Commerce Architects - Conversational Cart

The Conversational Cart provides the building blocks for creating an AI-enabled shopping experience using commercetools and ChatGPT. In this case we created a virtual assistant to guide a user through a shopping experience, but these tools can be used to tackle a number of use cases.

AI-powered ecommerce using an LLM
The conversational cart accelerator example allows a user to simply state what they want to buy, and then see products associated with their statement or question. For example, “What should I make for Thanksgiving dinner?” will be transcribed from voice to text, and passed to an LLM which will understand the user is searching for traditional Thanksgiving items. This will invoke a search against the commercetools search API with some popular Thanksgiving meal ingredients, and render the results to the user. Then the user can say “add the turkey, potatoes, cranberry sauce and pumpkin pie”. The Conversational Cart will add those items to the cart, and let the user know with a voice confirmation. The accelerator example includes:

• Function calls for invoking commercetools APIs for adding to cart, removing from cart, clearing cart, searching for items

• Code to set the context with the LLM for the type of shopping experience

• Code for interacting with voice-to-text and text-to-voice tools

• React frontend that pairs with backend code using tRPC

• Documentation for how this all fits together

Accelerator Implementation
We expect the Conversational Cart to be modified and integrated into an existing commercetools application. This is not a “click and play” solution. We are happy to engage with a commercetools merchant to integrate the accelerator example and create the AI-powered shopping experience. 

There are many potential applications of this general pattern. We welcome the opportunity to brainstorm ideas, and make them come to life.