<?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 Graphlinkz on Medium]]></title>
        <description><![CDATA[Stories by Graphlinkz on Medium]]></description>
        <link>https://medium.com/@graphlinkz?source=rss-6b50ee707ee0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*kKQXPijtyHX_5uruBrVgqw.jpeg</url>
            <title>Stories by Graphlinkz on Medium</title>
            <link>https://medium.com/@graphlinkz?source=rss-6b50ee707ee0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 20:46:53 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@graphlinkz/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[How a simple “AI Executive Summary” request became a project about judgment, context, and…]]></title>
            <link>https://graphlinkz.medium.com/how-a-simple-ai-executive-summary-request-became-a-project-about-judgment-context-and-87a2cc7c19c6?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/87a2cc7c19c6</guid>
            <category><![CDATA[ux-research]]></category>
            <category><![CDATA[product-discovery]]></category>
            <category><![CDATA[user-research]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Tue, 07 Jul 2026 20:07:09 GMT</pubDate>
            <atom:updated>2026-07-08T15:20:58.085Z</atom:updated>
            <content:encoded><![CDATA[<h3>When AI Wasn’t the Answer: A Product Design Case Study</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iDNr_i3fQtge3yOYd20Ojw.png" /></figure><p>Some product requirements arrive looking very innocent.</p><p>Polite, even.</p><p>They walk into the room wearing a nice little button-up shirt and say things like:</p><blockquote><em>“Customers are asking for an AI Executive Summary feature.”</em></blockquote><p>Very reasonable. Very modern. Very “let’s add AI because the market is hot and the competitors are moving.”</p><p>And honestly, on the surface, the request made sense.</p><p>The product was V-AI, an AI SaaS platform for enterprise legal and compliance teams. The users were reviewing large volumes of contracts across banks, insurance companies, healthcare networks, and other highly regulated organizations.</p><p>Long contracts. Busy teams. High stakes. Lots of documents.</p><p>So naturally, the original requirement was:</p><blockquote><em>“When users upload a contract, V-AI should automatically generate an AI Executive Summary.”</em></blockquote><p>At first glance, that sounds like a good feature.</p><p>Contracts are long.<br>People are busy.<br>AI can summarize text.<br>Boom. Product strategy.</p><p>Except… that’s not quite it.</p><p>Because after working through the problem using our process for problem-solving, the “AI Executive Summary” feature was discarded — totally.</p><p>And what we went for was something much more interesting:</p><blockquote><em>An Institutional Knowledge Layer with a persistent context panel that helps legal and compliance teams understand why a contract matters, not just what the contract says.</em></blockquote><p>This case study is about how that happened.</p><h3>Started as a very Reasonable Feature Request</h3><p>The product challenge started inside V-AI, an AI platform used by enterprise legal and compliance teams to review contracts faster.</p><p>The PM came with what sounded like a clear request:</p><blockquote><em>Enterprise customers are asking for AI Executive Summaries.<br>Sales says prospects ask about it in demos.<br>Leadership thinks it could be a differentiator.<br>We should explore shipping it.</em></blockquote><p>Classic product moment.</p><p>There is customer demand.<br>There is sales pressure.<br>There is leadership interest.<br>There is AI involved.</p><p>What could possibly go wrong?</p><p>Well, a lot actually.</p><p>The dangerous thing about “reasonable” feature requests is that they often hide a big assumption and that is:</p><blockquote><em>“The thing customers ask for is the thing they actually need”.</em></blockquote><p>That assumption is where the work began. Instead of jumping into screens, I started with the asking questions.</p><h3>1. Question the Requirement</h3><p>The first move was not to design.<br>No wireframes. No dashboard. No “AI summary card.” etc<br>The first move was to question the requirement.</p><p>I wanted to know:</p><ul><li>Who exactly asked for this?</li><li>Were they paying customers or prospects?</li><li>What kind of customers were asking?</li><li>What problem were they trying to solve?</li><li>What happens today without the feature?</li><li>What evidence do we have beyond “people mentioned it”?</li><li>Is this a real workflow problem or a demo-room request?</li><li>Are users asking for summaries because they need summaries, or because they do not know how else to describe the pain?</li></ul><p>That last question became very important. Because “we need summaries” can mean many things. It can mean:</p><blockquote><em>“I do not want to read 40 pages.”</em></blockquote><p>But it can also mean:</p><blockquote><em>“I am scared I will miss the part that matters.” etc.</em></blockquote><p>Those are very different problems. And if you solve the wrong one, you can build a very polished feature that still fails quietly.</p><h4>What the Stakeholders Revealed</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5lnxuEiD6K-Mo6lwMr21SQ.png" /></figure><p>As I questioned the requirement, the request started to wobble.</p><p>Sales was excited because prospects liked the idea in demos.</p><p>Of course they did. “AI Executive Summary” sounds great in a demo. It is the kind of feature that makes people nod. It feels obvious. It feels modern. It sounds like something your competitor might put on a landing page with a gradient background and a sentence like:</p><blockquote><em>“Unlock contract intelligence at scale.”</em></blockquote><p>Lovely.</p><p>But Customer Success had a different view.<br>Paying customers were not simply saying, “Please summarize my contracts.”</p><p>Their pain that Legal and compliance teams were reviewing many contracts, but the difficulty was deciding which contracts deserved deeper attention.</p><p>That was the moment the original requirement started to look suspicious.<br>Because an AI summary can tell you what is in the contract.</p><p>But the bigger problem was understanding why the contract mattered.<br>Those are not the same thing.</p><h4>The Real Problem Was Not Reading. It Was Judgment.</h4><p>We already saw that Legal and compliance teams were not merely asking:</p><blockquote><em>“What does this contract say?”</em></blockquote><p>They were asking:</p><blockquote><em>“Should I care about this contract?”<br>“Is this risky for our business?”<br>“Has someone reviewed something like this before?”</em></blockquote><p>That is a much more valuable problem to solve. And it reframed the job-to-be-done:</p><blockquote><em>When I receive a large batch of contracts, help me assemble enough organizational context to determine which contracts matter, why they matter, and who should review them, so that I can confidently allocate my team’s attention without worrying that I have overlooked something important.</em></blockquote><p>This was the turning point. The real job was about context and confidence.</p><h3>2. Delete</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PBTCwnQweB8L9Bj7bE972w.png" /><figcaption>Idea dumping</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TMj-96nvGDzumJqjzPLbrA.png" /><figcaption>Prioritizing solutions to find the sweet spot to focus on and Delete the rest</figcaption></figure><p>This is my favorite part of the process because it went wild.</p><p>Most product teams are comfortable adding features. But the our process forces us to question what we can delete.</p><p>At this stage, several ideas had to go.</p><ol><li><strong>Deleted AI Executive Summary: </strong>The original had to go, not because summarization is useless, but because it was not the right problem to solve.</li><li><strong>Deleted AI Prioritization:</strong> Another tempting idea was to let AI rank contracts automatically. This sounded powerful. But it introduced a trust problem. If the system says, “Review this one first,” the user will immediately ask <em>“Why?” </em>And if the system cannot explain that ranking using meaningful business context, the ranking becomes noise. Worse, it can become dangerous noise. For legal and compliance teams, false confidence is expensive.</li><li><strong>Deleted AI Ranking: </strong>Similar problem. Ranking contracts sounds efficient, but it risks pretending the system understands business importance better than the organization does.</li><li>And we kept deleting even more potential features.</li></ol><h3>3. Simplify and Optimize</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*eRZDSpv6rcQTvMv6.png" /><figcaption>Prioritizing the assumptions for the chosen idea</figcaption></figure><p>The surviving solution was based on one core idea:</p><blockquote><em>The system owns facts. Humans own meaning.</em></blockquote><p>That line became the design principle for the entire interaction.The product should not pretend to replace the reviewer’s judgment.</p><p>It should help the reviewer see the facts, surface relevant context, and capture what humans know.</p><p>The system should say:</p><blockquote><em>“Here is what is in the contract.”<br>“Here is what the organization already knows.”<br>“Here is what previous reviewers noted.”<br>“Here is what you may want to remember for next time.”</em></blockquote><p>But the human still decides what matters.</p><p>That distinction is everything.</p><p>Because in high-stakes enterprise workflows, the goal of AI is not always to “decide for me.”</p><p>Sometimes the better role is:</p><blockquote><em>Help me see clearly.<br>Help me remember.<br>Help me not miss important context.<br>Help me move faster without losing control.</em></blockquote><p>That led to a much simpler interaction.</p><h4>The Final Simplified Flow</h4><p>The core interface became:</p><pre>Contract List<br>    ↓<br>Selected Contract<br>    ↓<br>Persistent Side Panel<br>    ↓<br>Contract Facts<br>    ↓<br>What Reviewers Know<br>    ↓<br>What should you remember?<br>    ↓<br>Remember</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IWL4YNRzUWjq6qRj.png" /></figure><p>Just a focused review experience.</p><p>The reviewer moves through contracts, inspects each one, and sees a persistent context panel with the information needed to understand its significance: key facts, customer or vendor context, prior notes, institutional memory, reminders, and a way to add new context back into the system.</p><h4>Designing for the Moment of Review</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*iq44PcxUG_klNIZl.png" /><figcaption>Adding new context for IKL</figcaption></figure><p>Context only matters when it appears at the right moment.</p><p>Not in a dashboard. Not in another tab. Not buried in a knowledge base.</p><p>It needs to be present while the reviewer is making the decision.</p><p>That is why the persistent side panel became central: it keeps organizational memory beside the work.</p><p>Enterprise work is already fragmented.</p><p>The product should not add another fragment.</p><h3>4. Accelerate Cycle Time</h3><p>After simplifying the product, the next question was:</p><blockquote><em>How do we make this faster?</em></blockquote><p>Not by adding more features but by removing friction.</p><p>The target feeling became:</p><blockquote><em>Reviewing contracts should feel like moving through an email inbox, but with organizational memory beside every contract.</em></blockquote><p>That became the acceleration model.</p><p>Legal and compliance reviewers often move through many items. So the interface needed to support rhythm.</p><p>To support that, the design introduced several speed improvements.</p><ol><li><strong>Keyboard Navigation:</strong> Reviewers should be able to move through contracts with keyboard controls. Next contract. Previous contract.</li><li><strong>Focus preservation:</strong> Moving between contracts should not reset the reviewer’s mental state. The contract changes, but the interface stays consistent.</li><li><strong>Lazy Loading:</strong> Not every piece of context needs to load at once. Prioritize what the reviewer needs immediately, then load the rest progressively.</li><li><strong>Adjacent Contract Prefetching:</strong> If the reviewer is likely to open the next contract, load it quietly in the background.</li><li><strong>Optimistic Updates:</strong> When reviewers add a note or save something to memory, the interface should respond immediately while the system confirm in background.</li></ol><h4>What the MVP became</h4><p>By the end of the process, the MVP was no longer just AI contract summaries.</p><p>It became a contract review workspace that keeps institutional context beside every contract, helping reviewers understand business significance and capture what the organization should remember.</p><p>The MVP had three core parts:</p><ol><li>A focused contract review list.</li><li>A persistent context panel with relevant organizational knowledge.</li><li>A simple way for reviewers to answer one question: <strong>“What should we remember about this contract?”</strong></li></ol><p>Simple, human language that makes contributing knowledge feel natural.</p><h3>5. Automation Was Not the Starting Point</h3><p>Automation came last, not first.</p><p>AI products often rush to automate decisions before the workflow is understood. But if the process is unclear, automation only makes mistakes faster.</p><p>In this case, AI should first help reviewers retrieve the right context. Decision-making stays with the human.</p><p><strong>Automate retrieval before automating judgment.</strong></p><h4>The Six-Month Future: What This Could Become</h4><p>Six months into this projected future, V-AI is no longer just another AI contract summary tool.</p><p>Its advantage is organizational memory.</p><p>Reviewers see relevant context from past decisions, similar contracts, and team knowledge alongside every contract — helping them make faster, more consistent decisions without relying on tribal knowledge.</p><p>The result is better review speed, stronger collaboration, easier onboarding, and greater trust in AI-assisted workflows.</p><p>More importantly, V-AI stops being a tool that simply reads contracts.</p><p>It becomes a system that helps organizations remember what matters.</p><h3>What I Learned as a Product Designer</h3><p>This exercise reminded me that product design is not just about making features usable.</p><p>Sometimes the most valuable design work happens before the feature exists.</p><p>The real work is asking <strong><em>Are we solving the right problem? </em></strong>Because if the answer is no, every polished screen becomes useless decoration.</p><p>This simulation also reinforced a few product lessons I want to keep carrying forward.</p><ol><li>Customers often ask for solutions in the language available to them</li><li>Deleting is a product skill. Every deleted idea protects the user from complexity and protects the team from building unnecessary work.</li><li>Speed comes after clarity. Accelerating a bad workflow is reverse-progress.</li></ol><h3>The Case Study in One Sentence</h3><p>What began as an AI contract summarization request became an opportunity to help legal and compliance teams preserve and reuse institutional knowledge.</p><p>That shift came from asking a simple question:</p><blockquote>What problem are we actually solving?</blockquote><p>Sometimes the best product decisions come from challenging the original requirement.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=87a2cc7c19c6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[7 Questions That Saved Us From Building the Wrong AI Feature.]]></title>
            <link>https://graphlinkz.medium.com/7-questions-that-saved-us-from-building-the-wrong-ai-feature-67c575325343?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/67c575325343</guid>
            <category><![CDATA[product-discovery]]></category>
            <category><![CDATA[ux-research]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Wed, 01 Jul 2026 09:57:44 GMT</pubDate>
            <atom:updated>2026-07-01T09:57:44.459Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k7brdGQcJoaqEVoep5ZaGw.png" /><figcaption>Finding the right path</figcaption></figure><p>Some teams are very good at building clever solutions.</p><p>Beautiful dashboards. Smart AI features. Fancy automation. Polished flows. The kind of thing that makes people in a product review say, “Nice.”</p><p>The only small issue?</p><p>They sometimes build all of that for the wrong problem.</p><p>I recently worked through a product discovery challenge around an AI SaaS product. Let’s call the company <strong>“VAI”</strong>.</p><p>VAI helps enterprise legal and compliance teams review large volumes of contracts. Think banks, insurance companies, healthcare networks, and other organizations that deal with a frightening amount of documents.</p><p>The initial requirement from the PM sounded simple:</p><blockquote>“Customers are asking for an AI Executive Summary feature. When they upload a contract, VAI should automatically summarize it.”</blockquote><p>On the surface, this sounded reasonable.</p><p>Legal teams handle long contracts. Long contracts are boring. People do not have time. AI can summarize things. Boom. Ship it.</p><p>Very <strong><em>neat</em></strong>. Very <strong><em>modern</em></strong>. Very <strong><em>dangerous</em></strong>.</p><p>Because the moment I heard the requirement, something felt off.</p><p>Not because an AI summary is a bad idea. It might be useful. But useful is not the same as necessary. And “customers asked for it” is not the same as “this is the actual problem.”</p><p>So instead of opening Figma and pretending to be productive, I started asking questions.</p><p>The first question was simple:</p><blockquote><strong>Who is asking for this?</strong></blockquote><p>Turns out, the request was coming from multiple places.</p><p>Sales had heard it during demos. Some prospects asked, “Can the AI just give us a quick summary before we go into the details?” Fairrrrrrrr</p><p>Customer Success had heard similar things from a few enterprise customers.</p><p>Leadership was also interested because competitors were talking about AI summaries and AI agents.</p><p>Already, this was a yellow flag. Not a red flag. A yellow one.</p><p>Because “customers are asking” can mean many things. It can mean paying customers are blocked. It can mean prospects mentioned it once on a sales call. It can mean the CEO saw a competitor’s landing page and suddenly everybody is inspired.</p><p>So I kept digging.</p><p>I spoke with Sales. They said the request came up often during demos, especially with banking and insurance prospects. But when we looked closer, those prospects were not all asking for the same thing.</p><p>One wanted a one-page briefing for executives.</p><p>Another wanted to know which contracts required immediate legal review.</p><p>Another wanted to quickly scan thousands of vendor agreements during due diligence.</p><p>Another basically asked, “<strong>Can your AI tell me if I should care about this document?”</strong></p><p>That last one changed the direction of the whole investigation.</p><p>Because that is not the same as:</p><blockquote>“Summarize this contract.”</blockquote><p>That is closer to</p><blockquote>“Tell me whether this contract deserves my attention.”</blockquote><p>Different problem. Different product direction.</p><p>Then I spoke with Customer Success.</p><p>The explicit requests for “AI Executive Summary” were not that many. But when we looked for broader patterns, something more interesting appeared.</p><p>Customers were saying things like:</p><blockquote>“We upload hundreds of contracts and don’t know where to start.”</blockquote><blockquote>“Our analysts spend too much time figuring out which contracts matter.”</blockquote><blockquote>“Executives keep asking us to explain what is happening.”</blockquote><blockquote>“We don’t always know which contracts should go to Legal, Compliance, Risk, or Procurement.”</blockquote><p>At this point, the requirement was starting to crack.</p><p>The original assumption was: Customers need summaries because they don’t understand contracts.</p><p>But the evidence was pointing somewhere else: Customers need help deciding which contracts deserve attention.</p><p>Then we looked at usage data.</p><p>Banking and insurance customers were not the majority by customer count, but they represented a huge portion of revenue and document volume.</p><p>For banking customers, a large percentage of uploaded documents were never opened beyond the initial upload screen. Many others were opened once and never revisited.</p><p>That was not a “people don’t like reading” problem.</p><p>That looked more like a triage problem. So we interviewed actual customers. And this is where things got very interesting.</p><p>One customer explained that they receive contracts in batches. Not one or two documents. Hundreds. During vendor review cycles, they might receive 200 to 800 contracts over a few days.</p><p>Their first problem is not reading the contracts. Their first problem is deciding where to start.</p><p>They have to ask:</p><p>* Which contracts are urgent?<br>* Which contracts are high risk?<br>* Which contracts can wait?<br>* Which contracts need Legal?<br>* etc…</p><p>One customer said something that completely reframed the project:</p><blockquote>“Our problem isn’t understanding contracts. Our problem is deciding which contracts deserve a lawyer’s attention in the first place.”</blockquote><p>That revealed pretty much the Job to be done</p><p>The team was not asking for summaries because summaries were the goal. They were asking for summaries because they were trying to reduce uncertainty.</p><p>They were trying to answer questions like “What should I pay attention to?”, “Why does this matter?” , “Who should review this?” , “What happens if we miss it?”</p><p>So the requirement changed.</p><p>Before discovery, it was:</p><blockquote>Build AI Executive Summaries.</blockquote><p>After questioning the requirement, it became:</p><blockquote>Help enterprise legal and compliance teams confidently prioritize contracts by making the business significance of each contract visible.</blockquote><p>That is a very different design challenge.</p><p>An AI summary might still become part of the solution later. But it is no longer the center of the problem. And that distinction matters. Because if we had jumped straight into designing summaries, we might have built a very polished feature that summarized the wrong contracts beautifully.</p><p>Imagine that.</p><p>A clean UI. Great typography. Nice AI-generated bullets. Smooth loading states. And still, the customer misses the contract that actually matters.</p><p>That is how teams build <strong><em>clever solutions to the wrong problem.</em></strong></p><p>This is why designers need to question requirements.</p><p>Not to be difficult. Not to fight PMs. Not to sound smart in meetings.</p><p>But because requirements are often compressed assumptions.</p><p>When a PM says:</p><blockquote>“Build an executive summary feature,”</blockquote><p>inside that sentence might be hidden assumptions like:</p><p>* Customers know which contracts to open.<br>* The contract contains enough information to determine importance.<br>* The issue is reading speed.<br>* Summarization will solve prioritization.<br>* All customers mean the same thing when they ask for a summary.<br>* The feature request represents the actual pain.</p><p>Your job as a designer is not to accept all of that politely and go make boxes.</p><p>Your job is to unpack it. Ask questions like:</p><ul><li>Who asked for this?</li><li>What exactly did they say?</li><li>What were they trying to do?</li><li>What happens if we don’t build it?</li><li>Is this a problem for one customer or many?</li><li>Is this a sales request or an existing customer pain?</li><li>What behavior in the product supports this?</li></ul><p>That is where product design becomes more than interface design. Because sometimes the most valuable design work happens before the screen exists.</p><p>In this project, the biggest shift was not visual.</p><p>It was conceptual. We moved from:</p><blockquote>“How might we summarize contracts?” to “How might we help teams confidently decide which contracts deserve attention?”</blockquote><p>That one shift changes everything.</p><ul><li>It changes the data we need.</li><li>It changes who we interview.</li><li>It changes what we measure.</li><li>It changes what we design.</li></ul><p>It changes whether the solution is a summary, a prioritization system, a routing workflow, a context layer, an alerting model, or something else entirely.</p><p>So yes, AI can summarize contracts.</p><p>That is not the hard part. The hard part is knowing whether summarizing the contract actually solves the customer’s problem.</p><p>And this is the lesson I keep coming back to:</p><p>A requirement is not the truth.</p><p>A requirement is a clue.</p><p>Treat it like one.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=67c575325343" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What Gemini Taught Me About Designing “Everything AI” Products]]></title>
            <link>https://graphlinkz.medium.com/what-gemini-taught-me-about-designing-everything-ai-products-904097a99ea3?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/904097a99ea3</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[google-gemini]]></category>
            <category><![CDATA[trust]]></category>
            <category><![CDATA[hci]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Thu, 18 Jun 2026 04:59:11 GMT</pubDate>
            <atom:updated>2026-06-18T04:59:11.831Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZZxWw4PfC2wi_nG-AbjxQg.png" /></figure><p>I recently spent some time studying Gemini, not just as another chatbot, but as an AI product trying to play a much bigger role.</p><p>The more I explored it, the clearer the product ambition became: It is trying to be the place you go for almost anything AI can help with.</p><p>Research. Search. Writing. Image generation. Video generation. Android assistance. General questions. Creative exploration. Productivity. It is slowly becoming an “everything AI” layer within Google’s ecosystem.</p><p>That makes Gemini interesting to study from a product design and HCI perspective.</p><p>The question is not only: <strong>Can Gemini do many things?</strong> The better question is: Can users clearly understand what Gemini can do, trust what it gives them, and stay in control while using it?</p><h3>Gemini’s real job: becoming the “everything AI”</h3><p>At the surface level, Gemini is a general-purpose AI tool.</p><p>You can ask it questions, research topics, generate content, analyze information, create images, work across different formats, and interact with it through text, images, voice, and more.</p><p>But beneath that, Gemini seems to be doing something more strategic. It is has grown to become the AI layer for Google’s ecosystem.</p><p>If Gemini could describe itself to users, it might sound like: <strong><em>“Whatever you need, come to Gemini. We probably have a way to help.”</em></strong></p><p>That positioning is powerful, but it also creates a design challenge. The more things an AI product can do, the harder it becomes for users to understand its boundaries. A focused AI product has a clearer mental model. For example, a coding assistant helps you code. A research assistant helps you research. A writing assistant helps you write.</p><p>But Gemini wants to be many things at once: assistant, researcher, search engine, creative tool, productivity layer, and mobile companion. That breadth is useful, but it also means the interface has to work harder to help users understand what is possible.</p><h3>The mental model Gemini creates</h3><p>The mental model Gemini seems to encourage is simple: Ask naturally, and Gemini will figure it out. This is one of the strengths of conversational AI. Users do not need to learn a complex interface before they begin. They can simply express an intent, even if that intent is rough, incomplete, or exploratory.</p><p>That makes the experience feel open and flexible.But it also places a lot of responsibility on the system. When an AI product teaches users to “just ask anything,” the product must be good at interpreting vague intent, asking clarifying questions, and guiding users toward better outcomes.</p><p>This is where general-purpose AI products become tricky. Freedom is valuable, but too much freedom can also make the product feel unclear. <strong>If a user does not know everything Gemini can do, they may underuse it</strong>. If they overestimate what it can do, <strong>they may trust it in the wrong situations</strong>.</p><p>So the design challenge is not only about providing capabilities. It is about helping users form an accurate mental model of those capabilities.</p><h3>Trust is built through sources, correction, and restraint</h3><p>One thing Gemini does well is make an effort to show where information is coming from and that matters. In AI products, trust is not built only by sounding intelligent. It is built by helping users verify, inspect, and understand the output.</p><p>During my test, I noticed something important: Gemini did not simply agree with me when I insisted on something wrong. It was willing to push back, explain why the information was incorrect, or admit when the information I was looking for did not exist.</p><p>I asked it</p><blockquote>“Can you summarize the plot and critical reception of the 2025 movie <em>The Midnight Whisperer</em> starring Tom Holland?”</blockquote><p>The important detail: this movie does not exist and that is a strong signal for trust. A trustworthy AI product should not be designed to please the user at all costs. It should be designed to help the user move closer to what is true, useful, or accurate.</p><p>So Gemini’s ability to show sources, correct false assumptions, and avoid pretending that uncertain information is certain is an important part of the experience.</p><h3>The weakest point: uncertainty still feels under-designed</h3><p>The most interesting part of my study was how Gemini handles uncertainty. In some cases, Gemini did tell me when it was unsure. That is good.</p><p>But I noticed a gap.</p><p>Admitting uncertainty is not the same as designing for uncertainty. There is a difference between an AI saying “I’m not sure” and an AI helping the user understand “Here is what I know. Here is what I do not know. Here is how I can verify it. Here is what you should be careful about.”</p><p>That second version creates more confidence.</p><p>From my test, Gemini sometimes communicated uncertainty, but the experience did not always make me confident that it would seek clarification or resolve the uncertainty before continuing.</p><p>This matters because uncertainty is one of the core design problems in AI products. AI systems are <strong>probabilistic</strong>. They generate answers based on patterns, context, and available information. So uncertainty is not an edge case. It is part of the product experience.</p><p>A good AI interaction experience should not hide uncertainty, but it also should not leave the user with uncertainty alone.</p><p>It should help the user navigate it.</p><h3>Control and conversation architecture</h3><p>Gemini gives users a good amount of flexibility. You can interact through text, images, and other media. You can also redirect the conversation when the answer is going in the wrong direction. That makes the experience feel open and steerable.</p><p>This flexibility is important because AI conversations are rarely linear.</p><p>Users often begin with a rough idea, then refine it as the conversation develops. In that kind of interaction, control does not only mean clicking here and there It means the ability to correct, redirect, clarify, and reshape the output.</p><p>But as Gemini becomes more capable, control needs to become more visible too because when an AI product can search, reason, generate, summarize, and possibly take actions, users need to understand what the system is doing at each moment. Is it answering from memory? Is it searching? Is it using a tool? Is it making an assumption? Is it generating creatively? Is it grounding the response in external sources?</p><p>These differences matter because they affect how much the user should trust the output. The more powerful AI products become, <strong>the more important it becomes to make system behavior legible.</strong></p><h3>The bigger product design lesson</h3><p>Gemini reveals a larger challenge that many AI products will face.</p><p>As AI systems become more capable, the product design problem shifts from What can this AI do? to, Can users understand what it can do? Can they trust what it produces? Can they tell when it is uncertain? Can they correct it? Can they stay in control? Can they build the right mental model of the system?</p><p>For me, this is where AI product design becomes deeply connected to HCI.</p><p>The future of AI interfaces will not only be about adding more features or making models smarter. It will be about designing better relationships between humans and intelligent systems.</p><p>Gemini is interesting because it shows both the promise and the difficulty of becoming an “everything AI” product.</p><ol><li>Promise: one flexible interface for many kinds of tasks.</li><li>Challenge: the broader the product becomes, the harder the interface must work to make the system understandable, trustworthy, and steerable.</li></ol><p>That is the part I find most interesting. Not just what Gemini can do.</p><p>But what Gemini teaches us about how intelligent products should be designed.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=904097a99ea3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Constraints Aren’t Blockers; They’re What Make Good Design Possible]]></title>
            <link>https://graphlinkz.medium.com/constraints-arent-blockers-they-re-what-make-good-design-possible-0703a9778019?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/0703a9778019</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[ui-ux-design]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Fri, 23 Jan 2026 15:53:00 GMT</pubDate>
            <atom:updated>2026-01-23T15:53:00.602Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why the “simplest” solution is often the smartest, and how designers and engineers can find balance.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6qEkAgFfMgxE144Ii5BXnA.jpeg" /></figure><p>Not long ago, engineering constraints changed a UX flow I really liked. In my first year of design, I would have called that a compromise. I would have felt like I lost the argument.</p><p>Today, I see it differently. It’s not a failure; it’s the job.</p><p>As designers, we often work in a world of perfection. We look at the canvas and imagine what could be. But products don’t live in Figma; they live in code, in systems, and in long-term maintenance cycles.</p><h3>The Trio in Action</h3><p>I was recently working on an SSO (Single Sign-On) improvement initiative. I was part of a “Product Trio,” a structure I’ve come to value deeply, working alongside a thoughtful PM, and a brilliant engineer.</p><p>Our goal was simple: make the account selection process more intuitive.</p><p>I designed a state system that used colour-coding versus muted states to visually communicate account availability. From a pure UX lens, it was a win, a perfect design.</p><ul><li>It was clearer.</li><li>It was more expressive.</li><li>It was much easier to scan at a glance.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jLnRHEtlxGlJaxXiT-NH_A.jpeg" /></figure><p>I was ready to defend it. Then, engineering constraints showed up.</p><h3>The Cost of a “Small” Detail</h3><p>When the engineer looked at the flow, he didn’t see “expressive colours.” He saw logic.</p><p>That small visual distinction came with a massive hidden price tag:</p><ul><li><strong>Increased Complexity:</strong> The logic required to determine those states in real-time was non-trivial.</li><li><strong>Edge Cases:</strong> What happens if a state is unknown? What if the API lags?</li><li><strong>Long-term Maintenance:</strong> Every future update to the SSO would have to account for this custom state logic.</li></ul><p>From a design perspective, it was a 10% improvement in clarity. From an engineering perspective, it was a 50% increase in complexity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W4It6hFLY9Hklw4iSx0cBQ.png" /></figure><h3>Choosing the System over the Screen</h3><p>We didn’t ship the expressive version. We went with the simpler version where everything stayed coloured.</p><p>Was it the perfect UX I originally envisioned? Probably not.</p><p>Was it the right decision for the system, the scope, and the team? Absolutely.</p><p>Healthy collaboration isn’t about defending your idea until the engineer gives up. It’s about understanding the trade-offs well enough to move the product forward together.</p><h3>The Real Role of a Designer</h3><p>Constraints don’t block good designers; they shape better decisions.</p><p>If you can only design when you have unlimited dev resources and zero technical debt, you aren’t a product designer, you’re a visionary. Real product work happens in the mud of constraints.</p><p>It’s about asking: is this 10% improvement worth the 50% debt we are taking on?</p><p>Often, the answer is no. And being the one to say that’s okay is the ultimate sign of design maturity.</p><p>Where have you traded a “perfect” design for a “better” product recently?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0703a9778019" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I Killed My Favorite Feature. (And It Was the Best Decision I Made)]]></title>
            <link>https://graphlinkz.medium.com/i-killed-my-favorite-feature-and-it-was-the-best-decision-i-made-8ece26c06cdd?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/8ece26c06cdd</guid>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[ux-design]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Tue, 20 Jan 2026 16:43:13 GMT</pubDate>
            <atom:updated>2026-01-20T16:43:13.462Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>Early in my career, I thought good design meant defending your ideas. Experience taught me that good design is knowing when to delete them.</em></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hSfsoU81rovmhNHALmgVwg.png" /></figure><p>I changed a design I liked . Not because it was bad. But because it wasn’t the right call yet.</p><p>We are often taught that the job of a designer is to “fight for the user.” To advocate for the experience. To stand your ground against engineers who want to ship faster or PMs who want to cut scope.<br>And sure, sometimes that’s true.</p><p>But early in my career, I confused “fighting for the user” with “defending my ego.” I thought that if I backed down, I was failing.</p><p>Experience teaches you something else entirely:</p><ul><li>Knowing when to push.</li><li>And knowing when to step back for the team.</li></ul><p>This is a story about the latter.</p><h3>The “Perfect” Flow</h3><p>I’m currently working on a side project with some friends called <strong>Showup</strong>. It’s a mobility experiment.</p><p>We were building out the booking flow, and I designed a feature where users could see specific driver details while selecting their bus.</p><p>I didn’t just like it. I loved it. It felt reassuring. It felt thoughtful. It added that layer of “humanity” we always talk about in design critique.</p><p>In Figma, it was a winner. It answered the user question: <em>“Who is driving me?”<br></em>But my team pushed back.</p><h3>The Friction Trap</h3><p>They didn’t look at the aesthetics. They looked at the goal.<br>We were running an experiment. The goal was conversion. The goal was to get a user from Point A (Interest) to Point B (Booked) with zero friction.</p><p>Based on their experience with real-world usage, my “thoughtful” feature was actually a roadblock.</p><ul><li>More information.</li><li>More decisions.</li><li>More hesitation.</li></ul><p>In a vacuum, knowing the driver is nice. In a high-speed booking context, it’s a distraction. It forces the user to pause, evaluate, and compare.</p><p><em>Is a 4.8-star driver better than a 4.7-star driver? Does it matter for a 15-minute ride?</em></p><p>Suddenly, the user isn’t booking a seat. They are analyzing personnel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Paqc1siyEMG-aZSeMS268A.png" /></figure><h3>Protecting the Experiment</h3><p>This is the moment where the Designer Ego inside me wanted to argue. I wanted to bring up empathy. I wanted to talk about “delight.”</p><p>But I realized something crucial.</p><p>They weren’t attacking the design. They were protecting the experiment.</p><p>So I let go of my personal taste. Not because I was wrong. But because the goal wasn’t to “ship my idea.”</p><p>The goal was to run a cleaner test.</p><h3>The Permanence Fallacy</h3><p>There is a trap we fall into where we treat digital product design like architecture. We act as if once the code is shipped, it is poured in concrete, never to be changed again.</p><p>That fear makes us hoard features. If I don’t ship this now, it will never exist.</p><p>But here’s the part I find most interesting:</p><p><strong>Nothing is permanent.</strong></p><p>Software is soft. It is malleable.</p><p>If we ship the lean version now, and the data tells us later that users are anxious about who their driver is? We add it back. With evidence. With confidence.</p><h3>Making Room for Learning</h3><p>Good designers don’t cling to solutions. They cling to problems.</p><p>If the problem is “trust,” driver details are <em>a</em> solution, not <em>the</em> solution. If the problem is “speed,” driver details might actually be the anti-solution.</p><p>By removing the feature, I wasn’t lowering the quality of the design. I was increasing the quality of the signal we would get from the experiment.</p><p>So, I ask you: <strong>Where have you chosen judgment over preference recently?</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8ece26c06cdd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Crafting Your Unique Color Palette: A Designer’s Guide to HSB Manipulation]]></title>
            <link>https://graphlinkz.medium.com/crafting-your-unique-color-palette-a-designers-guide-to-hsb-manipulation-ead6d9e3cc70?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/ead6d9e3cc70</guid>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Tue, 12 Dec 2023 09:13:57 GMT</pubDate>
            <atom:updated>2023-12-12T09:13:57.516Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bIQrjnbf2g6pbkb2qXz_QA.png" /></figure><p>Do you ever wonder how you, as a designer, can construct a color palette that suits your vision and aligns seamlessly with the product you’re creating? I’m sure a lot of folks can relate to this or probably not. But anyway I want to share what I’ve learned with you. <em>(Credit goes to </em><a href="https://www.linkedin.com/in/zanderwhitehurst/"><em>Zander</em></a><em> from </em><a href="https://memorisely.com/"><em>memorisely</em></a><em>)</em></p><p>So as designers, we often find inspiration in the works of others right? why not apply the same principle to colors? The challenge I have faced is usually to not outrightly copy someone else’s color scheme entirely for my projects. But here’s the thing — you absolutely can (sort of) make use of someone else’s color scheme in the form of a reference and just change a single value on each color scale to get yours.</p><p>With the default use of hex code (#000000, #ff7000, and the like) it is almost impossible to guess what color should be like if you manipulate the values, and in my opinion, it has made people feel like they can’t control their color values to the best possible state in a near accurate manner.</p><blockquote><em>…drum rollssss 🥁</em></blockquote><p>That is where <strong>HSB</strong> comes in. HSB (Hue, Saturation, and Brightness) color picker is a powerful tool for manipulating colors. In a nutshell, it allows you to play with the saturation and brightness of your hue to achieve precisely the tone you’re after.</p><h3>Understanding HSB</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vZZYG4Joysm-rCEaYI_9Kg.png" /><figcaption>Saturation goes from left to right. Brightness goes from top to bottom</figcaption></figure><p>Before we dive into it properly, let’s quickly clarify what HSB is. In simple terms, it’s another way of selecting colors, similar to the hex code. However, with HSB, you can fine-tune the saturation and brightness of your hue to create the perfect color.</p><h3>And now for the process</h3><p>Let’s start by referencing a grey color scale. In the image below, we have a nice crafted grey scale.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PgvEEp3xFc6gfxL8KIw-ug.png" /><figcaption>A grey color scale for reference</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bdGKALpQg5iKbe2qrIytWw.png" /><figcaption>color picker with the same hue value but different saturation and brightness</figcaption></figure><p>If you observe closely, the hex codes differ, but the hue values remain the same. The only variations lie in the saturation and brightness (are you thinking what I’m thinking). The game-changing realization here is that by manipulating the hue value, you can morph the entire scale to fit your unique needs. Let me illustrate.</p><figure><img alt="Sample colors for a brand" src="https://cdn-images-1.medium.com/max/1024/1*Ha3o1JVhiMa4EZrn09QVJQ.png" /><figcaption>Sample colors for a brand</figcaption></figure><p>Imagine you’re working on a project with specific brand colors, and you want to use the first color to build neutral tones <em>(Yes, you don’t have to always use #000000 to build your greys or neutral tones)</em>. By switching from HEX to HSB on your color picker, you can identify the hue value — let’s say it’s “<strong>234.</strong>” Now, replace the hue value of the reference color scale, and voila! You’ve got a new, tailored greyscale based on your product’s brand colors.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YXQocGuLaGrutAFoZ9yfTQ.png" /><figcaption>Hue picked for the new color</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gnSjz992vtEvBbNOVeLPaQ.png" /></figure><p>Notice how the weight of the colors remains consistent because they share the same saturation and brightness. However, they look distinct because of their different hues.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Mkb1H_Fx-Ws9UeK35wGzSA.png" /><figcaption>New color scale with new hue</figcaption></figure><p>Now, putting it all together, you’ve effortlessly created a new color scale. This process is less tedious because all you have to do is replace the hue value, resulting in a fresh palette for your designs. This technique can be applied to semantic colors (success, warning, info, error), ensuring a consistent tone throughout your project.</p><h3>Next steps</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OCQpuaR37jHwkmUHxhx55A.png" /></figure><p>Incorporating these primitive grey colors into your variables is a fantastic way to kickstart the organization of your colors. This approach is beneficial when building your UI kit or a design system for the product you’re working on. Stay tuned for more insights on that in my next article!</p><p>Side note: <em>I might delve into UI kits and design systems in a future piece. Exciting stuff!</em></p><p>Feel free to practice changing your color scale with my scale value in this Figma community file: <a href="https://www.figma.com/community/file/1316182880970694750/color-scale-manipulation-with-hsb">https://www.figma.com/community/file/1316182880970694750/color-scale-manipulation-with-hsb</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ead6d9e3cc70" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Buzz of Web3.0 — how it differs from Web2.0]]></title>
            <link>https://graphlinkz.medium.com/buzz-of-web3-0-how-it-differs-from-web2-0-40a38f2bacee?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/40a38f2bacee</guid>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[block-game]]></category>
            <category><![CDATA[web3]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Tue, 01 Mar 2022 22:30:47 GMT</pubDate>
            <atom:updated>2022-03-01T22:36:26.565Z</atom:updated>
            <content:encoded><![CDATA[<h3>Buzz of Web3.0 — how it differs from Web2.0</h3><p>Before we go too far, we should probably get some things clear about what the web is in general.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Pug9HcURBNwo0Fz-LIlpzw.jpeg" /></figure><p>According to <a href="https://www.techopedia.com/definition/5613/web">Techopedia</a>, The web is commonly a way to refer to the World Wide Web(WWW), which is a part of what we know as the Internet which is made up of lots of pages that can be accessed through the use of Web browsers such as Google Chrome, Safari, etc. The web is often confused to be the Internet (which is wrong by the way) and has been used interchangeably.</p><p>Over time the Web has seen a series of development in how it functions and people interact with it. These various phases were what was tagged as Web1, Web2 and now we have Web3.</p><h4>Brief Timeline</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ORxdTEDlr7v9-x56-RNQNQ.jpeg" /></figure><p><strong>Web1.0:</strong> The form of the internet where you only consume content. it&#39;s just like reading from the newspaper. You take information in, and then take information in again, without contributing to the content shared.</p><p><strong>Web2.0:</strong> A big improvement from Web1.0. Here users can contribute to the content of the web. A simple example will be social media, where you can both create and consume content.</p><p><strong>Web3.0(the Elephant in the room):</strong> This is yet another improvement from Web2.0, but here users have more security and control over their content and data. In essence, there’s no middleman, no chairman, no boss. Web3.0 does not have a central authority like in Web2.0(e.g. Google, Facebook, etc, who handle and keep your data), the authority and control here are “Decentralized”. The core technologies of Web3.0 areBlockchain, Artificial Intelligence (AI), Augmented Reality (AR), and Virtual Reality (VR), users will be given more personalized content and experiences.</p><p><a href="https://www.huobilearn.com/en-us/what-is-the-difference-between-web2-and-web3/">-source: Huobo learn</a></p><p><em>The main concern of this article is to point out salient differences between Web3.0 and Web2.0</em></p><h3>Some ways Web3.0 differs from Web2.0</h3><ol><li><strong>Digital Identity</strong>: Unlike as it was with Web2.0, you will have more control over the privacy of your digital identity by using an alias and digital avatars that are only unique to you which is only possible through Web3.0 technologies.</li><li><strong>Personalization:</strong> With Web3.0, the Internet experience would become more customized to the users. Artificial intelligence, which selects the most relevant options for information (search engines are actively engaged in it, reducing the role of organic search results)</li><li><strong>Security</strong>: A new level of security and privacy through decentralization — freedom from censorship and surveillance due to the lack of a control center — distributed ledger, and decentralized finance (Defi)</li><li><strong>Data and Digital Asset Ownership: </strong>Through the use of blockchain technology, Web3 aims to move tech giants out of the way by placing complete ownership of data back to the users who own them. Companies like Google and Facebook store your personal information mostly to improve their target marketing. However, many have raised privacy concerns over such one-sided data governance and see Web3 as a solution. Blockchain technology is already being used to secure financial data in the crypto space, where transactions in the ledger are stored in a permanent and verifiable way. Access to the data can be collected back through really secured means, which is not controlled by the application used to create the data.</li></ol><p>There are still lots of ways Web3.0 differ from Web2.0 but these are a few points that can and I hope has helped in understanding the whole Web3.0 Buzz and how it differs from what has been used to(Web2.0)</p><blockquote>This article is actually a challenge😅 , The“Block Game” <a href="https://blockgames.gg/">Blockgame</a>s | <a href="https://nestcoin.com/">NestCoin</a> | <a href="https://zuri.team/">Zuri Team</a></blockquote><p><strong>References</strong></p><p>Techopedia: <a href="https://www.techopedia.com/definition/5613/web">https://www.techopedia.com/definition/5613/web</a></p><p>Huobi Global: <a href="https://www.huobilearn.com/en-us/what-is-the-difference-between-web2-and-web3/">https://www.huobilearn.com/en-us/what-is-the-difference-between-web2-and-web3/</a></p><p>The AppSolution: <a href="https://theappsolutions.com/blog/development/the-difference-between-web2-vs-web3/">https://theappsolutions.com/blog/development/the-difference-between-web2-vs-web3/</a></p><p>Internet with Brain: <a href="https://www.web3.lu/divide-the-web-timeline-in-nine-epochs/">https://www.web3.lu/divide-the-web-timeline-in-nine-epochs/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40a38f2bacee" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I hope people I know don’t see this.]]></title>
            <link>https://graphlinkz.medium.com/i-hope-people-i-know-dont-see-this-162a9e0459b1?source=rss-6b50ee707ee0------2</link>
            <guid isPermaLink="false">https://medium.com/p/162a9e0459b1</guid>
            <category><![CDATA[random-thoughts]]></category>
            <category><![CDATA[first-post]]></category>
            <dc:creator><![CDATA[Graphlinkz]]></dc:creator>
            <pubDate>Thu, 30 Dec 2021 14:55:52 GMT</pubDate>
            <atom:updated>2021-12-30T14:55:52.301Z</atom:updated>
            <content:encoded><![CDATA[<p>I’ve never published any article on medium and I just wanted to use this medium(no pun intended) to drop an article even if its just few seconds read before the year runs out.</p><p>To who ever sees this, be your own motivation and be the best you can be, just move at your own pace. It works.</p><p>Thanks for reading.</p><p>(Lol do they thank readers here😅)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=162a9e0459b1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>