Elevating E-Commerce: AI-Infused Product Recommendations on Shopify with Seaplane

The biggest e-commerce sites are able to leverage AI to personalize every customer interaction. But what about the millions of smaller sites? What if there was an easy way to drop in generative-AI powered personalization into, say, any Shopify-driven site in the world?  Today we’re thrilled to introduce you to something we’ve been working on - a fun (Demo) Shopify e-commerce site, Double Black Snowboards, powered by the Seaplane Hybrid AI Recommendation Engine and Hydrogen Open Source Project.  In this case, Hybrid AI means using both Very Large Language Models (70B+)(LLM) and a network of edge-based Low Latency Edge (7B) Large Language Models (LLMs).

In this blog post, we'll take you through the highlights of how Seaplane can transform the way Shopify store owners engage with their global customers, significantly boosting revenue, loyalty sign-ups, and recommendation accuracy, all while keeping costs down and improving customer experience.  In a follow-up blog we’ll be showing you exactly how easy this is with Seaplane.

Building the Seaplane LLM Recommendation Engine - A Shopify-Driven E-Commerce Site

Before we start, Seaplane is not getting into the Snowboard sales business - as much as some of our employees would love that!  We are building this to showcase Seaplane’s ability to dramatically simplify AI-infusion into a global-scale e-commence site. We picked Shopify as a terrific exemplar platform enabling anyone to create powerful web-stores; but adding hyper-personalized recommendation can apply to any e-commerce site. We think of our demo as a “concept car” to showcase the power of our platform; but components of this are already live today at some of Seaplane’s customers.

Let’s grab our snow boots and walk into a Virtual Snowboard Shop, shall we?

Imagine a traditional shopping experience at a brick-and-mortar store for a snowboard. When you're looking for a new snowboard, an associate helps you find the perfect match by suggesting alternatives based on your preferences. We aim to recreate this personalized experience on Double Black Snowboards' web store, leveraging the Seaplane Hybrid AI Recommendation engine, acting as a knowledgable sales agent who only has your best interests at heart.

Hyper-Personalized Product Discovery

Upon arriving at the website, users interact with a digital, AI-driven shopping clerk, supplying valuable insights into their preferences (name, weight, gender, proficiency, mountain types, and even favorite resorts and lodges). This information is then leveraged to populate the product page with the most relevant items, accompanied by hyper-personalized suggestions, with every description unique. Each user gets a detailed description of the board tailored to their specific interests and skill levels, generated by a low-latency LLM. The product recommendations continue to evolve as users explore the site, drawing from various data sources such as similar user behavior, historical transactions, and current inventory.

Seamless Customer Engagement

When users find a product they like, they can tell us by engaging with our AI expert through a chatbot. This chat interface initially utilizes a RAG-enabled low-latency LLM (Zephyr 7b) to provide real-time answers. Simultaneously, a recommendation request is sent to our powerful Llama2-70B parameter model, considering factors such as browsing behavior, intake form responses, and similar users, to deliver three super-relevant recommendations.

Highlighting Seaplane's Key Values for Our Customers

Our solution boasts several key features that set Seaplane apart, all wrapped in an intuitive and high-level Developer Experience [DX], making AI-infusion accessible to anyone:

  1. Global Infrastructure (including edge GPUs): Serving users across the globe with siloed data, ensuring low-latency performance, compliance and data sovereignty with Seaplane’s geofencing capabilities.
  2. Model Flexibility: From high-quality LLMs (Llama 2-70B) to low-latency edge LLMs (Zephyr 7B and Mistrial 7B Instruct), this hybrid-AI solution showcases the simplicity of leveraging Foundational Models and open-source models on the edge for best price-performance.
  3. Application Runtime: Integrated, managed data services including: vector databases, SQL services, and local object storage worldwide make deploying and operating a RAG implementation at scale wildly simple.

Unlock the Full Potential of Seaplane

To delve deeper into the capabilities of the Seaplane Hybrid AI Recommendation Engine demo, we invite you to follow us on socials (LinkedIn and X-Twitter); engage in a Q/A on our Discord page; or sign up for the beta. Shortly, we will release a YouTube video of the demo, technical blogs, and even the source-code as a reference design to help you incorporate this concept into your Shopify website in the future. Seaplane's LLM recommendation engine could be a new powerhouse for your e-commerce site and we would love to help you deliver it.

Join us in reshaping the online shopping landscape, one personalized recommendation at a time. The possibilities are limitless, and the future is here with Seaplane as we continue to help AI take flight in 2024.

Sign up for the beta here!

Subscribe to our newsletter
Please insert valid email address

Bring Your Apps to the World

Join the Seaplane Beta