Task processing and execution using large language models

Patent metadata

Field	Value
Publication number	US-2025298990-A1
Application number	US-202418611159-A
Country	US
Kind code	A1
Filing date	Mar 20, 2024
Priority date	Mar 20, 2024
Publication date	Sep 25, 2025
Grant date	—

Abstract

One example method includes receiving, by an artificial intelligence (“AI”) assistant, a user query comprising one or more tasks; determining one or more services based on the user query; obtaining, for each of the one or more services, a plurality of examples, each example providing an example command suitable for execution by the respective service; generating one or more instructions based on the user query, the one or more services, and the one or more pluralities of examples; providing the one or more instructions to a trained large language model (“LLM”); receiving, from the LLM, one or more commands corresponding to the user query; for each command of the plurality of commands, issuing the respective command to a corresponding service of the one or more services; generating a response to the user query based on results of the plurality of commands; and outputting the response

First claim

That which is claimed is:

1 . A method comprising:
receiving, by an artificial intelligence (“AI”) assistant, a user query comprising one or more tasks; determining one or more services based on the user query; obtaining, for each of the one or more services, a plurality of examples, each example providing an example command suitable for execution by the respective service; generating one or more instructions based on the user query, the one or more services, and the one or more pluralities of examples; providing the one or more instructions to a trained large language model (“LLM”); receiving, from the LLM, one or more commands corresponding to the user query; for each command of the one or more commands, issuing the respective command to a corresponding service of the one or more services; generating a response to the user query based on results of the one or more commands; and outputting the response.

2 . The method of claim 1 , further comprising:
generating, using a trained ML model, one or more first embeddings based on the user query; generating, using the trained ML model, one or more second embeddings based on descriptions of the one or more services; and wherein determining the one or more services is based on the one or more first embeddings and the one or more second embeddings.

3 . The method of claim 2 , further comprising:
determining, for each service, a confidence based on the one or more first embeddings and one or more second embeddings corresponding to the respective service; and determining the one or more services based on the respective confidences and a confidence threshold.

4 . The method of claim 2 , further comprising:
determining, for each service, a confidence based on the one or more first embeddings and one or more second embeddings corresponding to the respective service; and wherein obtaining, for each of the one or more services, the plurality of examples is based on the respective confidence for the respective service.

5 . The method of claim 1 , further comprising:
for each example, generating a relevancy based on the user query; and wherein obtaining, for each of the one or more services, the plurality of examples is based on the respective relevancies.

6 . The method of claim 5 , further comprising:
generating, using a trained ML model, one or more first embeddings based on the user query; generating, using the trained ML model, one or more second embeddings based on the pluralities of examples; and wherein generating the relevancy is based on at least a subset of the one or more first embeddings and the respective one or more second embeddings associated with the respective example.

7 . The method of claim 1 , wherein the issuing the respective command is performed by the AI assistant.

8 . The method of claim 1 , wherein the issuing the respective command is performed by the trained LLM.

9 . A system comprising:
a non-transitory computer-readable medium; and one or more processors communicatively connected to the non-transitory computer-readable medium, the one or more processors configured to execute processor-executable instructions stored in the non-transitory computer-readable medium to cause the one or more processors to:
receive, by an artificial intelligence (“AI”) assistant, a user query comprising one or more tasks;
determine one or more services based on the user query;
obtain, for each of the one or more services, a plurality of examples, each example providing an example command suitable for execution by the respective service;
generate one or more instructions based on the user query, the one or more services, and the one or more pluralities of examples;
provide the one or more instructions to a trained large language model (“LLM”);
receive, from the LLM, one or more commands corresponding to the user query;
for each command of the one or more commands, issue the respective command to a corresponding service of the one or more services;
generate a response to the user query based on results of the one or more commands; and
output the response.

10 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to:
generate, using a trained ML model, one or more first embeddings based on the user query; generate, using the trained ML model, one or more second embeddings based on descriptions of the one or more services; and wherein determining the one or more services is based on the one or more first embeddings and the one or more second embeddings.

11 . The system of claim 10 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to:
determine, for each service, a confidence based on the one or more first embeddings and one or more second embeddings corresponding to the respective service; and determine the one or more services based on the respective confidences and a confidence threshold.

12 . The system of claim 10 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to:
determine, for each service, a confidence based on the one or more first embeddings and one or more second embeddings corresponding to the respective service; and wherein obtaining, for each of the one or more services, the plurality of examples is based on the respective confidence for the respective service.

13 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to:
for each example, generate a relevancy based on the user query; and wherein obtaining, for each of the one or more services, the plurality of examples is based on the respective relevancies.

14 . The system of claim 13 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to:
generate, using a trained ML model, one or more first embeddings based on the user query; generate, using the trained ML model, one or more second embeddings based on the pluralities of examples; and wherein generating the relevancy is based on at least a subset of the one or more first embeddings and the respective one or more second embeddings associated with the respective example.

15 . The system of claim 9 , wherein the issuing the respective command is performed by the AI assistant.

16 . The system of claim 9 , wherein the issuing the respective command is performed by the trained LLM.

17 . A non-transitory computer-readable medium comprising processor-executable instructions configured to cause one or more processors to:
receive, by an artificial intelligence (“AI”) assistant, a user query comprising one or more tasks; determine one or more services based on the user query; obtain, for each of the one or more services, a plurality of examples, each example providing an example command suitable for execution by the respective service; generate one or more instructions based on the user query, the one or more services, and the one or more pluralities of examples; provide the one or more instructions to a trained large language model (“LLM”); receive, from the LLM, one or more commands corresponding to the user query; for each command of the one or

Assignees

• Zoom Video Communications Inc

Inventors

• Geng Yuanling
• Lee Bilung
• Tao Renjie

Classifications

• G06F40/40 Primary

  Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30) · CPC title

• G06F16/33295 Primary

  in dialogue systems · CPC title