Launching Generative AI Vision

Led the design for a set of Generative AI Vision features, all within tight deadlines and complex technical constraints. Attracted significant customer base for early trials. Featured in Cloud Next ‘23 Keynote by Executives.

Image generation

Background

Studio Vision empowers developers with generative vision AI capabilities. I led the design for a range of advanced Generative AI Vision features encompassing fine-tuning, style transfer, digital watermarking, prompt expansion, image captioning, image understanding, and video generation.

Goals

  • Unified platform: An integrated hub for exploring, envisioning, and experimenting with diverse vision generation capabilities.
  • Seamless Editing: A seamless experience of generating and editing media with prompts.
  • Better Prompting: Help users build better prompts with hands-on “learning by doing”.

Final design

From Words to Pixels: Unleash Your Creativity with Image Generation

Image generation

Embark on an artistic journey with image generation, where mere words ignite a symphony of pixels. Seamlessly refine your creations by steering them towards specific styles or subjects with a single click, unlocking endless creative possibilities.

Unlock Video Insights: Single-Click Access to Concise Video Descriptions

Video description

Effortlessly grasp the essence of a video with a concise description that encapsulates its key points, accessible within a single click.

Crafting Expressive Video Narratives with Scene-Based Generation

Video generation

Empower users with granular control over video creation through scene-based generation, enabling the crafting of personalized and nuanced video narratives.

Impact

  • Highlighted as top tier announcement at I/O and Cloud Next opening keynotes by Sundar, showcasing Google's leadership in Generative AI.
  • Signed deals with key customers like Canva and Typeface.
  • Gained substantial media attention, including TechCrunch and Engadget.

“We decided to integrate Imagen into our app marketplace to provide it to our users. You’ve made great progress since we first meet last April” - Danny Wu, Head of AI Products, Canva

"By combining Google Vertex AI’s Imagen with Typeface's brand-personalized AI, we are able to help enterprises to create 10x personalized content in a fraction of time." - Vishal Sood, Head of Product at Typeface.

Challenges

The paramount challenge in this project was striking a balance between rapid iterations within the evolving technical landscape and delivering an exemplary user experience. For instance, in the digital watermarking domain, the team initially proposed mandatory user setup for this feature due to technical constraints. By recognizing the suboptimal nature of this approach, I championed the complete removal of user setup through mockups, fostering cross-functional alignment across Cloud and DeepMind.

To effectively navigate this challenge, I harnessed my expertise in rapid iteration, constructing multiple prototypes to engage stakeholders and refine a diverse range of concepts. This iterative approach enabled us to continuously assess and refine the user experience while maintaining the pace of development.

Next project

Redesigning Teambition’s Mobile App

See case study