Facet.ai
Building generative AI tools for creatives.
JULY 2022 — MAY 2024
Facet.ai was a Series A startup acquired by Adobe in May 2024.
Originally an AI-powered, web-based photo editing tool, Facet transitioned to an AI-first creative studio with the launch of our flagship product, Generative Canvas.
After joining the 15-person team as a new grad in 2022, I grew into the senior design lead by developing multifaceted skills in design, growth, prototyping, and product strategy.
Originally an AI-powered, web-based photo editing tool, Facet transitioned to an AI-first creative studio with the launch of our flagship product, Generative Canvas.
After joining the 15-person team as a new grad in 2022, I grew into the senior design lead by developing multifaceted skills in design, growth, prototyping, and product strategy.
Designing Generative Canvas
While at Facet, I led the design of core features for Generative Canvas, our flagship tool for creatives.
The process often began by prototyping with AI/ML engineers to test novel experiences for greater precision in generative workflows. Working with our platform engineers, I saw the designs through to implementation. Building from 0 to 1, we launched with 10K+ MAUs.
Smart suggestions
To prevent cold starts and writer’s block, we created a ‘/’ interaction for the prompt bar that would dynamically surface suggestions based on what the user has already written.
To prevent cold starts and writer’s block, we created a ‘/’ interaction for the prompt bar that would dynamically surface suggestions based on what the user has already written.
Variance
The homogeneity of generative imagery poses a problem to creatives, who start from broad strokes of ideation.
The variance slider follows creatives throughout their process, allowing them to control how divergent they want the results to be.
The homogeneity of generative imagery poses a problem to creatives, who start from broad strokes of ideation.
The variance slider follows creatives throughout their process, allowing them to control how divergent they want the results to be.
Convergence
When creatives reach a result they like, they can use the “Generate more like this” feature to converge on an idea.
When creatives reach a result they like, they can use the “Generate more like this” feature to converge on an idea.
Region prompting
With the emergence of real-time diffusion results, region prompting became possible.
Users could now direct specific areas of their generative compositions and see the results update immediately.
With the emergence of real-time diffusion results, region prompting became possible.
Users could now direct specific areas of their generative compositions and see the results update immediately.