AI Shopping Galleries
Automatically recommend relevant affiliated products based on your content and display them in dynamic, customizable galleries.
For commerce-focused publishers, nothing is more important than delivering an elevated shopping experience that keeps your audience coming back for more. Shopping galleries are a popular choice for affiliate marketers because they help readers find and purchase products they love in one convenient location — and they’re proven to increase clicks and conversions.
The problem is building a traditional shopping gallery requires a significant amount of manual clicking, copying, and pasting. So if you’re publishing commerce content at scale — or you’re looking to add shopping features to your existing content — the effort can be time consuming and labor intensive.
AI Shopping Galleries automatically recommend relevant affiliated products based on your content and displays them in dynamic, customizable galleries.
How it works:
Content analysis
The system scans your page content, analyzing topics and keywords using natural language processing to identify the most relevant product categories.
Predictive analytics
It then generates smart search queries based on the content analysis. These queries are used to search our extensive product database.
Product recommendations
Once the search queries are executed, the system analyzes the search results and selects the most relevant products.
Gallery creation
Recommended products are displayed in an attractive, brand-agnostic, customizable gallery designed to meet IAB standards for responsive layouts.
Tech Stack / Design
Client
- Preact
- Vite
- Vanilla Extract
- TypeScript
- Runtime validation
- Functional programming
Constraints
- Speed. Distributed JavaScript must not negatively impact page load time.
- Privacy. System must be GDPR / CCPA compliant.
- Developer experience. Galleries must be usable in client-side rendering (CSR) and/or server-side rendering (SSR) environments and easy to install.
- Brand Agnostic. Publishers must be able to apply their own styling to comply with brand guidelines.
amongst many others...
Server
- Python
- FastAPI
- LlamaIndex
- Corrective Retrieval Augmented Generation (CRAG)
- Prompt engineering
- Auto-generated client SDKs
and much more...
Role
I led the client-side development for this project and worked closely with the product and design teams to build a product that was both beautiful and functional. I also contributed to the server-side development helping guide a thoughtful API design.
This was a very exciting project to work on. I learned a lot about AI and how to build a product that uses it.
TODO: Add more details 😃
Results
The AI Shopping Galleries feature was a success and was rolled out to all Sovrn Commerce publishers. The feature has been a major driver of traffic and revenue for many publishers.