<?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 Karyll on Medium]]></title>
        <description><![CDATA[Stories by Karyll on Medium]]></description>
        <link>https://medium.com/@kjcatz?source=rss-12512a2d6086------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*o6PE8d3_n9UUN0b4HUgIIg.png</url>
            <title>Stories by Karyll on Medium</title>
            <link>https://medium.com/@kjcatz?source=rss-12512a2d6086------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 21:43:53 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kjcatz/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[I’m Just a Designer. I Built a Story Engine Anyway.]]></title>
            <link>https://medium.com/@kjcatz/im-just-a-designer-i-built-a-story-engine-anyway-cb7a49e89819?source=rss-12512a2d6086------2</link>
            <guid isPermaLink="false">https://medium.com/p/cb7a49e89819</guid>
            <category><![CDATA[storytelling]]></category>
            <category><![CDATA[writing]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Karyll]]></dc:creator>
            <pubDate>Thu, 16 Apr 2026 22:49:03 GMT</pubDate>
            <atom:updated>2026-04-16T22:49:03.115Z</atom:updated>
            <content:encoded><![CDATA[<p><em>What AI fiction is getting architecturally wrong, and what I learned breaking it apart to find out.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G1MVIjGXfi37e6KT4aRayg.png" /></figure><p>In my last piece, I wrote about how I shipped Qento alone. No team, no engineering background, no one to hand off to. Just a designer who kept pulling on a thread until something real came out of it. You can read it <a href="https://medium.com/@kjcatz/i-shipped-a-bug-that-let-users-get-my-app-for-free-heres-what-i-missed-c7eb524bd603">here</a>.</p><p>I ended that piece with a sentence I’ve been sitting with ever since: <em>a great story isn’t just generated, it’s engineered.</em></p><p>Because I said it before I fully understood it. I knew it was true the way you know a layout is working before you can explain why. But the more I’ve dug into what makes AI-generated fiction fail <em>(and it fails constantly, in the same ways)</em> the more I’ve realized that sentence was pointing at a problem most people building in this space haven’t confronted yet.</p><p>The problem with AI fiction isn’t the writing. It never was.</p><h3>Plot is not story</h3><p>There’s a distinction that every novelist internalizes early but most AI systems never encounter: plot is what happens. Story is how it <em>feels</em> to watch it happen.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I0E9AYYMzbeNgmxVUB9YvA.png" /></figure><p><em>Plot is: </em><br>the character loses their job, meets a stranger, falls in love, gets betrayed, finds redemption.</p><p><em>Story is: </em><br>the specific texture of grief when the thing you lose was already halfway gone. The way hope feels dangerous to someone who’s been disappointed enough times. The particular silence after betrayal; not dramatic, just quiet and final.</p><p>You can generate plot with a list of events. Story requires something else. It requires that the system <em>understand what things mean to the person they’re happening to.</em> And that’s an entirely different design problem.</p><p>Most AI fiction engines are event generators. They’re good at sequence. They’re poor at significance.</p><h3>Characters forget who they are</h3><p>In most AI-generated fiction, characters are described at the beginning and then gradually forgotten. They have a bio <em>(introverted, ambitious, distrustful of authority)</em> and this information gets referenced when convenient and ignored when inconvenient.</p><p>Real characters aren’t a list of traits. They’re a <em>pattern of responses.</em> An introverted character doesn’t just appear in scenes labeled “quiet” — they experience every scene through the lens of someone who finds the world slightly too loud. Their introversion shapes what they notice, what they avoid saying, what costs them energy and what restores it.</p><p>When you lose track of that lens, the character becomes a placeholder. Their name is there. Their function in the plot is there. But the <em>person</em> has left.</p><p>This happens in AI fiction constantly. Not because the model can’t write introverted characters; it can, when asked. But because the system isn’t designed to maintain psychological <em>continuity</em> across scenes. It maintains narrative continuity (<em>what happened, but not emotional continuity) </em>what this costs someone, what they’re carrying, how they’ve changed.</p><p>The result is a character who resets. And a story you stop caring about.</p><h3>Tension is a pressure system, not a dial</h3><p>Tension isn’t a setting you turn up before a big moment. It’s a pressure that has to accumulate somewhere before it can release anywhere.</p><p>If you’ve ever read a thriller where you knew the character was in danger but didn’t feel it in your chest then you’ve experienced what it looks like when tension is delivered to the reader instead of built inside them. The reader is told to care. They don’t.</p><p>Tension builds through <em>restraint</em>, not escalation. The scene where nothing bad happens but you know something’s wrong. The conversation where everything said is true and everything unsaid is truer. The moment a character almost asks the question they need to ask, and then doesn’t.</p><p>Pacing is the art of knowing what to withhold and when. And this is almost entirely absent from how AI narrative systems are designed, because withholding is harder to optimize for than producing. The path of least resistance for a generative system is to give the reader the next thing. The craft move is often to not.</p><h3>What print taught me</h3><p>I have a background in graphic design before product design. And one of the things that transfers <em>more than I expected</em> is the principle of hierarchy through restraint.</p><p>In editorial design, you don’t create emphasis by making everything louder. You create it by making most things quieter. The pull quote stands out because everything around it recedes. The white space isn’t empty. It’s doing active work, telling the reader’s eye where to rest and where to move.</p><p>Good fiction works the same way. Every scene can’t be the most important scene. Every line of dialogue can’t carry maximum weight. The moments that land do so because the moments before them were calibrated to let them land.</p><p>This is a design problem. Not just a writing problem.</p><p>When we started building Qento’s story engine, this was the question we kept returning to: <em>what would it mean to design for pacing rather than for output?</em> To build something that understood not just what to generate next, but what to hold back, and why?</p><h3>I had no idea what I was building</h3><p><em>I should say something honest here.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TeHjyjUP87-V_QRnczwCvg.png" /></figure><p>When I started working on the story engine, I didn’t have a framework for any of this. I’m a product designer. I wasn’t trained in narrative theory or system architecture. I hadn’t read the canonical books on story structure. I just knew <em>(the way you know when a typeface is wrong, or when a layout has no clear hierarchy)</em> that something about AI-generated fiction <em>wasn’t working</em>.</p><p>So I started pulling on that thread.</p><p>The structure didn’t come from a plan. It came from refusing to accept vague answers to specific questions. Why does this character feel hollow here? Not “because AI” specifically, <em>mechanically</em>, why? What information would the system need to hold for this moment to land differently? What’s missing?</p><p>Most of what I know about narrative architecture I learned by breaking it. Writing test stories and identifying exactly where they fell apart. Reading craft books not as a student of writing but as a designer reverse-engineering a system. Having long conversations with readers about what made a story feel real to them; not the plot, not the prose, but the <em>feeling of being inside something coherent</em>.</p><p>I didn’t come in with a blueprint. I came in with a question and a tolerance for not knowing the answer yet.</p><p>What I’ve learned is that this is actually an advantage. Designers are trained to work from the outside in: from the felt experience backward toward the structure that creates it. That’s exactly the right direction for this problem. The reader doesn’t experience the architecture. They experience the story. So that’s where you start.</p><h3>The invisible architecture</h3><p>The best story engines are the ones you never notice. Not because they’re simple, but because when they’re working, all you feel is the story.</p><p>This is what we’re building toward. A system where the mechanical decisions happen underneath the surface — when to escalate, when to let a scene breathe, when a character needs pressure and when they need release. Invisible to the reader. What they experience is a story that feels like it was written with intention. Because it was.</p><p>We’re not there yet. Building something that genuinely understands emotional continuity <em>(that holds a character’s psychological texture across an entire arc)</em> is hard. Building something that knows the difference between a scene that’s slow because it’s building and one that’s just… slow, is harder. We’ve gotten some of it right. We’ve gotten some of it wrong and had to rebuild.</p><p>But the question is the right one. And I think it’s the question the whole field needs to start asking more seriously.</p><h3>The question everyone is avoiding</h3><p>The models are going to keep getting better. That part is settled. The gap between human and machine prose will keep shrinking until, eventually, it closes.</p><p>When it does, what separates a story worth reading from one you abandon won’t be sentence quality. It’ll be whether the story understands what it’s doing; whether there’s <em>intention</em> underneath the language. Whether it knows that this character is carrying something specific, and that everything in this scene should be shaped by that weight.</p><p>That’s an architecture problem. It’s a design problem.</p><p>And it turns out you don’t need an engineering background to work on it. You just need to be the kind of person who can’t leave a broken thing alone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*txNRKMMZXd_qkLteyoDODA.png" /></figure><p><em>If you haven’t read the first piece (on how I shipped Qento alone as a designer with no engineering background) </em><a href="https://medium.com/@kjcatz/i-shipped-an-app-alone-as-a-designer-no-team-no-handoff-heres-how-004daea1a586"><em>it’s here.</em></a><em> Qento is a story platform built around the idea that AI narrative should feel like craft, not generation. Check it out on the </em><a href="https://apps.apple.com/us/app/qento/id6761312674"><em>app store</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cb7a49e89819" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I Shipped a Bug That Let Users Get My App for Free. Here’s What I Missed.]]></title>
            <link>https://medium.com/@kjcatz/i-shipped-a-bug-that-let-users-get-my-app-for-free-heres-what-i-missed-c7eb524bd603?source=rss-12512a2d6086------2</link>
            <guid isPermaLink="false">https://medium.com/p/c7eb524bd603</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[indie-hacking]]></category>
            <category><![CDATA[app-development]]></category>
            <category><![CDATA[ux-design]]></category>
            <dc:creator><![CDATA[Karyll]]></dc:creator>
            <pubDate>Wed, 15 Apr 2026 18:17:20 GMT</pubDate>
            <atom:updated>2026-04-15T18:17:20.036Z</atom:updated>
            <content:encoded><![CDATA[<p><em>If you missed part one — </em><a href="https://medium.com/@kjcatz/i-shipped-an-app-alone-as-a-designer-no-team-no-handoff-heres-how-004daea1a586"><em>I shipped a full iOS app alone in a month, as a designer, with no handoff.</em></a><em> Start there.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9wIye10tOBCvMCyFYzHNYg.png" /><figcaption>The horrifying text message.</figcaption></figure><p>Qento has been live for less than a week.</p><p>I built Qento alone. No QA team. No second pair of eyes. Just me, Claude, and a lot of late nights convincing myself shipping fast was the right call.</p><p>It was. Until it wasn’t.</p><p>Here’s what was happening.</p><p>Qento gives free users one story. That’s the free trial: one full story, up to chapter 5, no credit card. After that, you subscribe, or pay to make that one story <em>endless</em>.</p><p>Except there was a backdoor I didn’t know I built.</p><p>A user could start their free story. Read it. Hit the paywall at chapter 5. Delete the story. And then start a brand new one for free.</p><p>Repeat forever.</p><p>The code I wrote to protect the paywall was checking whether a story <em>currently existed, </em>not whether one had <em>ever been created</em>. Delete the evidence, reset the counter. It was the digital equivalent of eating your meal and disputing the charge.</p><p>I’d written a bug that actively worked against me.</p><p>There was a second one.</p><p>Guests, users who hadn’t signed in yet, couldn’t create stories at all. Instead of seeing the app, they saw a subscription paywall the moment they tapped “+”.</p><p>I’d assumed you needed an account to do anything. So I blocked guests at the door and sent them straight to the payment screen.</p><p>A first-time user opening the app, tapping the big obvious button, and immediately seeing “subscribe for $9.99/month.”</p><p>No story. No context. No reason to stay.</p><p>The fix took a few hours. The embarrassment lasted longer.</p><p>I patched both; deleted stories now count toward the limit forever, and guests get their one free story before we ever ask them for anything.</p><p>But the real thing I fixed was my mental model.</p><p>I’d been thinking about the paywall as a wall. Something to enforce. I wasn’t thinking about it as a promise; <em>you get this much for free, and here’s what comes next.</em> Those are different designs.</p><p>The wall version blocks people. The promise version earns them.</p><p>I shipped version 1.1.2 the same day I found the bug.</p><p>No team. No handoff. Just me catching my own mistake before it cost more than it already had.</p><p>That’s the part they don’t tell you about building alone. You’re not just the designer. You’re the QA. The security audit. The person who has to read their own code like a stranger would.</p><p>It’s humbling. And honestly, it’s made me a better builder.</p><p><em>Follow for more. Next up: the hidden structure I built so Qento’s AI actually writes like a novelist instead of a chatbot.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c7eb524bd603" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I Shipped an App Alone. As a Designer. No Team. No Handoff. Here’s How.]]></title>
            <link>https://medium.com/@kjcatz/i-shipped-an-app-alone-as-a-designer-no-team-no-handoff-heres-how-004daea1a586?source=rss-12512a2d6086------2</link>
            <guid isPermaLink="false">https://medium.com/p/004daea1a586</guid>
            <category><![CDATA[app-development]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[indie-development]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Karyll]]></dc:creator>
            <pubDate>Mon, 13 Apr 2026 20:58:35 GMT</pubDate>
            <atom:updated>2026-04-13T20:58:35.570Z</atom:updated>
            <content:encoded><![CDATA[<p><em>I didn’t learn to code. I learned to think like a builder. That distinction matters more than you think.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gQR1sUIinymRVilv_wi3jw.png" /></figure><h3>The Idea That Wouldn’t Let Me Sleep</h3><p>It started the way most dangerous ideas do. At midnight, annoyed.</p><p>I’m a Product Designer. I’ve spent years turning other people’s ideas into pixels, writing briefs no one fully read, sitting in meetings where someone would say “that’s a great creative direction” and then absolutely nothing would happen. I was good at the job. I just kept wondering what would happen if I stopped waiting for permission to ship <em>my own</em> thing.</p><p>The idea that kept nagging at me: <em>What if I made an app that felt as fun to write in as my favorite games feel to play?</em> A storytelling experience where you’re the protagonist (not a chatbot, not another AI novelty) a real reading experience with emotional arcs, chapter gates, and a narrative that actually remembers your choices.</p><p>I knew how to design it. I didn’t know how to build it.</p><p>That used to be where the story ended. This time, it wasn’t.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n9djkTx3-rMA28X3B-clAQ.png" /><figcaption>The very first prompt entered in Perplexity.</figcaption></figure><h3>Step One: Stop Guessing. Start Interrogating.</h3><p>Before I opened Figma, I opened Perplexity.</p><p>Not to look up “how to build an iOS app.” I used it as a thinking partner to stress-test the idea itself. I threw the messy concept at it and said, <em>okay, help me make this real.</em></p><p>It didn’t give me a pitch deck. It gave me signal. Real data. Competing perspectives. The back-and-forth turned a vague itch into a full product brief: positioning, flows, interaction principles, even the edges of the brand.</p><p>I walked away with a brief so tight I could’ve walked into any meeting and defended every decision. There was no meeting. There was just me. That was the point.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_gQtz536jF62rQUsb8eJsA.png" /><figcaption>Preview of Qento v1 Figma file.</figcaption></figure><h3>Step Two: Build It Before You Build It</h3><p>Next, I went to Figma Make. I already live in Figma (it’s where I think). But Figma Make let me do something I’d never done before: test an interactive prototype that <em>felt</em> like the real thing before a single line of production code existed.</p><p>I designed the story setup flow: genre picker, protagonist name generator, archetype selection. I hooked up a Claude API to the prototype for the story generation to feel realistic. I tested the reading experience with real content. I sent it to around ten people and watched.</p><p>Two of them asked me where they could download it.</p><p>That was the moment I knew the idea was real. Not <em>viable, real</em>. There’s a difference. Viable is a spreadsheet. Real is someone trying to use a prototype like it’s a finished product because they forgot it wasn’t.</p><p>By the time I was ready to build, I wasn’t guessing at the experience anymore. I’d already lived it a hundred times in Figma. I mocked, iterated, and broke things on purpose. I watched where people hesitated, where they smiled, where they got lost, and then cleaned up the MVP until it felt genuinely fun.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fwKSaGBb_CnbQCcWsQvQSA.jpeg" /><figcaption>The MVP experience.</figcaption></figure><h3>Step Three: The Part That Was Supposed to Stop Me</h3><p>Here’s what I’ll tell you plainly: I am not a developer. I reached out to developer friends to see if they’d want to partner. Everyone’s busy. But I believed in the idea. So I looked into Claude.</p><p>I understand systems. I understand how apps are architected: auth flows, state management, API calls, navigation patterns. I’ve sat next to engineers for eleven years. But writing Swift? Configuring StoreKit? Integrating Firebase Auth, Firestore, Analytics, and a live Anthropic API proxy through Cloud Functions? CLI? Terminal?</p><p>Not in my job description.</p><p>I opened Claude Code anyway.</p><p>What followed was the most productive month of my creative life. You read that right, one month of building.</p><p>It felt less like “learning to code from scratch” and more like pair-programming with an infinite senior engineer who never gets tired of refactors. I described what Qento needed to do; it built the architecture. Screen by screen, flow by flow, the thing assembled itself: logic, state, all the invisible machinery behind the UI. I designed the plan.</p><p>Shipping Qento taught me something I didn’t expect: a great story isn’t just generated, it’s engineered.</p><p>We built what I think of as a craft engine: a set of narrative laws that govern how every paragraph is constructed. Rules most readers will never consciously notice, but will always <em>feel</em>. The difference between a sentence that lands and one that doesn’t. Beneath the surface, every story runs on what I call a Kinetic Narrative Language — a structured system that tracks tension, pacing, subtext, and emotional restraint at the paragraph level. Your protagonist has a Soul: a psychological profile built on wound, armor, and drive that evolves based on the choices you make. I designed the logic.</p><p>Then there was the backend. Authentication layers. Server-side rate limiting. Subscription enforcement. All of it sitting invisibly between the user and the story. Infrastructure I never thought I’d understand, let alone build myself.</p><p>What started as “I want to make a story app” became a full-stack education in narrative design, distributed systems, and the surprising overlap between the two. Turns out the principles aren’t that different: both are about controlling tension, managing state, and knowing exactly when to give the user what they want, and when to make them wait for it.</p><p><em>I shipped it. Fully signed, archived, submitted, and reviewed.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TgwVDCd20A6zuMuIBVURSQ.png" /><figcaption><em>Qento is available on the App Store. </em><a href="https://apps.apple.com/us/app/qento/id6761312674"><em>Click Here to D</em>ownload</a></figcaption></figure><h3>The Thing Nobody Told Me About Building</h3><p>Everyone talks about the gap between designers and developers like it’s about skill.</p><p>It’s not. It’s about permission.</p><p>Designers are trained to hand things off. To present, defend, and then wait. The build is someone else’s territory. Architecture decisions are above your pay grade. Infrastructure is not your concern.</p><p>That’s not a natural law. It’s just how the industry decided to organize itself.</p><p>AI didn’t just give me new tools. It dissolved the handoff. There’s no longer a moment where my contribution ends and someone else’s begins. The idea, the design, the logic, the code, the submission: it was one continuous act of creation. And that changes everything about what a single person can build.</p><p>Creatives are no longer limited to moodboards, decks, and pretty Figma files. We don’t have to stop at the concept. We can actually ship the thing we dreamed up.</p><h3>I’m Not Just a Product Designer Anymore</h3><p>I still think in systems. I still care obsessively about how something <em>feels</em> to use. I still sketch before I ship.</p><p>But “Product Designer” no longer fits.</p><p>A designer makes things that need someone else to exist. What I built doesn’t need anyone else. It has a backend. It has IAP. It has a widget that sits on your lock screen and asks <em>what happens next?</em> every time you look at your phone.</p><p>I am a Creative Builder.</p><p>And if you’re a designer, an illustrator, a writer, a filmmaker, a creative, or anyone who has spent their career generating ideas that other people execute, I want you to hear this clearly:</p><p>The execution gap is closing.</p><p>You don’t need to become an engineer. You just need to stop believing that building is someone else’s job. The tools exist. The only thing left is the idea. And this is where we shine as creatives.</p><p>You already have that part.</p><p><em>Follow for more! Up Next - I’ll be talking about the hidden structure behind how Qento crafts stories and builds worlds that feel immersive.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=004daea1a586" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>