<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Christos Anastasiades on Medium]]></title>
        <description><![CDATA[Stories by Christos Anastasiades on Medium]]></description>
        <link>https://medium.com/@anastasiades.chr?source=rss-4ec84cd7858------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Ti6dQ2oMSadI3-DdpfmzwQ.jpeg</url>
            <title>Stories by Christos Anastasiades on Medium</title>
            <link>https://medium.com/@anastasiades.chr?source=rss-4ec84cd7858------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 11:21:27 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@anastasiades.chr/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Building a digital democratic platform designed for civic matters]]></title>
            <link>https://uxdesign.cc/building-a-digital-democratic-platform-designed-for-civic-matters-107966f6dda6?source=rss-4ec84cd7858------2</link>
            <guid isPermaLink="false">https://medium.com/p/107966f6dda6</guid>
            <category><![CDATA[civictech]]></category>
            <category><![CDATA[algorithms]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[politics]]></category>
            <dc:creator><![CDATA[Christos Anastasiades]]></dc:creator>
            <pubDate>Tue, 18 Nov 2025 23:20:17 GMT</pubDate>
            <atom:updated>2025-11-18T23:20:17.922Z</atom:updated>
            <content:encoded><![CDATA[<h4>A great time to innovate democracy, while the technology is in place.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/812/1*Z60XLFCjRC8ahtlyGAydCw.png" /><figcaption>Wireframe illustrating how comments could be rated for clarity, insight, and inclusiveness in real time</figcaption></figure><p>About twenty years ago, just after turning eighteen, I tried to follow politics because I cared about living in a fair society. I wanted everyone to have food, healthcare, safety, education — the basics any decent person wishes for others. But politics felt toxic: shouting, accusations, and no real progress. Even the word began to feel heavy.</p><p>I know I’m not alone; many people feel frustrated or disconnected. Yet politics should be noble — collaboration, empathy, coexistence.</p><p>So I’m writing this to explore how technology and digital design might help us rebuild that connection and make democracy feel human again.</p><h3>The bad side of politics and technology</h3><p>It’s almost impossible to separate politics from technology today. Every major shift in communication has reshaped how we organise, argue, and imagine our future. New tools always begin as freedom — more voices, more access, more possibility — but over time they start shaping us in ways we barely notice. Social media felt like a new democracy at first, until algorithms pushed us into <strong>echo chambers, doom scrolling </strong>and trained us to react instead of <strong>reflect</strong>. Technology can amplify <strong>empathy or anger</strong>, truth or distortion. It depends on <strong>how we design it.</strong> If we build tools that <strong>reward curiosity over conflict,</strong> politics could start to feel human again.</p><p><strong>Propaganda</strong> isn’t new, but technology has made it <strong>faster</strong>, more <strong>precise</strong>, and far more <strong>personal</strong>. In the past, newspapers and radio spread broad, one-size-fits-all messages. Today, algorithms whisper tailored narratives to each of us. The Cambridge Analytica scandal highlighted how personal data can influence political persuasion, but similar patterns have also been observed in Europe. Ahead of the 2024 EU elections, researchers uncovered coordinated disinformation networks operating across France, Germany, and Italy, using fake accounts to quietly influence voter perception. This is the new propaganda: subtle, targeted, and often unnoticed — not designed to inform us, but to keep us engaged and predictable.</p><p>If we want more harmonious communities, we must rebuild <strong>respect</strong>, <strong>trust</strong>, and <strong>understanding</strong>. This isn’t just about better apps — it’s about better values. A healthy digital political system rests on a few core principles: a <strong>thirst for truth</strong> over partisanship; <strong>transparency by default</strong> in decisions and algorithms; <strong>effectiveness</strong> in solving real problems; <strong>equal access to voice</strong> regardless of power; and <strong>truth before speed</strong>, prioritising accuracy over virality. It also requires <strong>agility</strong>, allowing nuanced positions; <strong>holism</strong>, showing how issues connect; <strong>simplicity</strong>, making participation approachable; and <strong>decentralised ownership</strong> to prevent concentrated control. These values must guide the technology we design next.</p><h4>Designing a Political Platform That Supports Equilibrium</h4><p>A fictional glimpse of this future appears in <a href="https://turnerstories.medium.com/the-democr-app-96895f45b680">N. A. Turner’s short story <em>“The Democr App.”</em></a> In it, society runs on constant digital referendums, with citizens voting directly on everything from climate policy to economic reform. Turner captures the tension at the heart of radical digital democracy through one of his characters:</p><blockquote>Mouse87 still wondered if the majority could decide what is true and just for the entire nation. Could the people govern themselves? He believed they could. He had to. It was part of his design… to unite people, to let democracy shine in its truest form. But he made one mistake. Mouse87 wanted people to become equal, but no two persons are the same, let alone a whole nation.</blockquote><p>This fictional moment reveals a real-world challenge: if technology is going to serve democracy, it must be built for <strong>equilibrium, not engagement</strong> — helping people understand before they react, listen before they judge, and collaborate before they divide.</p><p>Imagine a platform where political discussion doesn’t lead to outrage or exhaustion, but to clarity. Instead of a chaotic feed, citizens move through a guided process that adds context, shows diverse perspectives, and explains how each decision touches real life. Every issue is presented holistically: the problem, public feelings, solutions, supporters, opposers, risks, and consequences. The platform wouldn’t punish disagreement; it would encourage <strong>informed disagreement</strong>.</p><p>The biggest risk to be addressed is that people are emotional thinkers. We rely on shortcuts, familiar ideas, and content that feels good. That makes building a political platform for understanding — not reaction — uniquely challenging. Echo chambers can quietly return. Misinformation spreads fast. Power can be centralised. People get overwhelmed or tune out. Privacy must be protected. Algorithms can inherit cultural bias. And cynicism can make citizens feel their participation doesn’t matter. These risks aren’t technical — they’re human. The goal isn’t perfection but <strong>transparency, safety, and adaptability</strong>, designing for how people actually think while gently helping them grow toward deeper, fairer public understanding.</p><figure><img alt="Cleisthenes the founder of Athenian democracy" src="https://cdn-images-1.medium.com/max/320/1*KG9RVtDI285nG1Vwz8De-w.gif" /><figcaption>Cleisthenes the founder of Athenian democracy (created with Sora)</figcaption></figure><h3>Proposal</h3><p>I propose to build a digital democratic platform designed for civic matters. A space where citizens explore issues through context, clarity, and diverse perspectives. The proposed solution is inspired by a collection of technologies and methodologies that already exist in the market.</p><h4>Modernising the democratic process</h4><p>Modern democracies could learn from how product teams build great systems: through continuous cycles of learning, testing, and improvement. In tech, decisions aren’t final — they’re iterated. After research, prototyping, and alignment, products launch, are monitored, refined, and improved again.</p><p>Our political systems rarely work this way. Voting is treated as the end of the process, not the beginning of a learning loop. Imagine instead a democratic cycle that mirrors product development: <strong>collect facts</strong>, <strong>collect people’s feelings</strong>, <strong>ideate</strong>, <strong>vote</strong>, <strong>enforce and monitor</strong>, then <strong>learn and iterate</strong>. Decisions would become living processes, shaped by data, experience, and citizen feedback.</p><figure><img alt="Modern Democratic Process contains 5 steps that repeat in a loop" src="https://cdn-images-1.medium.com/max/1024/1*Vg8TQB8kmhEWPoHg3_Ifrg.png" /><figcaption>Modern Democratic Process</figcaption></figure><p>This concept is not only mine to share; the minister of the digital affairs of Taiwan, Audrey Tang, explained with some major wins in his announcement.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FESW6D59eFuw%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DESW6D59eFuw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FESW6D59eFuw%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/378431ed5045273f32d50ae23e346c34/href">https://medium.com/media/378431ed5045273f32d50ae23e346c34/href</a></iframe><h4>Modernising democratic roles</h4><p>Modernising democracy also means modernising our <strong>roles</strong> within it. A healthier system depends on shared participation, not hierarchy</p><ul><li><strong>Citizens</strong> — Stay curious, seek holistic truth, and engage respectfully with the community. Understanding comes before opinion.</li><li><strong>Representatives &amp; Experts</strong> — Study issues deeply, gather reliable evidence, and communicate clearly. Their role is to <strong>guide</strong>, not rule.</li><li><strong>Journalists &amp; Communicators</strong> — Go beyond reporting outcomes. Explain processes, challenge assumptions, and make complex information accessible and accurate.</li><li><strong>Moderators &amp; Facilitators</strong> — Ensure discussions remain constructive, inclusive, and fact-based. Conversations need care, not control</li></ul><h4>Getting inspired by existing tools</h4><p>This proposal isn’t built from scratch; it stands on the shoulders of platforms already pushing civic understanding forward. Across the world, tools from deliberation to fact-checking show how technology can reduce bias, reveal context, and support healthier public dialogue. Below are the key inspirations:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/287411905cf22ca0843b2f5dbdd50386/href">https://medium.com/media/287411905cf22ca0843b2f5dbdd50386/href</a></iframe><h3>Platform Features</h3><h4>Civilian Contributions</h4><p>This is the heart of the platform — the <strong>civilian contributions</strong>, or what we usually call <em>posts</em>. Every user can contribute something meaningful: an idea, a problem, an opinion, a question, or even an opportunity for improvement. Contributions can be connected to existing topics or open entirely new ones.</p><p>But here, it’s not just <em>what</em> you say that matters — it’s <em>how</em> you say it.<br>The platform’s goal is to <strong>guide users toward expressing ideas in a fair, respectful, and inclusive way</strong>, helping people frame their thoughts with clarity and empathy. The tone should promote understanding, not hostility.</p><p>When creating a post, the system gently assists the user in identifying <strong>who might be affected</strong> — both positively and negatively — and what <strong>risks or consequences</strong> the idea might carry. This helps transform a simple complaint or reaction into a <strong>civic contribution</strong> — a statement grounded in awareness and responsibility.</p><p>Example:</p><ul><li><strong>Initial Post:</strong><br> <em>“It was about time for an accident. Our roads suck!”</em></li><li><strong>Guided Contribution:</strong><br> <em>“After the accident last week on James Bond Street, I’ve noticed how poor visibility is around that area. I completely understand the need to illuminate archaeological monuments beautifully, but perhaps we can adjust the lighting design to make the road safer for drivers and pedestrians.”</em></li></ul><h4>Engaging With Contributions</h4><p>Once a post is shared, the conversation truly begins. Engagement on this platform isn’t about collecting likes — it’s about <strong>building knowledge together</strong>. Users can interact with a post by adding <strong>further research</strong>, <strong>supporting documents</strong>, <strong>related ideas</strong>, or <strong>personal experiences</strong> that enrich the discussion. Reactions are not noise; they are part of the learning process. Every contribution adds a layer of understanding to the original idea.</p><p>As discussions grow, the system detects when a post gains meaningful traction — not through volume, but through depth, balance, and constructive participation.</p><p>For example, imagine the earlier post about the <strong>James Bond Street accident</strong>. As others contribute — a local architect sharing a visibility report, a cyclist uploading a nighttime video, a city planner linking to safety regulations — the post evolves into a valuable public thread. The system may then elevate it to the top of the <strong>#Transportation</strong> section, where local authorities, journalists, or experts can see it easily.</p><p>Contributing to engagements becomes a <strong>path to collaboration</strong>, connecting citizens, experts, and decision-makers around the same living issue.</p><h4>Feedback, Evaluation, and Reporting Layer</h4><p>Every contribution can be evaluated by the community through a <strong>structured feedback model</strong>. Instead of quick reactions, users can mark a post as <em>helpful</em>, <em>insightful</em>, <em>unclear</em>, or <em>biased</em>. They can also add short notes explaining why — offering specific, respectful observations rather than emotional responses.</p><p>Take the example of the <strong>James Bond Street accident post</strong>. As the discussion grows, others might respond like this:</p><ul><li><em>“Helpful — this identifies a recurring safety issue and offers context.”</em></li><li><em>“Unclear — it’s not clear who manages street lighting in that zone; can you clarify?”</em></li><li><em>“Insightful — linking the lighting to archaeological site preservation adds an interesting angle.”</em></li></ul><p>This structured feedback system <strong>evaluates quality</strong>, not <strong>popularity</strong>. Thoughtful, diverse responses elevate a post; unclear ones prompt clarification or community notes. Citizens can <strong>flag</strong> <strong>misinformation</strong> or harm, triggering contextual reviews rather than punishment. Over time, this<strong> two-way process</strong> teaches users to communicate better and helps the platform highlight substance over outrage.</p><h4>Enhanced Community Notes</h4><p>Community Notes has proven that <strong>context builds trust</strong>. When users add explanations instead of blunt “misleading” labels, people are more likely to believe the correction and share less misinformation. A study <a href="https://academic.oup.com/jcmc/article/29/2/taae006/7644305">published</a> by Oxford University Press found that contextual notes were <em>significantly more trusted</em> than standard misinformation flags.</p><p>A University of California, San Diego <a href="https://today.ucsd.edu/story/community-notes-accuracy-study">analysis</a> confirmed that Community Notes were <strong>highly accurate and credible</strong>, especially for vaccine-related posts. Similarly, research on <a href="https://arxiv.org/abs/2307.00916">arXiv reported </a>that when a note is attached to a misleading post, engagement drops sharply — reposts by ~45% and likes by ~43%.</p><p>Still, the model struggles with <strong>speed, diversity, and visibility</strong>.<br>Many viral posts never receive notes in time, and a <strong>small group of contributors</strong> writes the majority. This pattern was noted in a large-scale <a href="https://ddia.org/">data transparency report</a> and in follow-up analyses on Wikipedia.</p><p>Because the platform isn’t optimised for speed or emotional reactions, each post is first shared with a segment of users who have expertise in that topic. This allows knowledgeable contributors to add early context and Community Notes. If the post begins gaining traction, a <strong>community review mode</strong> activates, inviting a wider and more diverse group of users to contribute additional context before the content spreads further.</p><p>All notes appear directly within the post, with the system’s LLM providing clear, neutral summaries.</p><h4>Algorithm for Understanding</h4><p>Most digital platforms today are built around one metric — <em>engagement</em>. Their algorithms are designed to maximise <strong>emotion, outrage, and addictive scrolling</strong>, because the longer we stay online, the more ads we see, and the more profitable the system becomes.</p><p>Our platform must take the opposite path. Instead of optimising for emotion, it optimises for <strong>understanding</strong>. Instead of amplifying what divides, it surfaces what connects.</p><p>The algorithm’s purpose is to <strong>guide comprehension</strong>. It promotes content that helps people see the full picture, not just the side that fits their bias.</p><p><strong>Algorithm key principles:</strong></p><p><strong>1. Relevance First</strong><br> Content is served to the people who can meaningfully enrich it: citizens directly affected, experts with relevant insight, and communities equipped to help. This creates deeper engagement, allowing posts to gather context, sources, impacts, and risks — rather than shallow reactions or fragments.</p><p><strong>2. Exposure to Multiple Viewpoints</strong><br> The system deliberately broadens perspectives. It resurfaces skipped materials, introduces contrasting views, and reduces the blind spots that create echo chambers.</p><p><strong>3. Promote Thoughtful contributions</strong><br>Thoughtful contributions are highlighted above heated or repetitive ones, shifting the culture from reaction to reflection. This slows the rush toward instant opinions and encourages users to pause, explore context, and consider alternative viewpoints before responding.</p><h4>Join Discussion Rooms</h4><p>When a topic gains traction and enough information is gathered, it can evolve into a <strong>virtual discussion room</strong> — an open dialogue space where citizens, experts, and representatives come together to exchange ideas in real time.</p><p>Because these discussions are hosted within the platform, all previously gathered data helps <strong>structure the meeting</strong> through dedicated moderator tools. Insights generated during the session are then fed back into the platform, enriching the topic’s knowledge base and improving the ongoing civic process.</p><p>Each session is moderated by a <strong>certified facilitator</strong> from the platform’s community, whose role is to ensure inclusivity, respect, and focus. After the discussion ends, participants can rate the moderator’s fairness and clarity, creating a transparent feedback loop that strengthens future sessions.</p><h4>Open Source and Transparency</h4><p>At the foundation of this platform lies a commitment to <strong>transparency</strong>. Every major component — from the algorithm that ranks posts to the data that informs decisions — should be <strong>open source</strong> and publicly auditable. People must be able to see how information moves, how visibility is determined, and how their contributions are valued.</p><p>By keeping the code, funding sources, and moderation logic open, the platform earns trust through visibility, not promises. Just as citizens can participate in discussions, they can also participate in improving the system itself — submitting code, suggesting fixes, or analysing biases in real time.</p><p>Openness makes corruption harder and collaboration easier. It turns the system into a <strong>public good</strong>, owned and shaped by the very people it serves.</p><p>And importantly, as highlighted by the <a href="https://medium.com/@lyakova7?source=post_page---post_author_info--06def9faf439---------------------------------------">Association Redefine</a>, open source creates <strong>long-term resilience</strong>. Even if a project shuts down, communities can continue developing and maintaining the tools because the code belongs to everyone. Investments don’t vanish — they evolve into a living civic ecosystem that outlives any single platform</p><h3>Visualising Data for Clarity</h3><p>A civic platform can generate thousands of posts, threads, and interactions. To make this complexity understandable, we need simple, smart visualisations that reveal the shape of the conversation at a glance. These visuals help each person see where they stand, how the community thinks, and how decisions evolve.</p><p>We focus on <strong>five key visual tools</strong>:</p><ol><li><strong>Profile Map</strong> —helps you understand <em>how</em> you participate in the civic space.</li><li><strong>Press Landscape</strong> — compares how different media sources frame the same issue.</li><li><strong>Community Landscape</strong> — clusters related discussions to reveal priorities and tensions.</li><li><strong>Debate Pathways</strong> — maps the flow of arguments from context to risks.</li><li><strong>Voting Impact Map</strong> — shows how different votes influence outcomes and each other.</li></ol><h4><strong>Profile Map</strong></h4><p>Inspired by the philosophy behind <strong>Readocracy</strong>, it doesn’t measure popularity; it measures <strong>depth, diversity, and quality</strong> of your engagement.</p><p>Your Profile Map reflects things like:</p><ul><li>the topics you explore</li><li>the perspectives you’ve engaged with</li><li>the clarity and helpfulness of your contributions</li><li>your willingness to consider alternative views</li></ul><p>Over time, the Profile Map becomes a visual record of your learning journey — showing how your understanding evolves and where your strengths or blind spots may lie</p><figure><img alt="A dark dashboard showing “Profile Data” with sections on media bias, news ratio, mood, content tendencies, and professional interests. It includes a bias spectrum chart, percentages of news sources, emotional mood stats, topic categories, and a network graph connecting themes like technology, design, culture, and ethics." src="https://cdn-images-1.medium.com/max/1024/1*8byK2_CrRq2aOk8zAc8bKg.png" /><figcaption>Readocracy Personal Reading &amp; Content Patterns</figcaption></figure><h4>Press Landscape Map</h4><p>Inspired by <strong>AllSides</strong> and <strong>Improve the News</strong>, the Press Landscape Map shows how different media outlets frame the same issue. Instead of one narrative, you see a simple side-by-side comparison: what each source highlights, ignores, or emphasises emotionally.</p><p>A slider lets you move between left, centre, and right perspectives, revealing coverage gaps and bias patterns instantly. The goal isn’t to tell you what to believe, but to help you <strong>see the full information spectrum</strong> so you can form your own, well-rounded view.</p><figure><img alt="A media-analysis dashboard presenting several connected visuals. One area includes parallel text boxes that describe the same topic from contrasting perspectives — for example, one panel might emphasize personal freedoms while another highlights public safety concerns, illustrating how the same event can be framed in very different ways." src="https://cdn-images-1.medium.com/max/1024/1*qMIudegfRzaltAMpFsbdLg.png" /><figcaption>How Different Viewpoints Shape the Same Story (from verity.news)</figcaption></figure><h4>Community Landscape</h4><p>Inspired by <strong>Polis</strong>, the Community Landscape shows where people agree, where they diverge, and where new ideas are emerging. Instead of a chaotic comment feed, opinions are grouped into clusters, making patterns visible at a glance.</p><p>You can instantly see:</p><ul><li>areas of broad consensus</li><li>viewpoints gaining traction</li><li>minority positions that add nuance</li><li>topics that need more voices or clarity</li></ul><figure><img alt="A polling interface displays a statement about requiring passenger liability insurance for riders in private ride-share vehicles, with options to agree or disagree. Below, a map-like visualization shows clusters of user profile photos grouped by positions labeled with letters and participant counts. Horizontal bars above some clusters indicate the share of agreement (green) and disagreement (red)." src="https://cdn-images-1.medium.com/max/1024/1*K4tLH8W9kllwGrQePy0oPQ.png" /><figcaption><em>How People Cluster Around a Single Opinion (from Polis)</em></figcaption></figure><h3>Debate Pathways</h3><p>Inspired by <strong>Kialo</strong>, Debate Pathways turns discussions into <strong>clear, structured argument maps</strong>. Instead of scrolling through emotional, unstructured comments, you follow a simple, logical flow: Context, Problem, Proposed Solution, Supporting Arguments, Counterarguments, Risks and Trade-offs</p><p>Each point branches visually, so you can explore an idea from multiple angles without getting lost. This structure helps users see <em>why</em> people disagree, where the strongest reasoning sits, and what blind spots remain.</p><figure><img alt="A discussion interface shows a circular chart at the top with red and green segments indicating levels of disagreement and agreement on a statement about allowing 16-year-olds to vote. Below, user comments appear in stacked cards. One highlighted comment argues that 16-year-olds could be pressured into voting against their interests, with supporting points listed under “Pros” and opposing points under “Cons.”" src="https://cdn-images-1.medium.com/max/935/1*NLfSpYLEWNXi-XhsA7fxIg.png" /><figcaption>How a Community Debates Whether Teens Should Vote (from kialo)</figcaption></figure><h3>Voting Impact Map</h3><p>The Voting Impact Map shows how each contribution and engagement shapes real outcomes. Instead of pressing a button blindly, you can see the <strong>Immediate effects</strong> of each choice and <strong>who is positively or negatively affected. You can also spot how this decision connects to other issues and map Potential long-term consequences</strong></p><p>Visually, it works like a branching diagram: each voting option expands into predicted impacts, trade-offs, and ripple effects across related topics. This helps citizens understand not only <em>what</em> they are voting for, but <em>what it leads to</em>.</p><p>The goal is to make voting <strong>transparent, informed, and interconnected</strong> — turning every choice into a conscious, educated action rather than a guess in the dark.</p><figure><img alt="A voting interface displays a pie chart showing most participants selecting the supportive option, with a smaller portion registering an objection and none undecided. A table beside the chart lists percentages and the number of voters for each option, shown with small profile photos. Below, a comment from a participant explains their objection, expressing concern about fire safety provisions." src="https://cdn-images-1.medium.com/max/1024/1*oSylmba14pHnzqMy-PZOJw.png" /><figcaption><em>Visualizing group agreement, objections, and the reasoning behind them (from Loomio)</em></figcaption></figure><h3>Closing</h3><p>Over the past decade, we have watched technology reshape politics in ways both inspiring and alarming. Digital platforms have given millions of people a voice, but they have also enabled manipulation, polarisation, and propaganda at a scale that was once unimaginable. Elections, public opinion, and social movements are now influenced not just by policies and ideas, but by algorithms, recommendation feeds, and data-driven emotional triggers.</p><p>If democracy is to survive and evolve, it must adapt to this new reality. And to do that, we must ask a deeper question:</p><blockquote>What makes democracy strong in the first place?</blockquote><p>The answer is not just voting, nor simply participating. The answer is <strong>truth-seeking</strong>.</p><blockquote><strong><em>Democracy is only as strong as the willingness of people to go to the bottom of the “truth”.</em></strong></blockquote><p>If citizens don’t question assumptions, investigate claims, or look beyond the most entertaining narrative, democracy becomes vulnerable — not because people lack intelligence, but because <strong>human nature prefers comfort to complexity and there are not plenty of means to do so</strong>.</p><p>I am proposing a holistic platform that can support civilians to stay active, informed and respectful to their community.</p><p><strong>Thank you for reading!</strong><br>I’d love to connect and hear your views on how we can create the best possible experience for a platform like this. Your perspective matters. <a href="https://www.linkedin.com/in/christosanastasiades/">LinkedIn</a></p><p>And if you want to support more writing like this, <a href="http://buymeacoffee.com/chrana">feel free to buy me a coffee</a> ☕🙂</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=107966f6dda6" width="1" height="1" alt=""><hr><p><a href="https://uxdesign.cc/building-a-digital-democratic-platform-designed-for-civic-matters-107966f6dda6">Building a digital democratic platform designed for civic matters</a> was originally published in <a href="https://uxdesign.cc/">UX Collective</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How We Built an AI Ed-Tech Workflow That Turned Days of Work Into Seconds]]></title>
            <link>https://medium.com/design-bootcamp/how-we-built-an-ai-ed-tech-workflow-that-turned-days-of-work-into-seconds-0513f5526d75?source=rss-4ec84cd7858------2</link>
            <guid isPermaLink="false">https://medium.com/p/0513f5526d75</guid>
            <category><![CDATA[edtech]]></category>
            <category><![CDATA[ai-product-management]]></category>
            <category><![CDATA[n8n]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Christos Anastasiades]]></dc:creator>
            <pubDate>Tue, 21 Oct 2025 11:23:32 GMT</pubDate>
            <atom:updated>2025-10-21T11:23:32.793Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LOyDqJdUSHpfW7FGtBOU6A.png" /></figure><blockquote>The decision to use AI wasn’t the bold part — it was the only way the project could even start. The real bet was whether it could actually meet our quality bar and compress timelines enough to matter. Eight weeks later, what began as an experiment became a one-click workflow that turned days of work into seconds.</blockquote><h3>The Opportunity</h3><p>A new digital learning platform approached me (an individual AI Product contributor) with an ambitious goal: to launch a series of short, high-quality educational materials that could attract both <strong>new users</strong> and potential <strong>funding partners</strong>.</p><p>While the partner names remain confidential during the funding stage, the mission was clear — transform the university’s existing knowledge into consistent, <strong>publish-ready digital content</strong>.</p><p>The targets were ambitious: <strong>60 micro-courses in the first month</strong> to validate the concept, and <strong>500 courses within six months</strong> to build a competitive catalog.</p><p>Delivering at that scale meant there was no room for a traditional content-production approach. We needed a method that could merge academic depth with<strong> production efficiency </strong>— <strong>fast</strong>, <strong>reliable</strong>, and on <strong>budget</strong>. That’s where AI stopped being an experiment and became the foundation that made the <strong>entire project feasible</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dVwvp8TyOmkqGf2l9ObHlg.png" /></figure><h3>The Deliverable</h3><p>Each week, we committed to delivering one <strong>“thematic”</strong>, composed of <strong>3 bundles</strong>, each with <strong>5 courses and one quiz</strong>.</p><p>For example, a thematic titled Consumer Behaviour might include bundles such as Understanding the Consumer Journey, with courses like Behavioural Economics and Marketing Strategy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ESpB9irqSx3kD5x1pneJmg.png" /></figure><p>To make automation feasible, every element had to be precisely defined. Our core components were:</p><ul><li><strong>Course:</strong> A 9-minute, concise, high-value educational video — supported by slides, animations, and light sound effects when needed to emphasize key ideas.</li><li><strong>Voice-over &amp; Cue Document:</strong> One Google spreadsheet per course with three columns — (a) voice-over script, (b) visual cues, and (c) audio cues — used by the video production team.</li><li><strong>Detailed Document:</strong> A Google Doc describing course content, pricing, and related courses, also used to tag and organize material within the platform.</li><li><strong>Quiz:</strong> A short, five-question quiz per bundle to reinforce student learning.</li><li><strong>Organizational Files</strong>: The folder structure where all materials lived and tied everything together.</li></ul><p>Each week’s output included:</p><ul><li>15 videos</li><li>15 course description documents</li><li>15 voice-over scripts with cues</li><li>3 quizzes</li></ul><p>By keeping each deliverable modular and standardized, we built a system that could be automated end-to-end — making every new thematic as predictable as the last.</p><h3>The Team</h3><p>This project involved four main parties, each playing a distinctive role in bringing the content to life.</p><ul><li><strong>The University:</strong> Provided academic materials and subject expertise, while reviewing and proofreading the AI-generated drafts to ensure accuracy and tone.</li><li><strong>AI Product Manager (my role):</strong> Translated the university’s content into production-ready assets for the video team — defining workflows, building templates, and setting up the AI systems that automated repetitive tasks.</li><li><strong>Video Production Team:</strong> Turned scripts into concise 9-minute videos, using AI-assisted editing tools. Their feedback helped refine how scripts and cues were formatted for faster turnaround.</li><li><strong>The Digital Platform (Client):</strong> Managed final uploads and system tagging, coordinated feedback loops between teams, and ensured all materials met publishing standards.</li></ul><p>Together, this cross-functional setup made it possible to balance speed, accuracy, and creativity — ensuring each course met both production goals and academic standards.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mcsze3Y2xgj6XU0Qb_UXvA.png" /></figure><h3>Our Approach</h3><p>We structured the project<strong> like a product build </strong>— validating manually, scaling through automation, and continuously refining based on results.</p><p>The roadmap evolved through five key phases:</p><p><strong>1. Requirements Gathering</strong> — We aligned with the platform, video team, and university to define needs and output formats. During this stage, we collected all relevant university materials and clarified quality criteria for every deliverable.</p><p><strong>2. Manual AI Production (MVP)</strong> — Using ChatGPT, we manually created the first bundles to test prompt structures, refine templates, and understand how AI handled academic tone and content depth.</p><p><strong>3. Optimization</strong> — Through prompt engineering and the model’s adaptive memory, production time per bundle dropped from roughly <strong>1.5 days to about 1 hour</strong>, while output consistency improved significantly.</p><p><strong>4. Automation with n8n</strong> — We then built a “one-click” workflow that generated full course packages — including scripts, documents, and quizzes — automatically stored and organized in Google Drive. This turned the workflow into a repeatable system.</p><p><strong>5. Stabilization (In Progress)</strong> — The next phase focuses on strengthening reliability: adding error handling, automated alerts, and version tracking to make the system robust enough for continuous scaling.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zgL5W15X-23m_DUO6rgVUA.png" /></figure><p>Moving from phase one to full automation took approximately <strong>eight weeks</strong>, reducing thematic production time from <strong>hours to seconds</strong> — transforming what was once a bottleneck into a fully scalable process.</p><h3>Some Rules for Our Process</h3><p>I’ve seen how the right process and structure can transform delivery — and nowhere is that more true than in the AI era. When iteration happens in seconds, process design can’t just manage speed; it has to evolve with it. The faster things move, the more adaptable the structure behind them needs to be.</p><p>To keep our workflow scalable and aligned across teams, we established a few simple but powerful rules:</p><ul><li><strong>Embrace iteration:</strong> early outputs were imperfect by design — progress over perfection.</li><li><strong>Grace Period:</strong> The first few weeks were dedicated to testing and learning — intentionally allowing room for imperfection before production standards applied.</li><li><strong>Don’t Go Back:</strong> When something broke, we restarted the flow instead of patching it. It was faster to rebuild cleanly than to fix half-functional processes.</li><li><strong>Two-Day Feedback Loops:</strong> Every deliverable received feedback within 48 hours, ensuring no team was ever blocked or waiting.</li><li><strong>Weekly and Daily Milestones:</strong> Monday — creation, Tuesday — review, Wednesday — feedback, Thursday — iteration, Friday — delivery.</li><li><strong>Trust the Process:</strong> Early outputs weren’t perfect, but consistency and iteration proved far more valuable than chasing immediate polish.</li></ul><p>These principles created the rhythm that kept multiple teams in sync and ensured continuous progress — even when speed was the top priority</p><h3>Implementation</h3><h4>The Workflow Stack</h4><blockquote>This stack reflects the tools I used to design and automate the content creation workflow — from AI generation to structured delivery. It doesn’t include the video production tools used by the media team, but rather focuses on the systems that enabled my part of the process to run efficiently and at scale.</blockquote><p><strong>ChatGPT:</strong> After experimenting with other models, including Gemini, ChatGPT consistently produced higher-quality, more contextually accurate content. Gemini required more iterations to reach an acceptable tone and often drifted from the academic style we needed. ChatGPT handled longer prompts more reliably and integrated smoothly into our workflow, making it the clear choice for course generation and refinement.</p><p><strong>n8n:</strong> We needed a tool that could automate complex workflows visually and scale without heavy coding. <strong>n8n</strong> offered the flexibility to manage logic, data flow, and custom API connections as the project grew. We also tested Zapier, which works well for quick, simple automations, but at the time it didn’t support direct OpenAI integration. Even now, <strong>n8n</strong> provides greater adaptability for large-scale orchestration, making it the better long-term fit for our setup.</p><p><strong>Google Drive:</strong> All partners were already using it, and it worked smoothly with <strong>n8n</strong>, so it became the straightforward choice for collaboration and file management.</p><p><strong>OpenAI API:</strong> Powered the large-scale automation behind content generation. Integrating it with <strong>n8n</strong> allowed us to move beyond manually using the ChatGPT platform — generating complete course packages automatically, with consistent quality, structure, and formatting.</p><p>This simple ecosystem allowed us to move from idea to implementation quickly, with minimal overhead — balancing accessibility for non-technical collaborators with the scalability needed for automation.</p><h4>Execution</h4><p>Once the tools were connected, the workflow ran almost end-to-end. <strong>ChatGPT</strong> served as the content interface, feeding in the thematic glossary and university materials. Through custom GPT actions, it triggered the <strong>n8n</strong> automation workflow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w8GVU9_4tNTri8TDHGEqlw.png" /></figure><p>When <strong>n8n</strong> received the input, it automatically created a main folder with three bundle subfolders. Each folder contained all course documents, quizzes, and voice-over scripts — fully organized, named, and ready for the video production team.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LY6tcM9uILsUuyxW54E9OQ.png" /><figcaption>v0.8 — n8n workflow</figcaption></figure><p>Since the deliverables needed to be tightly controlled with minimal ambiguity, it was faster and more reliable to build <strong>Sequential LLM Chaining</strong> — feeding one model output directly into the next — rather than using autonomous <strong>Agents</strong>. Chaining provided greater consistency and traceability, ensuring each step followed a defined structure without unexpected deviations in tone or format.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*B0Yozy9bqvOTcyYIe9dYRw.gif" /></figure><p>This setup — our <strong>phase 4 workflow</strong> — marked the first version of a truly repeatable system. It wasn’t perfect, but it turned a multi-step manual process into a single automated flow. After submitting the input, the full workflow finished in about <strong>10 minutes</strong> — just enough time to brew a coffee.</p><p>Although it’s entirely possible to build such workflows with no coding experience (as in my case, using ChatGPT), a bit of scripting knowledge helps shorten the learning curve and makes refining or troubleshooting much easier.</p><h3>AI Economics</h3><p>Each course generation involved three <strong>GPT-4o</strong> calls, consuming roughly <strong>5,000 tokens</strong> per course. At an average cost of <strong>$0.06 per 1,000 output tokens</strong>, that came to about <strong>$0.30 per course</strong> — or <strong>$1.80 per bundle</strong> and <strong>$4.80 per full thematic set</strong>, including quizzes.</p><p>In each batch, about three courses typically failed QA and were manually recreated, but this rework remained minor since it occurred within the <strong>OpenAI</strong> platform’s fixed monthly usage tier. Building and testing the <strong>n8n</strong> workflows added around <strong>$50</strong> in setup costs.</p><p>From a <strong>unit economics</strong> perspective, the difference is significant. What once required professors or subject-matter experts days to curate could now be automated in hours — freeing resources for review and creative refinement instead of raw production. The result wasn’t just cost efficiency but also the freedom to <strong>experiment faster</strong> and validate ideas at scale.</p><p>As automation started to impact real budgets, these numbers became practical decision tools. They helped us see where automation delivered the most value and where keeping a human in the loop still made sense.</p><p>We also started thinking about the general <strong>sustainability of using LLMs</strong> for future projects — how to apply these systems responsibly and efficiently across new initiatives. A next step might be building a simple calculator to track automation ROI and process impact, translating efficiency gains into informed, sustainable growth.</p><h3>Learnings</h3><p>Looking back, a few clear lessons stood out from building and scaling this workflow:</p><ul><li><strong>Quality improves through interaction.</strong> ChatGPT’s memory and iteration cycles consistently raised output quality over time.</li><li><strong>Fresh chats can outperform long ones.</strong> Starting a new chat sometimes produced better results than continuing an old one — have you experienced the same? My assumption is that older threads can carry subtle biases from earlier context.</li><li><strong>Process and communication matter more than speed.</strong> In fast-moving, AI-assisted projects, well-defined feedback loops and clear ownership often make the difference between chaos and momentum.</li><li><strong>AI has real costs.</strong> Testing, iteration, and experimentation all consume tokens — and those costs should be included in every project budget, not treated as an afterthought.</li><li><strong>Automation must serve clear value.</strong> LLMs only make sense when they remove real bottlenecks or replace high-value, repetitive work. Automating for its own sake rarely pays off.</li><li><strong>LLMs are very bad at counting words.</strong> A simple prompt like “produce a 1,000-word document” doesn’t work reliably — extra implementation layers are needed to enforce length and structure.</li><li><strong>Use agents only when necessary.</strong> They can be powerful, but in structured production workflows, they often add more complexity than value.</li></ul><p>… Lastly,</p><p>One of the biggest takeaways was understanding <strong>prompt acceptance and process economics</strong> — how much it costs to fix an error manually versus automating its resolution. In this project, a strong prompt met acceptance criteria about <strong>80%</strong> of the time. Fixing the remaining 20% manually took roughly an hour, while adding a new process layer for automatic error handling would increase LLM costs. For our setup, that trade-off made sense, but in larger-scale products — where user LTV is lower and LLM cost per user higher — the equation might flip.</p><p>In the end, it all comes down to <strong>unit economics</strong>: balancing value, tokens, and time to decide where automation truly adds impact.</p><h3>Improvements</h3><p>Even though the workflow runs end-to-end, there’s still plenty of room to make it smarter, faster, and more reliable. The next steps focus on increasing quality, reducing maintenance effort, and improving system visibility — with technology serving as an enabler, not the goal.</p><ul><li><strong>Optimize Courses Through Real Feedback</strong><br>So far, prompt testing has focused on internal evaluation and quality assurance. The next step is to connect prompt performance to real-world learning outcomes. By testing courses with actual students — not just relying on university reviewers — we could measure engagement, comprehension, and quiz success rates to understand which content truly works. Combining this feedback with <strong>prompt versioning and A/B testing</strong> would make optimization data-driven and continuous.<br> <em>Example:</em> If two quiz-generation prompts produce similar academic results but one leads to higher student engagement or faster completion rates, that insight can directly guide future prompt tuning.</li><li><strong>Improve Quality Control</strong><br> As the number of generated courses grows, maintaining quality at scale becomes harder. An <strong>Evals pipeline</strong> could automatically assess drafts based on clarity, tone, and structure — highlighting those that need human review.<br> <em>Example:</em> If a course sounds too academic, the system could flag it for regeneration in simpler language before publishing.</li><li><strong>Increase Content Accuracy</strong><br> The main challenge with LLM-generated material is factual consistency. To strengthen accuracy, we could integrate a Retrieval Augmented Generation <strong>(RAG)</strong> layer that allows the model to reference verified university sources before writing.<br> <em>Example:</em> For a course on <em>Digital Ethics</em>, the system could automatically pull from the university’s ethics handbook, improving factual precision and reducing editing time by over 50%.</li><li><strong>Boost Reliability</strong><br> Failures in one workflow step can slow everything down. Implementing <strong>structured error handling and recovery</strong> would allow the automation to self-correct instead of stopping entirely.<br> <em>Example: </em>If a file upload fails due to a naming conflict or network issue, the system could automatically retry or rename the file to keep production running smoothly.</li><li><strong>Enhance System Awareness</strong><br> As the workflow scales, visibility becomes key. Adding <strong>real-time alerts </strong>and monitoring (via Slack or email) would allow quick action when something goes wrong.<br> <em>Example:</em> If a quiz fails to generate, the system could instantly notify the content lead so only that module is rerun, not the entire process.</li><li><strong>Make Decisions Data-Driven</strong><br> To guide further optimization, a <strong>performance dashboard</strong> could track token usage, generation times, and success rates across themes.<br> <em>Example:</em> If one topic consistently takes longer to produce, the data could highlight inefficiencies in the prompt or source material.</li></ul><p>Together, these improvements would shift the workflow from a functional automation into an <strong>intelligent, adaptive system</strong> — one that not only produces faster but continuously learns where to improve next.</p><h3>Bottleneck</h3><p>The main bottleneck for this project turned out to be as expected, the <strong>video creation</strong>. Even with automation in content and scripting, video production timelines remained a<strong> step behind</strong>. There are plenty of promising tools out there improving daily and we are hopeful that we will crack this.</p><p>I am currently exploring self-produced videos to understand whether individual creators can accelerate the process — possibly leading into our next project: “Inspires the Kids”, an educational show concept born from these experiments.</p><h3>Closing Thoughts</h3><p><strong>Beyond the hype,</strong> when used with purpose and structure, AI tools become true enablers, helping projects emerge that might<strong> never have been possible</strong> before. A new kind of endeavor is rising: one where a single person or a small team can do what once took entire organizations.</p><p>This new era of AI has its own rules. To succeed, a few things need to be in place: clarity of <strong>process</strong>, <strong>adaptability</strong>, and awareness of <em>AI </em><strong><em>economics</em></strong> — understanding when automation truly adds value and when human oversight still matters. It’s easy to use a hammer to crack a nut; the challenge is <strong>designing systems where technology amplifies, not overwhelms.</strong></p><p>Times are changing. Builders are using AI to launch ideas that reach the world. They don’t need big budgets or large teams, <strong>just willingness to experiment.</strong></p><p>Small is becoming the new big. AI isn’t replacing creativity; it’s <strong>expanding who gets to create</strong>, and how fast ideas can turn into <strong>reality</strong>.</p><p><em>If you’ve worked on similar AI-driven workflows or faced the same challenges, I’d love to hear what you learned — feel free to share your experience or perspective in the comments.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0513f5526d75" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/how-we-built-an-ai-ed-tech-workflow-that-turned-days-of-work-into-seconds-0513f5526d75">How We Built an AI Ed-Tech Workflow That Turned Days of Work Into Seconds</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Backlog Story [CRM Onboarding] ]]></title>
            <link>https://medium.com/@anastasiades.chr/backlog-story-crm-onboarding-4864c84b7965?source=rss-4ec84cd7858------2</link>
            <guid isPermaLink="false">https://medium.com/p/4864c84b7965</guid>
            <category><![CDATA[crm]]></category>
            <category><![CDATA[email-marketing]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[user-onboarding]]></category>
            <dc:creator><![CDATA[Christos Anastasiades]]></dc:creator>
            <pubDate>Fri, 27 Jul 2018 13:40:29 GMT</pubDate>
            <atom:updated>2018-07-27T13:40:29.579Z</atom:updated>
            <content:encoded><![CDATA[<h4>Mainly focus on the email Strategy 📬</h4><h3>Overview</h3><p>Onboarding is one of the things that most customers will experience. After understanding the users needs the CRM onboarding campaign aims to point show how your product can make users more successful.</p><p>Any CRM onboarding campaign must present the business offering solution. The trick though for the user to absorb the gist of your company better is to do that within multiple, short, clear, smart and at the right time touch points. Based on the point you want to make across you can use different channels either separately or either by combing them all together. To spread your content across multiple channels appropriately, you must have in mind each channels advantages and disadvantages.</p><p>Designing such a campaign and … BOOM 💥, more sign-ups will convert into a long lasting, churned resistant clients. In this post, I will try to explain how each CRM channel (Email, On-Site/In-App, Mobile notifications,Web-Push, Display, Chat apps) could be utilise appropriately in more depth with real examples.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*UtohbwDK2PqMTyb0rdOxnw.png" /><figcaption>image from chargify</figcaption></figure><h3>Channel — Email</h3><p>Email is a powerful and dynamic tool and it can be the core of the CRM onboarding experience. This is because the email open rate here in Greece at forky at least was up to 55%. Furthermore, each email has a big enough capacity to transfer the gist of the business and its small enough to keep the user’s attention.</p><p>At forky we identified that we could improve our conversion rate, so we embarked on a campaign to undertake that. We started working on this campaign by first listing what is important for our customers to know about Forky, and what micro conversions we would like them to complete. After we prioritise this two list and tried to combine them in a way to have a nice story flow. Looking back I found a plan draft that it might be helpful.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9yE3mUhdaHytemeY." /></figure><p><strong>Frequency &amp; duration: </strong>The campaign is being designed to last for 36 days, based on two facts. (A) fork’s previous onboarding email campaign lasted only for the first week upon user subscription where users received an email daily. That led to a 9% open rate drop off every day. (B) We saw that if the user wouldn’t make the first order in the first 36 days upon subscription, the chances of ever ordering was close to zero.</p><p><strong>Structure: </strong>Most of the emails are designed to have a primary and secondary objective which are meant to compliment each other. It’s important to highlight that most objectives have a unique Call To Action (CTA). Also, some emails have two versions. One version that it will be used to communicate an active voucher and the other version that contains no voucher.</p><p><strong>Email Content: </strong>A description of each email</p><p><em>Email - 1: </em><strong>Objective 1 </strong>We welcome the user and we tried to merge the company’s manifesto with the company’s promise. We tread the first email design with extra love while using the first singular in the copy. <strong>CTA 1 </strong>Visit forky’s page. It&#39;s nice to mention here, that a similar email was build at another startup called <a href="https://www.e-table.gr/">e-table</a>, where the CTA was a survey and the participation rate was 8%.</p><p><em>Email - 2: </em><strong>Objective 1 </strong>The Forky menu is curated from <a href="https://www.facebook.com/Peskias/">Christoforos Peskias </a>and all deserts from <a href="http://steliosparliaros.gr/">Stelios Parliaros</a>. Those two important chefs where one of the main forky marketing assets. The copy was curated in a way that the two Chefs were speaking to the audience. <strong>CTA 1 </strong>Chefs are inviting you to try forky. <strong>Objective 2 </strong>We emphasise for the second time in a row (also mentioned in email 1) that the menu is changing daily to achieve re-visits and to make known this unique aspect of our product. If users read the email and reach the end, it means that they are interested enough, and we have an opportunity to “hook” them. Discretely we mention an incentive, an offer to try our product. By being discrete with the offer here we are looking to “hook” users that are really interested in our product, and are not only focusing to eat cheap. Those users could be also segmented and use later on in other campaigns and further analysis. <strong>CTA 2 </strong>Use the offer.</p><p><em>Email - 3:</em> <strong>Objective 1</strong> We present real social media comments and testimonials to highlight our user’s love. Since this section was a .gif we alternated the comments at a speed of 2 seconds per frame. <strong>CTA 1 </strong>We invite users to make their first order by giving them a bold discount. Framed in a brilliant design <strong>Objective 2 </strong>To strengthen the bonds and at the same time avoid any users negative behaviour (like unsubscribing from the newsletter), we created this section where users can choose how and what to be communicated. <strong>CTA 2 </strong>Link to there profile at the opt-in section.</p><p><em>Email — 4:</em> <strong>Objective 1 </strong>At forky we do care about healthy eating and we pay extra attention to our nutritionist guidelines so our users could enjoy daily meals without any guilt. We made sure that the user knows about it with this email. <strong>CTA 1 </strong>Change your eating habits <strong>Objective 2 </strong>We created a Google form with basic feedback questions and we ask users to complete it.<strong> CTA 2 </strong>Give us feedback.</p><p><em>Email — 5:</em> <strong>Objective 1 </strong>If the user<strong> </strong>pre-orderers, the company is saving money for both food waste and logistics. To increase ordering repeatability and environmental responsibility, we are trying to convince users to pre-order by rewarding them with discounts from those company savings. <strong>CTA 1 </strong>Order before 12 pm</p><p><em>Email — 6:</em> <strong>Objective 1 </strong>We designed this email aiming to increase dinner orders. <strong>Objective 2 </strong>We found out that most dinner orders happened through the app so here it was the perfect time to mention about our app. Mobile app users are much more loyal compared to the rest and we urge them here to download it. <strong>CTA 2 </strong>Download our app</p><p><em>Email — 7:</em> <strong>Objective 1</strong>If you haven’t order by the 17th day upon sign-up, we want to know why and we are connecting you with the customer support. Furthermore, this email is served when a user abandons its cart. <strong>CTA 1 </strong>Chat with our customer support</p><p><em>Email — 8:</em> We decided not to implement it due to the limited resources.</p><p><em>Email — 9:</em> <strong>Objective 1 </strong>This is the last onboarding email and we remind the user about a big discount while using a sense of urgency. That was important because if the user would not convert after 36 days, data showed us that it would be a very little change to become a client after that. <strong>CTA 1 </strong>Use your discount.</p><p>After communicating the above plan with the team we came up with the following content presented in the “Email Template examples” section.</p><p><strong>Email Template examples</strong></p><p>Unfortunately, I can only provide 3 of the email templates only in their Greek version. The following examples are samples of the working files and are not the final templates. The first screenshot is the template for the first email, the second for the fifth one and the last screenshot for the ninth email.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/406/0*7dyP0SD_as7NKPgT." /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*6IraLJdc9kmFpxBdzYByyQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/680/0*HX0eMnuPU5vo0Q9x." /></figure><h3>Right Team</h3><p>As I think we achieved, each touchpoint must be appealing, so shout out to all awesome designers and copyrighters out there. Is also important to serve the content at a time that the user is ready to act upon it, so why not to give a thumbs up to the person that it will optimise those touch points.</p><blockquote>🤟Kudos 👏</blockquote><blockquote>I would like to give some love💚 to<a href="https://www.linkedin.com/in/katerinafoskolou/"> Katerina Foskolou</a> for being the shining colour 🌈 , <a href="https://www.linkedin.com/in/franceska/">Franceska Galafti</a> for being the speech 📢, Nikos Kouroupa for being the techie 🤖 and <a href="https://www.linkedin.com/in/maria-constantina-chardalia/">Marientina Chardalia </a>for being an excellent support ♠️ for this project. Also Kudos to <a href="https://www.linkedin.com/in/nana-kalafati-56988a19/">Nana Kalafati</a> for overseeing all the forky’s nutritionist 🍲communication knowledge and <a href="https://www.linkedin.com/in/giannakeasgeorge/">George Giannakeas</a> for being the inspiration 🧐 for this project. Finally, endless respect 💪for <a href="https://www.linkedin.com/in/michalisgk/">Michalis Gkontas</a> and <a href="https://www.linkedin.com/in/pierrepitsilis/">Petros Pitsilis</a> for being the founders of this tasty startup.</blockquote><p>The following picture, presents one design recommendation by Katerina that it was never used. I found it lovely and very creative job, and I thought to give it some life here 🤪</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9PjhcxAzEX8ip3u5." /></figure><h3>Channel — Display Marketing</h3><p>To ensure that the content reaches the maximum of users more than one time, your core content can be combined and served with other channels as well.</p><p>Display marketing can play a substantial role in onboarding. Using Google and Facebook display campaigns can remind the gist of the business in a more frequent manner. By implementing the right campaign set up, you can display the right message at the right time.</p><p>Having said that at forky, we created a Facebook onboarding campaign to strengthen the core message that was planned to communicate as demonstrated in the previous section. The user would enter this campaign after its subscription and pass into a different Ad-set each week. Each Ad-set had a topic like for example in the second week was “forky chefs” or in the third week was “food nutrition values”. Where appropriate each Ad-set was designed to contain more than 4 Ads while combining the 2 Objectives with the 2 versions (look at the previous section).</p><p>Furthermore to save some time, similar to the philosophy of sending the user to the right customised landing page after clicking onto a specific display campaign, you could trigger a customised pop-up message at there visit for each audience, to explain for example how similar audience (ex. Lawyers or teenagers) use your service/product in their everyday life.</p><p>Food for thought, if you decide to subsidise a specific audience and tap into their routines, you can create an awareness campaign message just for that. For example, if Uber finds out that those 18 years old men living in Athens, where had joined the army is a hot audience, an awareness campaign can be created with the motto “Don’t drink and drive”. This campaign can be accommodated with a 50% discount upon booking their first ride after midnight.</p><h3>Channel — On-Site / In-App</h3><p>The first on-site (or in-App) onboarding messages can be served at the first user visit. Here are some examples of how you can take advantage of this channel</p><p>While focusing to improve the subscription flow, metrics and the customer care team can help you identify where users struggle and where most of the sign-up drop-offs happen. Figuring out these inefficiencies you can customise messages to help user overcome there struggle. For example, that might be at the stage of the email confirmation or phone verification. This solution can become handy until the product team optimise the flow.</p><p>If you find out that the majority of the sign-up drop-offs are happening form a specific audience, then an automated process could be set up to connect them with the customer care team. The customer care team can investigate their situation further.</p><p>A survey can be placed to gather wealthy insights based on Avinash Kaushik <a href="https://www.kaushik.net/avinash/the-three-greatest-survey-questions-ever/">article</a> using the following three questions. What is the purpose of your visit to our App today? Were you able to complete your task today? If you were not able to complete your task today, why not?</p><h3>Channel — SMS and Chat apps</h3><p>SMS, Viber and mobile push notifications can be used for time-sensitive messages, where the user can see the message as soon as possible. For example, we can use efficiently those channels when we need to let users know that their offer is ending soon. Those channels can be supplementary and followed behind an email. By doing this at forky, the conversions were increased by +20% rather than using those channels individually.</p><h3>Onboarding Eligibility</h3><p>1. Exit — intent popup tools <br>2. Referred by friends<br>3. PR (ex. events)<br>4. 3rd Party databases</p><p>Should we include this leads to the onboarding campaign? Is it going to be the same campaign as the users that did subscribe? Is acceptable to use this emails?</p><p>Here are some Pros and Cons.</p><p><strong>Pros</strong><br>1. Whoever entered the campaign he/she be a subject to become a brand evangelist or even better an active user.</p><p>2. By having leads and users under one campaign it will be easier to avoid sending irrelevant emails or vouchers that users received before.</p><p><strong>Cons</strong> <br>1. The frequency of the lead emails will be smaller, but there is a high probability of finding the content irrelevant and unsubscribe. An even worse scenario is to mark the emails as spam.</p><p>2. The campaign would be more complicated and more bugs will consume time to be resolved.</p><p><strong>Questions that it will help us to decide:</strong></p><ul><li>Do you have the right to target or send emails to this addresses?</li><li>How many lead emails do you gather each month?</li><li>How many leads do they convert into users?</li><li>How many of those users use the service?</li><li>What would be the expected conversion that makes sense to spend additional resources?</li><li>Can we just include leads in the display campaign and test their conversion rate?</li></ul><p>My approach to this matter would be to use this emails and try to make the most out of them. Ideally, each of the above leads must have there owned onboarding campaign, while controlling the frequency, content and the CTA without being improper and annoying. Here it’s important to be really easy for the user to unsubscribe if he/she wants it to.</p><h3>Monitor the campaign</h3><p>Each channel above has its own measuring tools. At forky we used Intercom to set the email campaigns but its analysis tool wasn’t enough. For this reason, we created an onboarding email dashboard in Google Analytics to monitor the overall campaigns success. Please note that the metrics in the following picture is a testing snapshot serving the purpose of an example. Exporting the data from the dashboard, a grater analysis can happen using Excel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jg1RMhxASV_NXTWIgvvtBw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4864c84b7965" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[in-app quasi-financial “supermarkets” war]]></title>
            <link>https://medium.com/@anastasiades.chr/in-app-quasi-financial-supermarkets-war-596ae1b68359?source=rss-4ec84cd7858------2</link>
            <guid isPermaLink="false">https://medium.com/p/596ae1b68359</guid>
            <category><![CDATA[challenger-banks]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[neobanks]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Christos Anastasiades]]></dc:creator>
            <pubDate>Thu, 11 Jan 2018 15:34:29 GMT</pubDate>
            <atom:updated>2018-01-12T10:07:15.657Z</atom:updated>
            <content:encoded><![CDATA[<h4>🤑💴<strong> </strong>🌍<strong> </strong>💵<strong> </strong>💰<strong> </strong>💶<strong> </strong>🔫<strong> </strong>💷🔥</h4><h3>🏦 <strong>Background</strong></h3><p>The neobanks need: millennials trust brands like facebook and google rather than banks. So banks must be redefined, as <a href="https://vimeo.com/155098178">stated</a> by Eileen Burbude, Parner at Passion Capital.</p><p>In their efforts to redefine banks a lot of start-up companies were born; Some of which initiated to offer users a better financial management tool, others to resolve the banks foreign hidden withdrawal fees and others to offer easy investment tools to the public.</p><p>The market size is huge with the market leaders being the U.S, UK, and Germany and in that order.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cov8LXXq6nk1qe5Rj5ToQA.png" /></figure><h3>📈🧐 <strong>Analysis Introduction</strong></h3><p>Till date, big banks are unable to compete with neobanks in building a better product for their users due to the never <a href="http://www.redherring.com/europe/revolut-speed-efficiency-scale-fintech-heavyweight/">ending bureaucracy and red tape</a>.</p><p>The market is big but the competition is even greater. The neobanks are in war to establish the lion’s share. In my analysis it will be shown that in the attempt of companies to become profitable, they are all moving in the same direction which is to offer a one-stop-shop for their clients, that is to spend, save, and invest.</p><p>The survival of the fittest will be determined by the ability of companies to gain the trust of the people, and find the optimum monetisation model which aligns with their product offering. The companies’ right prioritisation will decide the winner of this war.</p><p>In the following report I chose three leading neobanks (Revolut, Monzo, Stash) that have a slightly different business model and status at the moment. The rest global neobanks, which are more than17, can fall under the documented categories.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4EvsalC63oDuDm4i58em7w.png" /></figure><blockquote>💰 <strong>Status</strong></blockquote><blockquote>HQ: London // Total Funding: £67M // Revenue: £2.3M //Total Employees: 150 //Users: 1M</blockquote><p>💼 <strong>Business model</strong></p><p><strong>Started</strong> with a free currency exchange rate and as a free money transfer up to £500. Competitive pricing plans were <a href="https://www.slideshare.net/NikolayStoronsky/revolut-deck-140115">available for greater flexibility</a>.</p><p><strong>Currently</strong> at the standard freemium plan, they implement £5,000 a month cap on their foreign exchange transfer with interbank FX rates. Additionally, they offer £200 free ATM withdrawals per month. Pivoted to freemium model that is generating revenue £10 per month, the premium plan withdrawal limit reaches £400, unlimited FX volumes, plus the user benefits from a travelling insurance as well. Furthermore one can borrow from their partner, Leading Works Limited in a fully integrated manner. Peering also with Simplesurance GmBH can offer device insurance schemes. Last but not least, Revolut’s physical card costs £5 plus express delivery if needed.</p><p><strong>Future:</strong> The company plans to expand into the United States and Asia plus they <a href="http://www.independent.co.uk/news/business/indyventure/revolut-uk-fintech-european-banking-licence-apply-brexit-leave-eu-a8043306.html">expect their European bank license by 2018</a>. Furthermore, an integrated investment platform will be added, from which users will be able to buy stocks and bonds but also a loan. Finally, in addition to their integration with Bitcoin, the company will invest further in cryptocurrencies introducing Ethereum and Litecoin.</p><p>📱 <strong>Product milestones</strong></p><p><strong>01/17 </strong>Bill Splitting Functionality <strong>02/17 </strong>Current Accounts <strong>02/17 </strong>RITA chatbot <strong>03/17 </strong>Partnership with Lending Works <strong>03/17</strong> Revolut Premium <strong>04/17 </strong>Partnership with Trussle mortgages <strong>06/17 </strong>Business Current Account <strong>07/17 </strong>Euro Accounts <strong>09/17 </strong>Recurring Payments <strong>??/18</strong> In-house payment processing service</p><p>⚠️ <strong>Opportunities &amp; Risks</strong></p><p>👍🏼 Free FX served as the marketing channel has given them more than 1M subscribers. Pivoting the model was an essential effort of making the business sustainable.</p><p>👍🏼 European banking license in 2018. Will expand product offering to include interest-bearing savings and consumer credit products. <a href="https://www.bloomberg.com/news/articles/2017-11-08/payments-startup-revolut-wants-to-become-a-bank">It will also inspire trust to acquire more users</a></p><p>👍🏼 Revolut offers the most competitive solution to transfer money</p><p>👍🏼 Provision of Physical cards to the company’s subscribers is an additional revenue stream</p><p>👍🏼 One of the few neobanks that can offer value to Bitcoin owners</p><p>👎🏽 Brexit will cost time re-figuring licensing</p><p>👎🏽 Fair but fewer Freebies will prove if existing users will stay loyal</p><p>🤔 <strong>Opportunity Analysis</strong></p><ul><li>Revolut is one of the most competitive and complete product, and as a result is now the main player in the UK. There is still space though to dominate Revolut in European markets.</li><li>Following the above, If non Britain neobanks chooses to aggressively compete in the UK market, they can take advantage of Brexit.</li><li>Is the European market excluding Germany and Britain big enough? A further exploration of the global market is definitely worth the shot.</li><li>Competitors must make a plan to capitalise on banking products such us N26 with the announcement of the <a href="https://n26.com/introducing-n26-metal/">Black Metal</a> cart.</li><li>Competitors must implement product priority to accept cryptocurrencies</li><li>Free FX is an enormous success in marketing and attracts higher earners. Does other neobanks need to walk that path?</li><li>A question which remains, will Revolut manage to pivot away from being an FX card?</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tDpbetf7ZfRAC1e4n_O2sQ.png" /></figure><blockquote>💰 <strong>Status</strong></blockquote><blockquote>HQ: London // Total Funding: 138M$ // Total Employees: 150 // Users: 250K</blockquote><p>💼 <strong>Business model</strong></p><p><strong>Started:</strong> in similar fashion to Revolut with zero withdrawal fee and zero-commission in foreign exchange. It has positioned itself to allow greater control and visibility of your finances. A <a href="https://www.maparesearch.com/monzo-updates/">beautiful UX </a>was built to analyse the spending of the users current account.</p><p><strong>Currently:</strong> Monzo operates only in the UK. They focus in building a valuable product first while ignoring their revenue streams. Monzo still believes that it can keep the ATM zero withdrawal fee but it’s currently charging 3% fee outside the UK after a 30 day limit of 200£.</p><p><strong>Future: </strong>Monzo<strong> </strong>wants to be a financial marketplace control centre for each user. It started already partnering up with energy providers, and allows its users to switch their contracts to a new provider. Monzo is also working to peer with other lenders or investment funds to help you save or borrow money. Monzo is entitled to a commission on the earning of these third-party products (8).</p><p>📱 <strong>Product milestones</strong></p><p><strong>02/16 </strong>Golden Tickets Referral scheme <strong>12/16 </strong>Split bills &amp; money request <strong>04/17 </strong>Fully authorised banking licence <strong>04/17 </strong>Available on Android Pay <strong>07/17 </strong>Early phase current accounts <strong>07/17</strong> AI at<strong> </strong>Help screen <strong>07/17 </strong>Energy provider consternation <strong>09/17 </strong>Monzo pots (savings feature) <strong>??/18 </strong>Public API <strong>??/18</strong> Monzo Uni. (financial education) <strong>??/??</strong> Mental health support</p><p>⚠️ <strong>Opportunities &amp; Risks</strong></p><p>👍🏼 Winning their customers/community trust through their excellent customer engagement and full transparency culture</p><p>👍🏼 The second Payment Services Directive (<a href="https://www.evry.com/en/news/articles/psd2-the-directive-that-will-change-banking-as-we-know-it/">PSD2</a>) can come in handy, complimenting its vision to be the best financial management tool</p><p>👎🏽 Brexit will cost time refiguring licensing</p><p>👎🏽 Starling Bank charges zero fees with regards to ATM abroad and Revolut offers cheaper rates</p><p>👎🏽 Still rolling out current accounts, it risks falling behind its rivals — neobanks</p><p>👎🏽 Monzo is losing £50 per year per pre-paid customer, and £250 million has been spent on Monzo’s pre-paid cards to date.</p><p>🤔 <strong>Opportunity Analysis</strong></p><ul><li>Monzo has created a very strong brand and has earned the trust of the community. Competitors must incorporate Monzo’s customer engagement and transparency culture into their DNA.</li><li>Can competitors or new entires replicate the brand values in other global markets like the US?</li><li>Competitors can take advantage of the post-PSD2 era like making a public API and further prepare a strategic plan and key partnerships. A further analysis is required for this task.</li><li>We can not say with confidence if Monzo’s marketplace model is working as it currently lacks integration. Monzo though believes in it and is investing technical resources.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e_N4EVVlb4tHn85yzBHFdA.png" /></figure><blockquote>💰 <strong>Status</strong></blockquote><blockquote>HQ: Ney York // Total Funding: 79M$ // Total Employees: 80 //Users: 2.5M subscriptions &amp; 1M clients</blockquote><p>💼<strong> Business model</strong></p><p><strong>Started:</strong> in the US, offering a mobile app which aims to make micro-investing easier at low-risk for people that are inexperienced and lack financial skills. Stash only charges a fee of $1 per month for accounts under $5,000, and 0.25% per year for those with balances over $5,000.</p><p><strong>Currently:</strong> the company is offering Roth and Traditional IRA accounts to start saving for retirement. Stash Retire program offers tax — advantages and long-term investing tools. The subscription fee is $2/month for accounts up to $5,000, and 0.25% for accounts with a balance higher than $5K.</p><p><strong>Future: </strong>Early 2018 Stash Banking will be launched. This Bank challenger will have free accounts, FDIC-insured, with no fees, no minimum balance requirements and will be accessible via the largest ATM network in the US. Following the previous slides, not by surprise, CEO Krieg Brandon mentioned “Stash will be a one-stop-shop for our clients, to spend, save, and invest.</p><p>📱 <strong>Product </strong><a href="https://investorjunkie.com/48501/acorns-vs-stash/"><strong>milestones</strong></a></p><ul><li><strong>Stash Invest</strong> simply makes investment <em>recommendations</em>, leaving it up to you to do the actual investing.</li><li>Supplement your portfolio with your own <strong>investment choices</strong>.</li><li><strong>Auto-Stash </strong>(autopilot investment)</li><li><strong>Stash Learn</strong> (financial education)</li><li><strong>Retirement</strong> long-term investing tool</li></ul><p>⚠️ <strong>Opportunities &amp; Risks</strong></p><p>👍🏼 It’s highly likely that existing Stash Invest users will use Stash bank for their everyday finance</p><p>👍🏼 In-house investment algorithms <a href="http://www.bankingtech.com/1042322/stash-reveals-plans-to-launch-mobile-first-banking-service/">omit the need to share revenue with other third-party solutions</a>. Other neobanks share this income</p><p>👎🏽 Ongoing fees make Stash inferior to Robinhood, <a href="https://techcrunch.com/2016/12/14/micro-investing-app-stash-raises-25-million-series-b/">which generates revenue from interest on uninvested cash and margin accounts</a></p><p>👎🏽 Established US competition like Qapital, Moven, Bank Mobile, and Simple</p><p>🤔 <strong>Opportunity Analysis</strong></p><ul><li>Stash value proposition focused on saving and investing. Most neobanks based their monetisation by sharing these services</li><li>Competitors must follow this model closely and evaluate the opportunity to build these services in-house</li><li>Stash has grown fast and might defend the USA well. Having said that, at a TechCrunch (<a href="https://www.youtube.com/watch?v=ncWRe-JIWRs">Disrupt Berlin 2017</a>) event, the founder of N26 Valentin Stalf announced his company expansion in the USA while there is a pool of great VC opportunities.</li><li>If an expansion in the USA is in companies future plans, companies might want to focus promoting their day to day finance services, rather than investment services. Monzo’s model might be an attractive brand for the USA.</li><li>Last but not least, Stash will soon answer the question, <strong><em>whether neobanks are more trustworthy as investment platforms or as banks</em></strong></li></ul><blockquote>🤘🏼 Kudos 👏🏽</blockquote><blockquote>I would like to give some 💚 to <a href="https://www.linkedin.com/in/vassilis-panayiotou-6b0200147/">Vassilis Panayiotou</a>, Lawyer for proofreading this post and <a href="https://www.linkedin.com/in/victortrokoudes/">Victor Trokoudes</a>, <a href="https://withplum.com/?utm_content=inapp_quasifinancial_supermarket_war">Plum</a> CEO for giving me insides and a head start for the new-age FinTech industry.</blockquote><blockquote><a href="https://withplum.com/?utm_content=inapp_quasifinancial_supermarket_war">Plum</a> has a smart 🧠 algorithm, and a rocking 😎 Facebook interface to help you save money automatically. Being my personal favourite solution to save money, Plum entered with the right foot the quasi-financial supermarkets war!</blockquote><h3>👋👋Closing 👋👋</h3><p>Currently neobanks are sweating to get the millennials trust with their income salary. While doing that, mature neobanks are already making revenue from overdraft interest, credit &amp; loans interest, card transactions interchange fees, investment plans and insurance plans in that order. In addition to that, foreign exchange is an other sort of income excluding Revolut which is subsidising this service for marketing purposes.</p><p>From the above revenue streams mature neobanks can pay their bills and already be sustainable while they are not involved with any physical store costs. Their main cost is the growth expenses, card printing (while some of the neobanks are giving them out from free), the subsidisation for the localised transactional fees and the salaries to attract the high end engineers.</p><p>Based on the above analysis, the neobanks that would make it to the finish line first, would be the ones that from one side ⚖ ️ will maintain a transparent, supportive and innovative service and from the other side ⚖ will set the correct growth pace while managing their costs.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=596ae1b68359" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>