MG Motors. Generative AI in service of live experience.
Project Manager
Tomasz Cholewa (Dev)
The launch of the MGS5 EV focused on innovation. The audience, mainly automotive press is experienced and difficult to impress. To extend the technological narrative of the car, we introduced an interactive kiosk powered by Generative AI. Guests could create and share imaginative visuals featuring the MGS5 EV, extending the experience beyond the event itself.
Stage One, catch the Attention
Creative AI tools often fail at the starting point. A blank input field can feel intimidating, especially for first-time users. To avoid this, the standby screen showcased a looping gallery of example generations. These visuals showed what was possible, sparked ideas, and made the system feel accessible before anyone typed a word.
Stage two, Engage
The first interaction began with a initial CTA that opened the creation interface. A large input field in a center was designed to capture intent quickly and with minimal effort. Since most users struggle to craft detailed prompts, we supported the input with predefined style selections. These added structured visual direction to the prompt, allowing users to refine results without needing prompt-engineering skills.
Lower the Entry Bar
To make starting even easier, interface included a Randomizer. With one click, the system generated a seed and a starter prompt that users could tweak and run. It gave people a way in without overthinking. The feature subtly demonstrated how prompts and styles interact, helping users move from curiosity straight to iteration.
Encourage exploraion
Generative outputs rarely feel perfect on the first try. To support exploration, we introduced Regenerate and Edit options. Regenerate kept the same prompt but changed the seed, producing new variations. Edit allowed users to refine their prompt while maintaining consistency. Together, these tools balanced surprise with control and encouraged users to keep experimenting.
Final stage, take it with you, or share
Generative outputs rarely feel perfect on the first try. To support exploration, we introduced Regenerate and Edit options. Regenerate kept the same prompt but changed the seed, producing new variations. Edit allowed users to refine their prompt while maintaining consistency. Together, these tools balanced surprise with control and encouraged users to keep experimenting.
