Defining a character-driven identity for MGID's AI assistant

Overview

MIA is the intelligent assistant inside MGID's advertising platform โ€” and the first time the brand needed a face, not just an interface. As Head of Product Design, I initiated and led the identity work: there was no brief for "what an AI assistant should look like," so the first job was defining the problem itself. I set the two constraints everything else had to answer to โ€” the character had to carry real personality in marketing and hero moments, and reduce cleanly to a small UI icon without losing recognition. Most mascot systems fail at one of these ends; I made solving both a non-negotiable from the start.

I drove the work from strategy through execution: positioning the character against the crowded "friendly blob" pattern, defining a two-tier identity system โ€” glossy 3D for brand surfaces, flat with gradient hints for product UI โ€” and locking the recognition anchors that hold both tiers together. Beyond the character itself, I established the production approach: a generative pipeline combining prompt-engineered image models, AI video, and code-based animation, documented so the design team can extend MIA's states and assets without me in the loop. The outcome is not just a mascot but a system โ€” identity rules, motion language, and a repeatable way to produce both.

Categories

Brand & Character

Leadership

Company

MGID

Year

2026

Defining a character-driven identity for MGID's AI assistant

Overview

MIA is the intelligent assistant inside MGID's advertising platform โ€” and the first time the brand needed a face, not just an interface. As Head of Product Design, I initiated and led the identity work: there was no brief for "what an AI assistant should look like," so the first job was defining the problem itself. I set the two constraints everything else had to answer to โ€” the character had to carry real personality in marketing and hero moments, and reduce cleanly to a small UI icon without losing recognition. Most mascot systems fail at one of these ends; I made solving both a non-negotiable from the start.

I drove the work from strategy through execution: positioning the character against the crowded "friendly blob" pattern, defining a two-tier identity system โ€” glossy 3D for brand surfaces, flat with gradient hints for product UI โ€” and locking the recognition anchors that hold both tiers together. Beyond the character itself, I established the production approach: a generative pipeline combining prompt-engineered image models, AI video, and code-based animation, documented so the design team can extend MIA's states and assets without me in the loop. The outcome is not just a mascot but a system โ€” identity rules, motion language, and a repeatable way to produce both.

Categories

Brand & Character

Leadership

Company

MGID

Year

2026

When I deliver, I deliver ๐Ÿ’…๐Ÿป

ยฉ 2024

Warsaw

2:55 PM

When I deliver, I deliver ๐Ÿ’…๐Ÿป

ยฉ 2024

Warsaw

2:55 PM

When I deliver, I deliver ๐Ÿ’…๐Ÿป

ยฉ 2024

Warsaw

2:55 PM