Multimodal AI
Built a library of multimodal generation patterns now used across Nano Banana, Gemini, Genie, Whisk, and Pomelli.
Design preview
New patterns for image generation
Most AI users struggle to write prompts. What if you could prompt with more than words?
Over six months, I helped Google identify and refine a library of new interaction patterns. Using iterative global research, I crafted new prompting and editing affordances that empower users to prompt with specificity and correct inherent biases.
We built new flows for professionals, who need precise controls for consistent characters and brand assets
My role
I lead the client relationship and was the primary IC across all activities, from workshops and global research, to design, synthesis, and socialization.
My work brought together teams across Google including Deepmind, Labs, Gemini, Research, YouTube, Search, and Ads.
Team + timeline
Design lead (me)
Jr. Researcher
Strategy advisor
10 months
Activities
Research
Expert interviews
Landscape audit
Opportunity definition
Global recruiting
Generative IDI’s
Insights & synthesis
Strategy
Ideation workshop
Strategic frameworks
Product Design
Concept generation
Stimuli development
Flow design & prototyping
Product testing
Design toolkit
Pattern & token development
Development
Prompt engineering
Model architecture design
Confidential work.
Details are available on request