T-minus 4 weeks to the public launch of my book Marketing Plan for Tech Startups!
As you can imagine, I’m in the middle of final edits, layout decisions, and preparing content for every format readers might want.
There will be a paperback. There will be a digital edition on Kindle.
But of course, that wasn’t enough for me. I also wanted my own e-reader. One that actually had the features I always wished an e-reader offered.
So I vibe-coded one with GitHub Spark.
Inspired by Priyanka Vergadia’s demo showing how she built a full-stack app in minutes with GitHub Spark, I gave it a try. Spark is GitHub’s new AI-powered app builder that runs entirely in your browser. No setup. No config. No need to remember Java classpath from my mobile and web app developer days. 😉
Just describe what you want, and Spark builds it end-to-end: front-end, database, authentication, etc. As always, Priyanka did an awesome job walking her YouTube channel viewers through all the steps of using GitHub Spark to go from zero-to-app, so I thought: why not?
The PRD aka my wish list for a book reader
I mostly wanted three things:
- A two-page view so if you read on a big screen it feels like an actual book in front of you.
- A search function so you can instantly jump to “positioning,” “pricing,” or “Anthropic case study.”
- Bookmarks and notes, so readers can mark sections and write down thoughts as they read (my paperback margins are always full of notes and post-its 😉)
Three features I dreamed up, let’s see what I got.
How GitHub Spark turned my PRD into a working e-reader
I typed my requirements in natural language, hit submit, and Spark went into “think mode.”
A few minutes later, I had a working prototype: two-page display.
A couple of hours later (and with a few vibe-coding-hacks I’ll detail below) I added keyword search and a bookmark system. Here’s the finished product:
My e-reader vibe-coded in an afternoon looks very promising but is not quote ready to ship just yet. Here’s why:
Lessons learned from vibe-coding an e-reader in GitHub Spark
First, while Spark gave me the basic app scaffolding quickly, it struggled to render a PDF heavy with graphics. Sometimes it showed only text, other times it spit out binary data.
Spark’s default PDF handling just wasn’t built for a manuscript like mine. My book isn’t a typical wall of text. I wrote it in Google Workspace Slides to make it as much a tool as a book, packed with frameworks, diagrams, and visuals that startup founders and marketers can apply right away. The format was deliberate: keep the text lean, rely on visuals, and use slides as a constraint so every word carries weight.
I knew from a previous vibe-coding session that v0 by Vercel could handle a heavyweight manuscript like mine, so I thought: why not ask Vercel how it did it? The answer was pdfjs-dist, the distributable version of Mozilla’s PDF.js, which renders PDFs natively in the browser without plugins. I plugged it into Spark and—yay—I was unblocked!
Second, as I layered on more prompts and features, I learned that Spark projects can hit limits and stop accepting prompts.
The first prototype was quick and easy, refining it took patience… and some help from ChatGPT. When Spark stopped accepting prompts, I pulled down the GitHub ZIP, then used ChatGPT to reverse-engineer Spark’s app architecture and rebuild the project with more detailed instructions.
This experience pretty much sums up today’s vibe-coding scene: vibe-hackers are out of the box thinkers who juggle multiple tools; when one doesn’t do what you need, you pick up another.
My final lesson: vibe-coding is a lightning-fast way to prototype and experiment but it still takes time to create a production-grade app ready to be shared with others. That’s why for now, I’m only sharing screenshots.
Just like with my “Slide Tools” hackathon two weeks ago, I was reminded of the real promise of AI-driven coding:
The future of software with AI: everyone can be a creator.
The next generation of apps — whether e-readers or enterprise apps — will be powered by AI, built faster than ever, and customizable to fit customer needs with precision.
And some of those apps will be built by marketers.
Marketers as vibe-coders
“Vibe Marketers” are already starting to appear on job boards:
“We’re looking for a Vibe Growth Marketing Manager who is a builder who prototypes and ships faster than most teams can spec a brief. You’ll use AI tools, LLMs, no-code/low-code platforms, and smart automation to rapidly unlock new growth channels, improve operational efficiency, and experiment with new marketing ideas end-to-end.”
It’s clear that vibe-coding is becoming essential for speed and efficiency in marketing workflows.
But why stop at workflows? What if marketers could also be the first prototypers of new product ideas?
Marketers as product prototypers
Marketers are already customer advocates and trend spotters. Vibe-coding tools now give them the ability to turn insights directly into working prototypes, bridging the gap between customer voice and product innovation.
With vibe-coding, marketers can also extend existing products with new features requested by their customers, as I demonstrated in my “Slide Tools” hackathon.
A sneak peek into my book’s vision
Elevating marketers into co-creators of product is central to my book’s vision. My goal is to restore marketing to Kotler’s full “4 Ps” (product, price, place, promotion), rather than the narrow “1 P” of promotion it’s often reduced to. Vibe-coding tools may be the superpower that helps marketers reclaim all four.
If you’re a startup founder or marketing leader, my upcoming book Marketing Plan for Tech Startups project distills lessons from Fortune 500 companies and startups into practical frameworks to break through the noise and turn innovation into revenue.
I’m also thrilled to share that the one and only Priyanka Vergadia is among its distinguished contributors! 😀
Reserve your copy here:
https://marketingwithjustyna.gumroad.com/l/MARKETINGPLAN
PS: If you want to try Spark yourself, watch Priyanka’s excellent demo:
Tech Surprises 