Creative Direction, Art Direction, Strategy

Facebook Illustration System: Caleo

Evolution

The refresh was about refinement, not reinvention. Silhouettes were streamlined, forms made sleeker. A matte, ceramic-like material replaced shinier surfaces. The goal was sophistication that felt effortless rather than polished.

Color decisions were intentional down to the specific value. A single blue anchor kept illustrations vibrant and distinctly Facebook rather than drifting toward pastels. Tertiary bounce light colors were pulled from the Reactions system, creating subtle cohesion across two visual languages that needed to coexist.

After

Before

Comprehensive styleguide

With the visual language defined, the next challenge was making it teachable. Working with a team of three, the system was codified into a comprehensive style guide, a living reference covering everything from material and texture application to color rules, form principles, and lighting behavior.

The guide was written to be instructive rather than prescriptive: grounded enough to maintain cohesion, flexible enough to support creative judgment. The goal was that anyone coming into the work could understand not just the what, but the why behind every decision.

AI tooling

Built on the foundation of Caleo's visual language, the next challenge was making the system generative. Co-leading a team of four, the work involved developing an AI pipeline using Figma Weave that could produce on-brand illustrations on demand, supporting idea exploration and cutting production time without sacrificing the craft standards the system was built on. The tool shipped as a Figma plugin, living right where designers worked.

We defined the style. Then we taught a machine to speak it.

But generation was only part of the process. Once Figma Weave produced its output, the real work began: reviewing, selecting, and deciding what actually worked. Not every generated illustration had the right feel, the right weight, or the right personality. That's where taste came in. Knowing which one to keep, which to discard, and why, that judgment is what separates a system from a craft.

Final outputs aren't available to share due to confidentiality.