Challenge

Appley Rotten was a character brand I created over twenty years ago, with a rich history of media and products. In the last few years, the property had gone dormant, with no retail or online presence to speak of. I wanted to bring the character back online and deliver new content and experiences for modern audiences, but lacked the time and resources to fully develop the experience as I had in previous years.

With the emergence of powerful AI tools, I saw a path to moving quickly to utilize existing assets and build a clean, scalable platform for a relaunch.

History

Online development

Appley Rotten began as a one-off Flash animation on my private website in 2004, featuring a cute character with a large, apple-shaped head and a bad attitude. The character proved popular, leading me to form a character goods company with business partners later that year — expanding it to a full brand with a dedicated Flash website featuring animation, games, digital downloads and product previews. As Flash became unviable, we migrated to the TypePad blogging platform, where I created a weekly webcomic that ran for years and comprised almost 200 individual comics.

Product development

Starting as simple graphic t-shirts and accessories sold in local San Francisco boutiques and street fairs, the product line evolved to hand-made dolls and one-of-a-kind gallery pieces, and finally to a successful series of on-demand art prints with richly textured artwork and diverse thematic content.

Current state

In 2020, the global pandemic brought an abrupt end to in-person sales events, and TypePad shuttered shortly thereafter — taking all of the online content for the brand offline. For the first time in almost two decades, Appley Rotten was barely visible online. In 2026, I decided to rebuild the brand using modern AI tools to quickly build and test a bespoke interactive experience drawing on the rich history of the character.

Discovery

Comics had proven to be a strong traffic driver for Appley Rotten and represented the deepest well of content for a new online experience. After losing the blog, I explored migrating to WordPress or a modern comics platform like WEBTOON, but both meant working within a standardized experience with dependency on a third party.

With modern AI tools, I determined I could build a uniquely independent experience from scratch. I provided the same prompt to the three leading candidates — ChatGPT, Google Gemini, and Claude — using a folder of a few hundred comic image files and reference materials to create a prototype for a fully branded, responsive image viewer.

Key findings:

• While all models produced serviceable first prototypes, Gemini hit closest to the target functionality and presentation
• Gemini's image and video capabilities made it the primary tool throughout implementation
• Claude was generally better at suggesting and revising written text; OpenAI's Sora produced slightly more believable character acting in video

Design

The core experience

Working directly from the HTML prototype provided by Gemini, I refined key functionality and presentation through a short series of prompts. From previous iterations of the site and competitive samples, I knew exactly what I wanted to build and proceeded directly into code without creating any static mocks or flows. Most assets and branding had previously been designed, so it was simply a matter of folding them into the code. The use of AI greatly increased speed and allowed me to focus on what I was building instead of how to build it — reaching a functional, polished product in hours instead of days.

Key features:

• Streamlined controls: Users can navigate sequentially or choose specific comics, and zoom into panels to appreciate the small details
• Branding: All fonts, colors, and brand assets were applied cleanly without any dilution or modification
• Animation: Clear transition cues and a randomized breathing background added a layer of polish often cut for time in other projects
• HD content: Time savings in implementation allowed me to return to source files for hundreds of comics and update them for modern displays

Added functionality

The site was ready to launch in a comparable state to most comic sites, but I wanted to go further — creating a truly unique experience that extended the playful, off-kilter humor of the comic to every aspect of the interaction. I prototyped and tested a number of interactions with AI assistance, developing the most successful ideas into full features.

Key features:

• Achievements: A mysterious counter tracks secrets discovered across multiple visits
• Interactive game: A simple click-based game slowly reveals itself, offering rewards and increasing challenge
• Unlockable bonus comics and videos: Users unlock and view bonus content seamlessly within the existing site
• Hidden UI features: Additional items around the site trigger unique animations and jokes
• Variable outcome comics: Certain comics have alternate outcomes, presenting a different experience between users — and opening the door to a user-guided comic experience in future installments
• Safe mode toggle: A site-wide switch replaces adult comics with alternate versions, effectively doubling the content and adding a surreal, awkward tone throughout

New content

On the creative content side, I continued to draw all comics and brand artwork using traditional tools for digital illustration, updating hundreds of older comics to HD and creating new art in Adobe Illustrator and Photoshop. The one exception was Appley's first comic back after the long gap — where it seemed timely to poke fun at the proliferation of AI art online with an obviously cheap AI-generated panel.

As both a creator and consumer of commercial art, I share common concerns about using AI for creative work. I believe the worst outcomes occur when AI is used to mislead viewers or misrepresent the creative process — so I committed to clearly labeling and properly attributing any AI-created content within the media itself or on the hosting platform.

This led to studies with production work from old product photography, including a 3D spinning model of a papercraft toy and a video test for Instagram Reels to promote the character to new audiences. The reel hit almost 2,000 impressions in a single day, prompting me to explore a longer-format video using AI.

I approached the video from a real source: before Appley had ever appeared in comics, she existed as a physical doll, often alongside my cat Minou. From these references, I generated a "puppet" and stop-motion stage setting with Google Nano Banana 2 as part of pre-production. I then drew illustrations and translated them into rendered keyframes for a series of escalating vignettes between the tiny doll and a real cat, tentatively titled "The Nine Lives of Appley Rotten."

As I rendered dozens of short clips over a month, the AI frequently struggled with subtle character acting and more outlandish scenarios — but one vignette was starting to work. I reduced scope and edited the footage in Adobe Premiere into a focused animated short: "Appley Knows Kung Fu," where the small doll throws a strong punch at the bewildered cat and strides away in triumph.

Key features:

• The finished video runs nearly a minute with full audio and animation
• Failed animation attempts were compiled into an “outtakes” video — an unexpected side benefit that escalates from slightly off to completely unhinged, and may be funnier than the original

Delivery

The original Appley Rotten animation had launched on Valentine's Day 2004, so I soft launched the new site the week before the holiday to test the live experience, then expanded to existing Facebook fans with a celebratory 200th comic on the holiday itself. Concurrently, I experimented with new video formats on YouTube and Instagram Reels. In this final phase, I used AI to optimize code and implement analytics — enabling me to track individual interactions in more detail than ever before — and completed a week-long SEO pass resulting in full indexing of all site content and a publishing script for future updates.

Results

With a fairly limited launch, early analytics already tell an encouraging story. Dozens of visitors during the soft launch averaged over twenty comics per session — a strong signal that the reading experience is doing exactly what it was designed to do. Nearly a third returned within the first four weeks without any active promotion, and these early users show signs of gradual secret discovery and achievement completion across return visits.

Most of all, I'm personally happy to have the character back in my life and look forward to creating new art and experiences to share with new and old fans alike.