Superland

Superlandwas a creative pivot born inside AWDIO, a startup originally focused on interactive audio adventures for kids. What began as an experimental AI workshop became a full-on product pivot, culminating in a screen-light entertainment app for children aged 4 to 12. At its heart, Superland was built to be an alternative to passive screen-based media. It allowed kids to generate characters, objects, and places and use them in magical, AI-powered audio stories. We created a product that entertained and encouraged creativity and exploration by combining playful UI, innovative storytelling systems, and a lean but determined team.

We started from a shared pain point: most kids’ entertainment lives at two extremes: screen-heavy or entirely analogue. We believed there was room for something in between that used technology to spark creativity rather than stifle it. The trigger moment came during a three-day internal “AI sandbox” session. I built a silly superhero backstory generator for fun. The results were so surprisingly compelling the team collectively decided to pivot away from our original idea of static, choose-your-own-adventure audiobooks. That experiment became the seed of Superland.

I served as CTO of AWDIO. Officially, that meant owning everything tech-related, app development, infrastructure, tooling, and performance. In practice, because of my hybrid background in design and engineering, I also played a key role in product ideation, experience design, and feature development.

The core team included:

• Matthias (CEO) strategy and operations

• Jonas (CPO) product design and storytelling

• Daniel & Sorin frontend and backend engineers

We worked remotely, coordinating through Linear, Slack, and weekly standups. Despite our small size, decision-making was fast and organic. Ownership was clear: if something broke in tech, it was mine to fix.

Like any startup, we faced tight constraints:

• Time about 8 months of the runway when we pivoted to Superland

• Budget stretched via grants to last longer but still lean

• Tech Complexity early AI systems weren’t as plug-and-play as promised

• Offline-first we wanted the app to work seamlessly even without internet

We also underestimated the friction of working with LLMs regarding cost (computing is expensive) and behaviour (AI doesn’t always play nice). Designing around those realities became a defining challenge.

We started with a character creation flow that gave kids meaningful input and immediate narrative output. Created characters were stored in a persistent collection 📷 superland_02 📷 superland_01 📷 superland_03 📷 superland_00 📷 superland_04. Later, we introduced standalone story generation using existing characters, objects, and places 📷 superland_06 📷 superland_07 📷 superland_12. We visualized progress using a "Snake" timeline 📷 superland_10 📷 superland_11, added a read-along mode 📷 superland_08 📷 superland_09, and redesigned the homescreen to serve as a creative hub 📷 superland_14 📷 superland_13.

As CTO, I architected and implemented:

• Amodular backendusing Node.js after edge functions proved unstable for long prompts and audio generation

• Asentence-by-sentence TTS playbackapproach to minimize loading time and support “long mode” narratives

• Local caching and sync systemsto support offline-first design

• Adynamic template engineusing JSON, allowing modular backstory and story content generation

Synchronization between local and remote databases for kid-created content

All of it was designed to scale and avoid the classic trap of overbuilding. I aimed for flexible, adaptable systems rather than tightly coupled perfection.

We ran two-week sprints anchored by planning and wrap-up meetings. I collaborated closely with the CPO to break ideas into specs and tasks. Our culture was one of autonomy: I trusted the team to ship but kept tabs to prevent over-engineering spirals (especially with developers who needed frequent nudges).

Our approach evolved from scrappy solo work to a more structured cadence. We stayed reactive, shipping often and folding in honest feedback fast, including essays from one of our most dedicated young users.

The biggest pivot, moving from pre-recorded content to generative AI, unlocked a whole new product at a cost. We shelved months of carefully crafted content. That was hard. But it taught us to be less precious.

Other shifts included:

• Simplifying the creation UIs for younger users

• Moving audio generation from edge functions to dedicated servers

• Adapting the story engine to manage continuity between stories

That story engine became our most complex system, handling character states, narrative consistency, dynamic updates to the “world,” and multiple feedback loops to ensure story coherence.

We launched multiple updates of Superland and ended up with a small but loyal audience, around 4,000 monthly active users. Some kids came back every week. Some sent us detailed feedback. And for a few, we became part of their creative routine. That’s more than enough.

We didn’t just build a toy. We built a sandbox where imagination could run wild without dopamine loops or ads.

Looking back, I’m proud of our ambition and willingness to prototype ideas others might have dismissed as “too weird” or “too hard.” I’m also proud of my growth, teaching myself React Native on the job, building complex backend systems, and staying flexible through fast-moving change.

That said, I would’ve pushed us to ship sooner. We cared too much about making it perfect for investors. We should’ve let more kids break it earlier.

• Launch early, learn faster

• Trust your instincts, but test them

• Gen AI is a creative partner, not a plug-and-play fix

• Kids are smarter and more imaginative than most adults design for

• Magic happens in the messy middle

If I had to pick one thing only I could have done, it’s this: I wasn’t afraid to tackle the impossible. From wrangling dozens of prompt chains into a working narrative engine to figuring out how to generate and stream AI stories on the fly, I kept pushing. And I encouraged others to do the same. Because at the end of the day, design should feel like magic, even if you know how the trick works.