AI-powered Customer Review Highlights
Explored and designed new applications for Amazon's AI-generated review highlights to drive product discovery across the Smart Home category.
Amazon launched AI-generated review highlights as a company-wide capability. Leadership asked me to figure out what that could mean specifically for Smart Home customers. This kind of work sit in an interesting space because we wouldn't be building from scratch nor implementing someone else's spec with all the baggage that that implies. Thankfully, our org had always had certain room and influence to break away from company-level initiatives, so we were asked to ideate yet new ways to showcase this technology. The artifacts shown below are the exact slides I used to pitch these concepts to leadership, reflecting how product strategy and design communication actually happen at Amazon. One concept shipped, and two moved to the roadmap upstream at the org and company level.
Amazon launched AI-generated review highlights as a platform-wide capability, surfacing a short AI-written paragraph on product detail pages that summarized common themes across customer reviews. Leadership asked me to explore how our Smart Home Devices team could leverage this feature for our specific customers and business goals.
Customer reviews are one of the most important inputs to a purchase decision, but at Amazon's scale, they create their own problem. Customers want social proof, but not the work of reading through hundreds of reviews to find it. My job was to figure out where AI summarization could create the most value for Smart Home customers specifically.
The Problem
Internal data showed customers felt overwhelmed by large numbers of reviews and wished the information was summarized for them. For Smart Home specifically, this problem was compounded: customers were not just evaluating sentiment, they were trying to extract compatibility and use-case signals from unstructured review text.
The platform's new AI capability solved the summarization problem globally on the product detail page. The open question was how to bring it into other Smart Home surfaces in a way that served our customers and supported our commercial goals.
Exploration Concepts
I designed three applications of the AI review highlights feature for Smart Home, each targeting a different surface and a different moment in the customer journey.
- Concept 1: Homepage Widget
The idea was to bring AI-generated review sentiment upstream into the homepage, before a customer had even selected a product. Top-selling Smart Home devices surfaced alongside an LLM-generated headline and the most praised attributes across that product line, turning social proof into a discovery driver. This concept successfully shipped.
Leadership pitch slide detailing the Homepage Widget concept and AI attribute mapping.
Earlier explorations for integrating AI highlights.
Shipped homepage widget layout with product line attribute summary.
- Concept 2: Immersive Video Feed
Our Smart Home team had built an experience surfacing videos for products relevant to a customer's interests. I designed a method to merge top video reviews with our propensity models, then explored how AI-generated review highlights could be layered into that immersive surface.
Pitch slide showing the immersive video feed context where review highlights would surface.
- Concept 3: ASIN Cards
A third concept applied AI review highlights at the product card level, earlier in the funnel than the detail page. Each card surfaced an LLM-generated benefit message and curated customer quotes. The framing was that the voice of the customer generates the value message, removing the need for standard marketing copy.
Customer benefit ASIN Cards concept pitched to leadership.
Finalized ASIN card UI designs.
Outcome & Reflection
Contributed to the global Ai-powered review highlights feature.
The most useful thing a designer can do with a new platform capability is ask who it actually helps and where. Not every feature belongs on every surface. Designing three applications of the same capability forced a useful discipline: each concept had to justify itself on its own terms.
This project reinforced how much leverage there is in being the person who translates a broad platform capability into a specific team's context. The AI feature existed, allthough was new to everyone. My design work was finding where they intersected with enough precision to be worth building.