The Chronicle Build
The Setup
I used to build websites for fun since the WordPress days of early 2010. Then came Webflow and its humorous tutorial videos. I was hooked. I took Awwward-winning web UX/UI courses to learn about interactive design. The last one I built was in 2021 when my daughter was born. My wife and I had shift rotations to get uninterrupted sleep as a survival tactic, and I built her announcement website while on duty at 2am bouncing on a yoga ball with my daughter strapped to me in a baby k’tan wrap (I loved those things).
Those builds took available sleeping times. A personal project might eat a full weekend of four or five-hour late-night sessions and leave me running on fumes the following week. A bigger build for the nonprofit I co-founded could stretch across a month of stolen evenings.
My web builds nowadays are burn report dashboards at work. Not exactly the creative outlet of my choosing.
Then my direct report’s last day was coming up and I was met with a conundrum:
Force a virtual farewell that she said she didn’t want and do the only awkward thing most remote teams know
or…
Build something she’d actually want to keep.
My Approach + AI Role
The farewells I remember weren’t about what I’d built. They were about being known. Not the clever SmartSheet data mesh I’d set up, not the perfectly architected multi-market program estimate spreadsheet with pivot tables and charts beautifully designed with sharp insights available. 😏
It’s how I’ve made others feel.
She’s a climber (the boulder ones, not the rope ones, I’ve found). She played Dungeons and Dragons.
I think I’ve got my starting point, though I’ve never played D&D.
I knew what to do this time because of the knowledge system approach I have with AI. The first thing I built was a design system. Not the artifact. Not the story. The system.
I have a library of design knowledge files… a typography codex, a layout codex, a color codex, a motion codex, a web-component-architecture skill… They encode domain-specific decisions so I don’t fight them downstream.
Pass 1 of this build was turning what I had into a fantasy-sourcebook-codex with a type system, a parchment palette, rules for every decorative element, the structure of a character sheet, and a measurement system where changing one number resized everything else automatically.
None of this had anything to do with the person yet. That was the point. I front-loaded every design calculation so the sessions that followed could spend themselves entirely on her.
My stack: personal AI app for the narrative, the HTML architecture, and the design system. Midjourney for skipping hours of searching and comping graphics in Photoshop. Two distinct apps (one doubles with a separate design interface), each doing what it does best, connected by knowledge files between them.
What Actually Happened
Iterative builds across roughly 2 days. Bulk of it done day 1 after the kids went to bed for about 3 hours. Second day was finalizing and polishing in between meetings leading up to my last one-on-one with her.
The build itself felt different from anything I’d done before. Fourteen HTML versions. A 2MB single-file document with 30 scenes, 26 embedded assets, 11 fonts, all encoded so the whole experience is one file you can open anywhere. Chapter breaks with oil-painted plates I generated in Midjourney.
This is how building this website felt compared to the old way: I could focus on describing what I wanted in the language I would have built it in. The vocabulary was the same. The grammar was the same. But instead of hand-coding every interaction and hunting for hard-coded values I’d accidentally created, the design system caught it. A type-size change was one edit and everything downstream updated. The entire parchment color palette was five reusable rules. When I wanted the gold accent restricted to three uses across the whole document, I could state that as a design decision and track it, not hunt for where the HTML or CSS code was.
The old me would have been three hours deep, realized I’d hard-coded a hex color in nine places, broken my creative focus to go find them all, feeling frustrated, and questioning why I was building the thing.
This time, I was orchestrating. Like a conductor.
I’m still learning the boundary, though. I threw too much at the model in single passes. Concept, design, tone, and content all at once. This specific build taught me that framing comes first, then dedicated lenses to iterate. You can’t ask one prompt to get everything right simultaneously in this kind of work, no matter how well made the knowledge files.
That’s the tactical thing I’d give away to anyone doing this.
I got on our last one-on-one and narrated the whole thing live to her. Walked her through the chronicle for about 20 minutes. I could tell she was touched. Not many people get a last one-on-one with their departing manager that’s 20 minutes of someone saying, in the most dramatic reading voice way they can, I noticed what you did and it mattered to many people.
Then I sent the firm-wide thank you note email with the PDF and the website attached.
Shortly after, a senior colleague messaged our informal group chat: “When I leave, I want you to make me a chronicle.”
The Real Insight
The design system didn’t make the creative work faster. I still spent maybe 6 hours total between prep, research, and iterative builds.
But because I’d been collecting and assembling knowledge systems for months, the creative energy went to the right places.
The old builder me would spend half the build on infrastructure that had nothing to do with the person. Hunting hex values, creating CSS classes I should have made earlier, breaking creative focus to fix structural debt I’d accumulated by moving too fast (and watching youtube videos on how to fix them). Figuring out in 5% increments the parallax speeds of textures and animated interactions. The time was comparable to a weekend Webflow build. What changed was where my attention went.
Pass 1: design system. Pass 2: the narrative. Passes 3 through 7: the visual experience, the iteration, the visuals, the emotional pacing. The human material… teammate quotes, my own observations captured over months, the detail about the scythe that became a structural interpretive key for her entire character… that’s where the hours went. Not pixels. Not infrastructure. The person.
The old builds were creatively satisfying, but they were primarily about the mechanics. This one was primarily about meaning. Same hours. Same late night (the design knowledge session ended around 12:30am, some things don’t change).
Try This If…
If you have something to build and a system of reusable design decisions available to you, try separating the build into three passes and doing them in order. Design system first, with no content. Content second, with no design. Integration third.
The front-loading feels unnatural. You want to start with the thing. But every design decision you make before the human material enters is a decision you won’t have to make while you’re trying to hold an emotional story. And when you’re building something for a person, the emotional storyline is the whole point.
Two specific things to watch for.
First: the model will fabricate to fill gaps. It will invent plausible details that sound right and feel kind. Those will be the fastest things to ring false, because they’ll be the only claims not traceable to someone who was actually there. Always check its work against what you know to be true. Pressure test it.
Second: you will want to keep designing past the point where the artifact is done. I was fourteen versions deep and heading toward a from-scratch rebuild of the PDF version when the right approach was a browser conversion that took a minute.
The builder instinct at the end is to keep building.
Shipping is a different muscle.
Systems Lens
The colleague’s message was the feedback I needed. She didn’t want a D&D chronicle. She wanted what the chronicle represented: something intimate, curated, and built by hand for the person receiving it. Something that said I noticed you specifically.
That’s the promise of having a system underneath creative work. Not that the system makes you creative. But that it makes creative generosity repeatable. The design knowledge file I built for this project is still in my library. The measurement system still works. The next time another idea strikes, the design layer is solved. And I get to focus on the more meaningful aspects of what sparked my idea to begin with. Sharing an experience.
Watching someone feel seen by the words their teammates wrote about them, appearing on something you built with your hands and the systems you’ve been quietly assembling… that’s the part I’d build for again.