, we conducted extensive customer interviews to validate the market need, ensuring that our MVP addressed real pain points. This gave us credibility and confidence during investor pitches. Third, we focused on building a strong team with a proven track record, which resonated well with investors. Anton: Mary: Any advice for CTOs, especially those in startups? Mary: Anton: Absolutely. My advice would be to prioritize scalability and adaptability from the outset. Build systems that can grow with your user base and evolve as the market changes. Anton: Absolutely. My advice would be to prioritize scalability and adaptability from the outset. Build systems that can grow with your user base and evolve as the market changes. Anton: Secondly, don’t underestimate the value of collaboration—engage with other teams early to ensure alignment across product, engineering, and business goals. Lastly, stay curious. The tech landscape evolves rapidly, and staying ahead means constantly learning and experimenting. Mary: Thank you for sharing your journey and insights, Anton. I look forward to seeing the impact of your work in this exciting field. Mary: Anton: Thank you, Mary. It’s been a pleasure discussing these topics. The intersection of AI and healthtech holds immense potential, and I’m excited to be part of this journey. Anton:

Exec Building the Spotify of Healthcare Says Only Startups That Address Real Issues Will Thrive 

DeepSeek Releases Cheapest Ever LLM In The World

Stories, News & Mentions about SUREIN

Categories

Trending Topics

Stories, News & Mentions about SUREIN

Surein’s Stories on HackerNoon

Categories

Trending Topics