US 2025/0053430 A1Published application

Large language model-based virtual assistant for high-level goal contextualized action recommendations

AssigneeMeta Platforms Technologies, LLC

InventorsBrennan David Gorham Jones, Yan Xu, Hamid Eghbalzadeh

Life of this patent

Estimating the term…

Loading patent…

Abstract

The present disclosure relates to using a large language model (LLM), provided with user context information and user high-level goal information to generate contextualized action recommendations that can help the user achieve the high-level goal(s). In one exemplary embodiment, a user system, and an AI action recommendation system that is associated with an LLM, are communicatively coupled and cooperatively implement a contextualized action recommendation virtual assistant. Input data comprising personal information data of the user that includes at least one high-level goal of the user, and user context data obtained from the user system can be collected and used to generate a prompt that is input to the LLM. The LLM can then generate a contextualized action recommendation for the user based on the prompt, and the contextualized action recommendation can be presented to the user via a virtual assistant user interface on a display of the user system.

Claims20

Dependent claims

A contextualized action recommendation virtual assistant comprising:

a user system comprising a display to display content to a user, one or more sensors to capture input data, and a virtual assistant application;
an AI action recommendation system that is associated with a large language model and includes a virtual assistant engine that is cooperative with the virtual assistant application of the user system to implement the virtual assistant;
one or more processors;
and
one or more memories accessible to the one or more processors, the one or more memories storing a plurality of instructions executable by the one or more processors, the plurality of instructions comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform processing comprising:
collecting input data comprising personal information data of the user that includes at least one high-level goal of the user, and user context data from the one or more sensors of the user system;
generating, using the input data, a prompt for the large language model;
inputting the prompt to the large language model;
generating, by the large language model, a contextualized action recommendation for the user based on the prompt, wherein the contextualized action recommendation is predicted to help the user achieve the at least one high-level goal;
and
presenting the contextualized action recommendation to the user via a virtual assistant user interface on the display of the user system.

8 more elements

A non-transitory computer-readable memory storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to:

implement a virtual assistant through a user system comprising a display to display content to a user, one or more sensors to capture input data, and a virtual assistant application, in combination with an AI action recommendation system that is associated with a large language model and includes a virtual assistant engine that is cooperative with the virtual assistant application of the user system to implement the virtual assistant;
collect input data comprising personal information data of the user that includes at least one high-level goal of the user, and data from the one or more sensors of the user system that indicates a user context;
generate, using the input data, a prompt for the large language model;
input the prompt to the large language model;
generate, by the large language model, a contextualized action recommendation for the user based on the prompt, wherein the contextualized action recommendation is predicted to help the user achieve the at least one high-level goal;
and
present the contextualized action recommendation to the user via a virtual assistant user interface on the display of the user system.

4 more elements

Dependencies are inferred from the claim text.