The Challenge
MSLA branding reproduced for portfolio purposes. Copyright © MSLA Racing
For several years, I produced much of the visual material used by MSLA, an online community built around marble racing through Marbles on Stream and inspired by its founder's love of motorsport and iRacing. What began as paid freelance work developed into a long-running creative partnership with MSLA's founder, Enderzworld.
By 2026, the circumstances around that work had changed. Budgets had reduced, Enderz had less time to devote to MSLA, and I was less involved in the day-to-day community than I had once been. The volume of commissioned design that had previously been possible was no longer sustainable.
But the need for visual content had not disappeared. MSLA still ran events, communicated with its community and needed graphics for its broadcasts.
The problem wasn't how to replace a designer. It was how to help something we had spent years building continue without depending on one.
Context
I began working with Enderz in 2021. Initially, I charged for my time while we worked out what the recurring design work actually involved. Eventually we settled into a simpler model: a flat fee for regular race logos, with larger pieces such as animation and broadcast assets priced according to the additional work involved.
The pace was unusual. MSLA regularly needed a new identity for its next event, sometimes almost weekly. Enderz and I would bounce around ideas, look at dates, anniversaries, sponsors, community jokes or whatever else gave us an excuse for a theme, and turn those ideas into graphics.
Over time, individual commissions accumulated into a visual language. Community volunteers became illustrated pit crew. Characters returned across different events. In-game designs found their way into artwork. Graphics became merchandise, broadcast elements and animations. A joke from one event might reappear years later.
My surviving MSLA archive alone contains more than 900 files and almost 17 GB of material. That scale matters less as a measure of output than as evidence of how much shared visual history had accumulated.
Constraints
Reduced budget The volume of regular commissioned design work that had supported MSLA's visual identity was no longer financially sustainable.
Limited time Enderz had less time available to run MSLA, so any new approach needed to reduce rather than add to the production workload.
Visual continuity New assets needed to remain recognisably MSLA, retaining established characters, colours and visual conventions developed over several years.
Community meaning Characters represented real volunteers, recurring jokes and shared history. Preserving their appearance alone was not enough to preserve what they meant.
Accessible production The new workflow needed to be usable by people without professional design skills or specialist creative software knowledge.
Reduced dependency The process needed to allow Enderz, and eventually others, to create and adapt assets without requiring me to produce every new graphic.
Copyright and reference Community jokes and themed events often referenced films, games and popular culture. The challenge was to make those references immediately understandable without simply reproducing existing characters, logos or artwork.
Production cadence MSLA ran frequent races and events, sometimes requiring a new identity almost every week. Assets needed to be produced efficiently while still giving each event its own character rather than simply repeating a template.
Design Strategy
The starting point was not AI. It was the work that already existed.
Years of MSLA artwork provided established characters, colours, compositions and visual conventions. Rather than asking generative tools to invent a replacement identity, I treated that existing library as the reference point for anything new.
AI could then perform a much narrower and more useful role: helping us extend, adapt and animate things that already belonged to MSLA.
That distinction became central to the approach. The objective wasn't maximum automation or maximum output. It was to lower the skill and time required to make useful new assets while retaining enough human judgement to decide what actually belonged.
Just as importantly, the workflow needed to be teachable. If Enderz could use the existing assets as references and understand where AI helped and where it needed correcting, he would no longer need to commission me for every variation.
Design Process
The process evolved gradually rather than through a formal technology rollout. I experimented first on my own work, learned where the tools were useful, and retained manual control where consistency or detail mattered.
As the tools improved, existing artwork could be taken further. Finished graphics could become animated assets. Established characters could be placed into new situations without rebuilding them from scratch.
The archive became part of the system.
In MSLA's early years, much of the community's history lived in shared spreadsheets. My original portfolio gradually became a visual record: every race logo was archived with its name, date and sponsor. Enderz began using it during streams as a visual reference, looking back through previous races and quickly identifying when an event had taken place.
What began as a way of documenting my own work had unintentionally become a visual source of truth for the community.
When MSLA's circumstances changed, that experience became something I could transfer. I showed Enderz how to use the existing visual material as the starting point for new AI-assisted work rather than relying on text prompts to reinvent the identity.
The strongest evidence that the process transferred successfully came later. Tiarra, who now sometimes runs MSLA when Enderz is busy, began developing her own uses. She created variants of her established unicorn character with eyes open and closed and mouth open and closed, allowing the character to appear to talk during broadcasts.
At that point, she wasn't simply following a process I had demonstrated. She was applying the principle to solve a new problem herself.
The Solution
The resulting solution isn't a piece of software or a fixed template. It is a different way of working.
MSLA now has a substantial library of established visual assets, people who understand what those assets represent, and accessible tools that allow them to adapt and extend the library without requiring the original designer for every piece of production.
AI provides speed and accessibility. The archive provides continuity. The people running MSLA provide the judgement.
That combination has allowed a visual system originally dependent on regular commissioned design to become something its community can increasingly maintain for itself.
"The goal wasn't to replace the designer. It was to make sure what we'd built could continue without one."
Impact
5+ years of continuous visual development
900+ retained project files
2+ people now independently extending established assets with AI
The most meaningful outcome is harder to quantify.
The previous arrangement depended on MSLA having enough budget to commission regular design work from me. The newer model reduces that dependency. Enderz can create material himself, and the capability has begun spreading to others rather than remaining with the person I originally supported.
From a purely commercial perspective, teaching a client to need less of my paid time might seem counter-intuitive. But by then Enderz was also a friend, and MSLA was a community I cared about. I would rather transfer what I had learned than watch something we had spent years building disappear because its finances had changed.
Success meant making myself less necessary.
Design Artefacts
Looking Ahead
I don't spend as much time rolling with MSLA as I once did. People have occasionally asked where I've gone, and some have said that a little of the old spirit has gone with me.
I think that observation matters.
AI has made more of the visual system transferable than I might once have expected. It can help someone adapt an established character, create a new variation or animate something that previously required considerably more time and specialist skill.
The technology changed alongside the project. When I began working with MSLA in 2021, generative AI was not part of my design process. Over the following years, the technology developed dramatically. By the time budgets and circumstances meant I needed to step back, it had reached a point where it could help others work with the characters, conventions and assets I had already created.
But MSLA's identity was never contained entirely in its files.
It also came from knowing the people behind the characters, understanding the jokes, spotting the absurd idea that would make a good race, arguing over details with Enderz and occasionally deciding that the correct creative direction was to take over the entire community, turn everything pink and sparkly and fill the merchandise store accordingly.
Those things are harder to transfer because they come from participation.
For me, that has become one of the most useful lessons from adopting AI: understanding what can be handed over is only half the job. You also need to recognise what was human all along.