Amazon Connect enhances AI-powered predictive insights

Amazon Connect has announced new enhancements to its AI-powered predictive insights, aimed at helping businesses deliver more personalized and proactive customer experiences on a larger scale. This development builds on the five recommendation algorithms introduced at re:Invent 2025. The improvements include support for up to 40 million product catalog items, an eightfold increase, and integration with message templates for trigger-based campaigns. Additionally, the updates promise up to a 14% improvement in model accuracy.

With these enhancements, businesses can automatically engage customers with timely and relevant messages, significantly reducing the time and effort needed to implement AI-driven personalization. The system now allows for trigger-based campaigns that initiate personalized outreach based on customer behavior and predictive signals. For instance, businesses can send product recommendations when a customer abandons their shopping cart or offer complementary services following a purchase.

Moreover, the enhanced model accuracy and reduced training time enable businesses to deploy personalized experiences more quickly and with greater confidence in the recommendations provided. Companies can now target specific customer cohorts with campaigns tailored to predicted preferences and behaviors.

The Amazon Connect Customer Profiles feature offers a pay-as-you-go model, meaning businesses only pay for the profiles they utilize. The public preview of these AI-powered predictive insights enhancements is available in several regions, including Europe (Frankfurt), US East (N. Virginia), Asia Pacific (Seoul and Tokyo), US West (Oregon), and Asia Pacific (Singapore and Sydney), as well as Canada (Central).

For more information, businesses can visit the Customer Profiles webpage and delve into the AI-powered predictive insights documentation.