Google's AI-based product recommendations tool is now available in beta

Google's new service that uses artificial intelligence to recommend products to online shoppers is now available in beta to everyone. Recommendations AI borrows heavily from technology used by Google to provide recommendations across its services like YouTube and Search.

The tool is designed to help retailers analyze a customer's buying history in order to provide personalized product suggestions. That means it highlights an individual customer instead of a product. In a blog post, Pallav Mehta, a Product Manager at Google, explained how the technology works:

"Recommendations AI also excels at handling recommendations in scenarios with long-tail products and cold-start users and items. Its “context hungry” deep learning models use item and user metadata to draw insights across millions of items at scale and constantly iterate on those insights in real-time in a way that is impossible for manually curated rules to keep up with."

Retailers can start implementing the tool by importing catalog and user events data when creating a Google Cloud project. After that, they can select a model type, set their optimization objective, and start training the model. The training process will last for two to five days before recommendations can be served to customers.

Alongside the beta release of Recommendations AI, Google announced a new pricing structure for the tool, including three volume-based price tiers for predictions and a separate fee for model training and tuning. The company is also giving a $600 credit to new customers who will sign up for the service. That's in addition to the regular $300 free credit for new Google Cloud customers, and the search giant says this is enough to train and evaluate a model through a two-week A/B test.

Report a problem with article
Next Article

Samsung Galaxy Note20 details leak in full, reportedly sports plastic back and flat display

Previous Article

Microsoft releases a Windows 10 Team 2020 Update preview build

0 Comments - Add comment