Persona Systems
A set of persona-driven agents I built to think and react like real players, used to pressure-test every key visual against the audience it's actually made for, before it ships.
I've spent years pointing games toward a feeling before the tools existed to reach it. Now the tool is generative. My job hasn't changed: hold the vision, break the model until it surprises you, and ship the idea that makes a player stop scrolling.
Generative tools changed the pipeline. They didn't change what makes direction good.
When anyone can generate anything, the real risk isn't a wrong idea. It's an ordinary one. My job is picking the version that makes the room go quiet, not the one that plays it safe.
A generative pipeline isn't a shortcut to the final frame. It's a thousand rough ones, and the discipline to recognize the one that wasn't the plan.
I start from feel (mood, silhouette, light), and only then build the prompts, constraints, and systems that let a model reach for it on demand.
Anyone can generate. Direction is knowing, instantly, which output belongs in your world, and which one is just noise wearing your palette.
Six ways of working at the creative direction level, so everything downstream (key art, animation, the whole world) is easier to art direct well.
A set of persona-driven agents I built to think and react like real players, used to pressure-test every key visual against the audience it's actually made for, before it ships.
A console I built that reads a game's design bible and steps through its own production process (key art, 3D, animation, marketing), generating on-brief content across formats with a director still on every decision.
A system I built that turns one game's world into podcasts, spoken stories, serialized dramas, comics and short-form video: the same characters, rewritten from scratch for whatever format someone's actually going to watch, hear or read.
Filling in the gaps a game design leaves open, so every visual decision downstream has somewhere consistent to stand: I built this out as JUDAS, an original IP spanning ancient gods, engineered humans and the crime dynasties and corporations fighting to own both.
Taking the most over-familiar mechanic in mobile games and asking what happens when someone actually directs it, rather than assembles it: I built this out as Peachy Airways, a two-week sprint through Midjourney, ChatGPT and Claude to push a merge board's visuals, animation and feel past what the genre usually settles for.
Play the prototype ↗After Peachy Airways proved a merge game could be directed in two weeks, the next brief was tighter: two days to build a polished match-3 core board, the sixty seconds the whole genre lives or dies on. The result is Splatisserie, a patisserie food fight where every match splats across the tablecloth and the mess is framed as a painting at the end.
Play the prototype ↗Empathy first, then AI: deep research on who I'm building for, solid enough to trust completely.
Gather the emotional target before touching a model.
Push the generative system past its defaults until it surprises.
Build the taste-filter that scales past one director.
Turn the best output into rules the pipeline keeps.
The ideas live in the game, not the deck.
Thirty-two years across two disciplines, one eye. Formal training at two art schools came first, then seven years as an illustrator, represented by agencies in London, New York and Tokyo — commissioned work I built entirely by hand, long before any of it touched a screen. Twenty-five years ago that same eye moved into art direction inside the games industry, and it never left: I've led visual direction on mobile titles played by hundreds of millions, run art departments across more than twenty productions, and shaped the visual language of major licensed entertainment brands along the way — work that has, between it, touched billion-dollar franchises. Recently that same instinct has turned toward generative AI: I build the creative workflows, systems and pipelines that let a small team explore, iterate and ship at a pace that used to need ten times the headcount. Other Matters is where that thinking lives now — three decades of trained hand and trained eye, aimed at the newest tool either has ever been given.
Several of the systems above were built under contract, for specific studios — the tooling, its interface and its real outputs belong to them, not to this site, under the terms of those engagements. So they're described here the same way everything else is: the problem it solved, the mistake that taught the lesson, what it protects now that it's running. Where a screen recording is missing, that's not an oversight. It's the same discretion a studio should expect its own work gets, once it's mine to protect instead of show.
Many of these come from systems I've built for consistent, repeatable creation: one visual direction held steady across marketing content, 3D assets, animation, and transmedia storytelling. Filter by category, or select a piece to open it large.