LENS LIVE
Building Multimodal AI Experiences for Product Discovery on Amazon
LENS LIVE
Building Multimodal AI Experiences for Product Discovery on Amazon
LENS LIVE
Building Multimodal AI Experiences for Product Discovery on Amazon
Executive Summary
Lens Live is a multimodal extension of Rufus, Amazon’s AI-powered shopping assistant, that brings visual search into real time. Launched in September 2025 to tens of millions of U.S. customers, Lens Live allows users to pan their phone camera across products and see matches update instantly in a swipeable carousel. By integrating Rufus directly into the interface, customers can go beyond recognition—asking questions, exploring compatibility, and receiving AI-powered insights without leaving the camera view. As lead UX visual designer, I extended Rufus’s design system into this new modality, defining scalable patterns for real-time interactions and ensuring that AI responses felt clear, trustworthy, and easy to act on.
Executive Summary
Lens Live is a multimodal extension of Rufus, Amazon’s AI-powered shopping assistant, that brings visual search into real time. Launched in September 2025 to tens of millions of U.S. customers, Lens Live allows users to pan their phone camera across products and see matches update instantly in a swipeable carousel. By integrating Rufus directly into the interface, customers can go beyond recognition—asking questions, exploring compatibility, and receiving AI-powered insights without leaving the camera view. As lead UX visual designer, I extended Rufus’s design system into this new modality, defining scalable patterns for real-time interactions and ensuring that AI responses felt clear, trustworthy, and easy to act on.
Executive Summary
Lens Live is a multimodal extension of Rufus, Amazon’s AI-powered shopping assistant, that brings visual search into real time. Launched in September 2025 to tens of millions of U.S. customers, Lens Live allows users to pan their phone camera across products and see matches update instantly in a swipeable carousel. By integrating Rufus directly into the interface, customers can go beyond recognition—asking questions, exploring compatibility, and receiving AI-powered insights without leaving the camera view. As lead UX visual designer, I extended Rufus’s design system into this new modality, defining scalable patterns for real-time interactions and ensuring that AI responses felt clear, trustworthy, and easy to act on.
Context
Amazon Lens had already given customers the ability to search with images and barcodes, but those experiences were static—requiring a photo to be taken or uploaded before results appeared. With the launch of Lens Live in September 2025, Amazon introduced a new paradigm: real-time, continuous scanning powered by computer vision and generative AI.
Now, instead of waiting for a result, customers could pan their phone camera across a scene and see product matches update instantly in a swipeable carousel. By integrating Rufus directly into the Lens interface, the experience also unlocked new depth: customers could tap into conversational prompts, ask follow-up questions, and receive AI-powered insights about compatibility, features, or use—all without leaving the camera view.
Context
Amazon Lens had already given customers the ability to search with images and barcodes, but those experiences were static—requiring a photo to be taken or uploaded before results appeared. With the launch of Lens Live in September 2025, Amazon introduced a new paradigm: real-time, continuous scanning powered by computer vision and generative AI.
Now, instead of waiting for a result, customers could pan their phone camera across a scene and see product matches update instantly in a swipeable carousel. By integrating Rufus directly into the Lens interface, the experience also unlocked new depth: customers could tap into conversational prompts, ask follow-up questions, and receive AI-powered insights about compatibility, features, or use—all without leaving the camera view.
Context
Amazon Lens had already given customers the ability to search with images and barcodes, but those experiences were static—requiring a photo to be taken or uploaded before results appeared. With the launch of Lens Live in September 2025, Amazon introduced a new paradigm: real-time, continuous scanning powered by computer vision and generative AI.
Now, instead of waiting for a result, customers could pan their phone camera across a scene and see product matches update instantly in a swipeable carousel. By integrating Rufus directly into the Lens interface, the experience also unlocked new depth: customers could tap into conversational prompts, ask follow-up questions, and receive AI-powered insights about compatibility, features, or use—all without leaving the camera view.
Design Challenges
Multimodal consistency
Extending Rufus’s conversational system into a camera-driven experience without fragmenting the identity.
Real-time UI clarity
Designing scanning, focus, and carousel states that updated continuously without overwhelming the user.
Trust in AI outputs
Ensuring that AI-generated insights and suggested questions felt contextual, transparent, and actionable.
Accessibility in an immersive view
Making sure UI elements remained perceivable and legible in the dynamic, live camera environment.
Compactness on mobile
Balancing the density of real-time results, conversational prompts, and product interactions in a constrained viewport.
Multimodal consistency
Extending Rufus’s conversational system into a camera-driven experience without fragmenting the identity.
Real-time UI clarity
Designing scanning, focus, and carousel states that updated continuously without overwhelming the user.
Trust in AI outputs
Ensuring that AI-generated insights and suggested questions felt contextual, transparent, and actionable.
Accessibility in an immersive view
Making sure UI elements remained perceivable and legible in the dynamic, live camera environment.
Compactness on mobile
Balancing the density of real-time results, conversational prompts, and product interactions in a constrained viewport.
Multimodal consistency
Extending Rufus’s conversational system into a camera-driven experience without fragmenting the identity.
Real-time UI clarity
Designing scanning, focus, and carousel states that updated continuously without overwhelming the user.
Trust in AI outputs
Ensuring that AI-generated insights and suggested questions felt contextual, transparent, and actionable.
Accessibility in an immersive view
Making sure UI elements remained perceivable and legible in the dynamic, live camera environment.
Compactness on mobile
Balancing the density of real-time results, conversational prompts, and product interactions in a constrained viewport.
My Role
As lead UX visual designer for Rufus' extension into Lens Live, I:
Extended Rufus’s design system and visual language into a camera-based modality.
Defined the UI for real-time scanning, focus states, and product carousels.
Integrated Rufus’s conversational prompts and insights into the Lens interface.
Collaborated with product, engineering, and conversation design partners to validate feasibility, refine interaction models, and test trust and usability in early prototypes.
My Role
As lead UX visual designer for Rufus' extension into Lens Live, I:
Extended Rufus’s design system and visual language into a camera-based modality.
Defined the UI for real-time scanning, focus states, and product carousels.
Integrated Rufus’s conversational prompts and insights into the Lens interface.
Collaborated with product, engineering, and conversation design partners to validate feasibility, refine interaction models, and test trust and usability in early prototypes.
My Role
As lead UX visual designer for Rufus' extension into Lens Live, I:
Extended Rufus’s design system and visual language into a camera-based modality.
Defined the UI for real-time scanning, focus states, and product carousels.
Integrated Rufus’s conversational prompts and insights into the Lens interface.
Collaborated with product, engineering, and conversation design partners to validate feasibility, refine interaction models, and test trust and usability in early prototypes.
Process
Building on the foundation established with Rufus, I adapted the design system to support real-time camera interactions. This began with analyzing the unique demands of Lens Live: the interface had to display scanning feedback, update product carousels dynamically, and present AI-generated insights—all while keeping the camera view unobstructed.
To address this, I designed clear state changes for scanning and product detection, using lightweight animations and focused highlighting to help users track what the system was recognizing. Product results were presented in a swipeable carousel anchored at the bottom of the camera view, minimizing intrusion while remaining immediately accessible.
Integrating Rufus’s conversational capabilities posed another challenge: how could customers ask questions or view AI-powered product insights without breaking the visual search flow? I worked closely with product, engineering, and conversation design partners to design inline prompts and summaries that appeared contextually beneath the carousel. These conversational touchpoints provided additional guidance while keeping interactions efficient and lightweight.
Amazon sets a high bar for accessibility, but existing standards hadn’t anticipated immersive, camera-based interfaces. I translated those requirements into new principles for Lens Live, ensuring sufficient contrast, legible type, and inclusive interactions across dynamic backgrounds. This allowed the experience to meet Amazon’s high bar for accessibility even in uncharted territory.
Process
Building on the foundation established with Rufus, I adapted the design system to support real-time camera interactions. This began with analyzing the unique demands of Lens Live: the interface had to display scanning feedback, update product carousels dynamically, and present AI-generated insights—all while keeping the camera view unobstructed.
To address this, I designed clear state changes for scanning and product detection, using lightweight animations and focused highlighting to help users track what the system was recognizing. Product results were presented in a swipeable carousel anchored at the bottom of the camera view, minimizing intrusion while remaining immediately accessible.
Integrating Rufus’s conversational capabilities posed another challenge: how could customers ask questions or view AI-powered product insights without breaking the visual search flow? I worked closely with product, engineering, and conversation design partners to design inline prompts and summaries that appeared contextually beneath the carousel. These conversational touchpoints provided additional guidance while keeping interactions efficient and lightweight.
Amazon sets a high bar for accessibility, but existing standards hadn’t anticipated immersive, camera-based interfaces. I translated those requirements into new principles for Lens Live, ensuring sufficient contrast, legible type, and inclusive interactions across dynamic backgrounds. This allowed the experience to meet Amazon’s high bar for accessibility even in uncharted territory.
Process
Building on the foundation established with Rufus, I adapted the design system to support real-time camera interactions. This began with analyzing the unique demands of Lens Live: the interface had to display scanning feedback, update product carousels dynamically, and present AI-generated insights—all while keeping the camera view unobstructed.
To address this, I designed clear state changes for scanning and product detection, using lightweight animations and focused highlighting to help users track what the system was recognizing. Product results were presented in a swipeable carousel anchored at the bottom of the camera view, minimizing intrusion while remaining immediately accessible.
Integrating Rufus’s conversational capabilities posed another challenge: how could customers ask questions or view AI-powered product insights without breaking the visual search flow? I worked closely with product, engineering, and conversation design partners to design inline prompts and summaries that appeared contextually beneath the carousel. These conversational touchpoints provided additional guidance while keeping interactions efficient and lightweight.
Amazon sets a high bar for accessibility, but existing standards hadn’t anticipated immersive, camera-based interfaces. I translated those requirements into new principles for Lens Live, ensuring sufficient contrast, legible type, and inclusive interactions across dynamic backgrounds. This allowed the experience to meet Amazon’s high bar for accessibility even in uncharted territory.
Outcome
Lens Live launched in September 2025 to tens of millions of U.S. iOS customers, with rollout planned across Amazon’s broader customer base. The feature not only made visual search faster and more seamless, but also introduced a new layer of depth—allowing customers to ask questions, explore compatibility, and make decisions in real time. By extending Rufus into Lens Live, I helped define interaction patterns for continuous, multimodal AI experiences, ensuring customers could shop, learn, and decide with greater clarity and confidence.
Outcome
Lens Live launched in September 2025 to tens of millions of U.S. iOS customers, with rollout planned across Amazon’s broader customer base. The feature not only made visual search faster and more seamless, but also introduced a new layer of depth—allowing customers to ask questions, explore compatibility, and make decisions in real time. By extending Rufus into Lens Live, I helped define interaction patterns for continuous, multimodal AI experiences, ensuring customers could shop, learn, and decide with greater clarity and confidence.
Outcome
Lens Live launched in September 2025 to tens of millions of U.S. iOS customers, with rollout planned across Amazon’s broader customer base. The feature not only made visual search faster and more seamless, but also introduced a new layer of depth—allowing customers to ask questions, explore compatibility, and make decisions in real time. By extending Rufus into Lens Live, I helped define interaction patterns for continuous, multimodal AI experiences, ensuring customers could shop, learn, and decide with greater clarity and confidence.