I had the opportunity to be one of the few cross-functional leaders tasked with dreaming and building eBay.ai, a foundational initiative designed to weave responsible, human-centered AI into the fabric of eBay’s platform. My work wasn’t just about building features — it was about crafting the design language, principles, and reusable components that would allow AI to enhance discovery, simplify selling, and enrich user interactions, all while preserving user agency and ensuring inclusivity at every step.
With millions of constantly changing listings, eBay’s marketplace presents both a challenge and an opportunity for AI. Buyers increasingly expect smarter personalization and frictionless discovery, while sellers need tools that simplify their workflows without compromising control. The goal wasn’t to just sprinkle in AI — it was to thoughtfully integrate it into the experience, making eBay feel more intuitive and capable, without losing the human element that makes our platform special.
That duality — technology enhancing, not replacing, human decision-making — guided every aspect of this work. We needed to balance automation with transparency, personalization with control, and innovation with responsibility.
We began with an intensive AI sprint, where we identified potential AI use cases. After validating concepts, we consolidated patterns and sought feedback from stakeholders.
We developed prototypes to test different AI flows, fine-tuning elements such as personality, volume, and user control. These prototypes helped define how AI should "show up" in the eBay ecosystem.
We developed prototypes to test different AI flows, fine-tuning elements such as personality, volume, and user control. These prototypes helped define how AI should "show up" in the eBay ecosystem.
The AI system was gradually rolled out across the organization. We collected ongoing feedback, iterated on components, and expanded AI education initiatives within eBay.
We use AI as a tool to build enhanced experiences that proactively anticipate someone’s goals in personalized and meaningful ways.
We give people control over their AI-powered experiences through transparency and awareness.
We design AI-powered experiences that can be used and understood by a diverse group of people of all abilities.
We mitigate prejudice and bias when using GAI to create new images, videos, or texts with a human in the loop.
We confront the uncanny valley & spread of inaccurate information when using GAI through analysis, testing, and planning.
We engage in a system of checks and balances with our XFN partners, so we can move fast and fix things.
eBay.ai introduces a suite of AI-powered tools designed to streamline interactions and enhance personalization. The AI-powered shopping assistant provides real-time recommendations and helps users navigate the platform with ease. Smart search and AI-enhanced product discovery leverage natural language processing and predictive analytics to improve search relevance. Additionally, eBay.ai incorporates conversational AI through chat-based interactions, enabling seamless buyer-seller engagement. AI-powered visual search allows users to find items based on images, making shopping more intuitive. Lastly, AI-enhanced motion and interactions bring fluidity and responsiveness to the platform, ensuring a seamless experience.
We developed a decibel level system for the AI foundations that allow for us to modulate brand emphasis, motion, and messaging dependent on the use case.
Education modules provide users a short description of a feature with a visual connection to the icon. They can use a single icon or an icon with background gradient.
Assistant modules offer guidance and ingresses into AI generated features. They utilize the full spectrum filled icon to denote they are powered by AI.
The chat interface is where users interface with eBay AI. Similar to messaging, where several types of content can be sent and received. In addition to text messages users can send and receive listings, categories and image carousels.
Access to the AI chat interface will be accessible in several areas and is contextual based on where the user is in their shopping journey. Ingresses may be found throughout the product in-page in AI specific modules. Check out our showcase page to see more examples.
The foundations and strategy laid out in this sprint provided the much-needed springboard to ideate and realize the foundations, principles, and features needed for an integrated AI future. Some of the new features that came out of this sprint include AI descriptions, AI photo background generator, AI shop the look, and the AI agent. It was an honor to help lead this initiative - spanning 100+ participants and 5+ locations.
New
Colors
New
AI Icons
New Components
Sprint Participants
Read next: ebay magic moments framework elevating design craft in a global marketplace