A.I. Experiments

Like most people, I started with Generative AI, exploring realism and motion in image and video. Shona was a dancer I created back in the day with Midjourney and the music remix I created with Suno. It was fun, but Generative was only the beginning.

I quite quickly become obsessed with trying to capture human emotion. Could AI be used to make us feel something? This led to a more mixed-media approach where I used real photos and brought them to life specifically to go from happy-to-sad and back again in the shortest space of time. The original imagery were selected for their intimate studio setting and black backdrops to really pull the viewer into the muse’s face. Music and Lyrics are AI too.

The next challenge was to use one AI to control another. By doing so, I was able to not just be more efficient in the creative process, but to gain new powers in areas were I previously had no expertise, in this case it was 3D. These examples show my use of the LLM Claude, and it’s Model Control Protocol (MCP), to instruct the 3D software Blender. Crazy really, creating 3D design using just words.

The next step was to use AI to create bespoke tools and platforms to improve and enhance creative and strategic workflows and processes.

I’ve personally built tools to address specific problems in my creative and agency work.
Supersystem is creative ideation and production platform that supports strategists, creatives and producers in creating unexpected ideas and executions.

Frankly analyses live Reddit conversation to produce sentiment, audience and brand-safety reports, so campaign and creative briefs are based on what people actually say rather than assumed personas.

Selecta is a review tool for AI-generated image and video variants: clients open a share link, react to each option, and pick one winner per shot, replacing open-ended feedback threads with a single decision.

I also built MoodAI, a Chrome extension powered by GPT-4o mini that turns web browsing into an AI-assisted moodboard.

As a creative leader, I believe it’s no longer enough to just have great instincts and inspire a room.  ECDs and CCOs need to be genuinely fluent in the tools reshaping the work itself, not just delegate that fluency downward. That means staying hands-on, prototyping alongside the team, and having the technical curiosity to know what’s actually possible before briefing it.

Agency and Client leaders will increasingly value creatives who treat AI fluency as a commercial skill, not a creative side project, because clients are already asking, and the leaders who can answer with real, hands-on experience are the ones leading the room.

Because none of this started with a brief, just curiosity and a willingness to break things until they worked. If that’s the kind of collaborator you’re after, let’s build something.