The software used by Google to recommend new content to users (via YouTube and other services) is now in public beta and available for retailers to use on their own websites.
In a blog post published earlier this week, Google Product Manager Pallav Mehta said the tool, called Recommendations AI, will allow retailers to shift the emphasis onto each individual customer instead of the item, creating a more personalized experience for the visitor.
Retailers start by importing catalog and user events data, then the second step is to decide on a model type and optimization objective, required to train the model. According to Google, the initial training period can last as long as five days, after which the tool can begin recommending new items to returning visitors.
The tool is “context hungry”, claims Mehta, and can successfully avoid bias that usually occurs with popular, on-sale and season-dependent items.
It was also added that Recommendations AI can be used in conjunction with existing recommendation tools and doesn't demand a rip-and-replace approach.
According to ZDNet, Recommendations AI is a direct competitor to Amazon’s Personalize and Adobe’s AI-powered product recommendations tool.