Granted Patent B1

US 12,436,957 · App. 18/785,964 · Granted Oct 7, 2025

Inventors: Samrendra Kumar (Perumbakkam, IN); Karthik Mohan (Scarborough, CA); Frenchy George Thomas (Chennai, IN); Rohit Khanna (Cranbury, NJ); Leslie Dias (West Windsor, NJ); Winnie Chacko (Dix Hills, NY)

Assignee: Bank of America Corporation

G06F16/24575G06F16/2425G06F16/24539

Abstract

Claims (121)

1. A computing platform comprising:

at least one processor;

a communication interface communicatively coupled to the at least one processor; and

a memory storing computer-readable instructions that, when executed by the at least one processor, configure the computing platform to:

train, based on historical query information, a query response model, wherein training the query response model configures the query response model to output query responses based on input of query prompts;

receive, from a user device, a user query;

generate, based on one or more historical embeddings, an embedding corresponding to the user query;

generate a query prompt corresponding to the user query, wherein the query prompt comprises:

the embedding corresponding to the user query; and

a plurality of embeddings corresponding to a subset of the historical query information;

execute, using as inputs the query prompts, the query response model to output a query response;

send, to the user device, the query response, wherein sending the query response to the user device causes display of a user interface including the query response on a display of the user device;

identify, based on sending the query response, a status of the query response indicating whether the query response satisfies the user query, and in response:

based on identifying that the query response satisfies the user query, store, to a long-term memory component, the query response, or

based on identifying that the query response does not satisfy the user query, output an alternative query response; and

update, based on the query response and the status of the query response, the query response model.

2. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by at least one processor, further configure the computing platform to:

output, to a second user device and based on identifying that the status of the query response indicates that the query response does not satisfy the user query, the query prompt;

receive, from the second user device, a modified query response; and

output, to the user device, the modified query response.

3. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, further configure the computing platform to:

identify, based on identifying that the status of the query response indicates that the query response does not satisfy the user query, the alternative query response,

wherein identifying the alternative query response comprises selecting, from one or more historical query responses stored at the long-term memory component, a historical query response corresponding to the user query.

4. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by at least one processor, further configure the computing platform to:

maintain, at the long-term memory component, one or more historical query responses,

wherein executing the query response model to output the query response comprises identifying, based on one or more stored correlations to the one or more historical query responses, a historical query response corresponding to the user query.

5. The computing platform of claim 1 , wherein the subset of the historical query information comprises one or more of the following:

text inputs,

chat logs,

photographic images,

audio files,

historical data logs, or

digital files.

6. The computing platform of claim 1 , wherein the user query comprises one or more of the following:

text inputs,

chat logs,

photographic images,

audio files,

historical data logs, or

digital files.

7. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by at least one processor, further configure the computing platform to:

maintain, at the long-term memory component:

one or more historical embeddings corresponding to historical user queries, and

one or more historical query prompts,

wherein generating the embedding corresponding to the user query comprises selecting, from the one or more historical embeddings, a historical embedding corresponding to the user query, and

wherein generating the query prompt comprises selecting, from the one or more historical query prompts, a query prompt corresponding to the historical embedding.

8. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by at least one processor, further configure the computing platform to:

generate, based on the historical query information, the one or more historical embeddings;

identify, based on the one or more historical embeddings, one or more categories of historical query information; and

store, at a long-term memory component and based on the one or more categories of historical query information, one or more correlations between the one or more categories and the one or more historical embeddings.

9. The computing platform of claim 1 , wherein identifying the status of the query response comprises:

receiving, from the user device and via the user interface, user feedback; and

identifying, based on the user feedback, whether the query response satisfies the user query.

10. The computing platform of claim 1 , wherein the memory stores additional computer-readable instructions that, when executed by at least one processor, further configure the computing platform to:

receive, from one or more sources of industry information, wherein the industry information comprises one or more of:

policy information corresponding to one or more regulatory entities, or

cybersecurity information; and

update, based on the industry information, the query response model.

11. The computing platform of claim 1 , wherein the query response comprises at least one of:

information responsive to the user query, wherein the information responsive to the user query comprises at least one of:

account information,

regulatory information, or

asset information, or

a visual representation of the query response, wherein the visual representation comprises:

a text output;

an email;

a chat response;

a graphical representation;

an audio file; or

a data log.

12. A method comprising:

at a computing device comprising at least one processor, a communication interface, and memory:

training, based on historical query information, a query response model, wherein training the query response model configures the query response model to output query responses based on input of query prompts;

receiving, from a user device, a user query;

generating, based on one or more historical embeddings, an embedding corresponding to the user query;

generating a query prompt corresponding to the user query, wherein the query prompt comprises:

the embedding corresponding to the user query; and

a plurality of embeddings corresponding to a subset of the historical query information;

executing, using as inputs the query prompts, the query response model to output a query response;

sending, to the user device, the query response, wherein sending the query response to the user device causes display of a user interface including the query response on a display of the user device;

identifying, based on sending the query response, a status of the query response indicating whether the query response satisfies the user query, and in response:

based on identifying that the query response satisfies the user query, store, to a long-term memory component, the query response, or

based on identifying that the query response does not satisfy the user query, output an alternative query response; and

updating, based on the query response and the status of the query response, the query response model.

13. The method of claim 12 , further comprising:

maintaining, at the long-term memory component, one or more historical query responses,

wherein executing the query response model to output the query response comprises identifying, based on one or more stored correlations to the one or more historical query responses, a historical query response corresponding to the user query.

14. The method of claim 12 , further comprising:

outputting, to a second user device and based on identifying that the status of the query response indicates that the query response does not satisfy the user query, the query prompt;

receiving, from the second user device, a modified query response; and

outputting, to the user device, the modified query response.

15. The method of claim 12 , further comprising:

identifying, based on identifying that the status of the query response indicates that the query response does not satisfy the user query, the alternative query response,

wherein identifying the alternative query response comprises selecting, from one or more historical query responses stored at the long-term memory component, a historical query response corresponding to the user query.

16. The method of claim 12 , further comprising:

receiving, from one or more sources of industry information, wherein the industry information comprises one or more of:

policy information corresponding to one or more regulatory entities, or cybersecurity information; and

updating, based on the industry information, the query response model.

17. One or more non-transitory computer-readable media storing instructions that, when executed by a computing platform comprising at least one processor, a communication interface, and memory, cause the computing platform to:

train, based on historical query information, a query response model, wherein training the query response model configures the query response model to output query responses based on input of query prompts;

receive, from a user device, a user query;

generate, based on one or more historical embeddings, an embedding corresponding to the user query;

generate a query prompt corresponding to the user query, wherein the query prompt comprises:

the embedding corresponding to the user query; and

a plurality of embeddings corresponding to a subset of the historical query information;

execute, using as inputs the query prompts, the query response model to output a query response;

send, to the user device, the query response, wherein sending the query response to the user device causes display of a user interface including the query response on a display of the user device;

identify, based on sending the query response, a status of the query response indicating whether the query response satisfies the user query, and in response:

based on identifying that the query response satisfies the user query, store, to a long-term memory component, the query response, or

based on identifying that the query response does not satisfy the user query, output an alternative query response; and

update, based on the query response and the status of the query response, the query response model.

18. The one or more non-transitory computer-readable media of claim 17 , wherein the instructions, when executed, further cause the computing platform to:

maintain, at the long-term memory component, one or more historical query responses,

wherein executing the query response model to output the query response comprises identifying, based on one or more stored correlations to the one or more historical query responses, a historical query response corresponding to the user query.

19. The one or more non-transitory computer-readable media of claim 17 , wherein the instructions, when executed, further cause the computing platform to:

output, to a second user device and based on identifying that the status of the query response indicates that the query response does not satisfy the user query, the query prompt;

receive, from the second user device, a modified query response; and

output, to the user device, the modified query response.

20. The one or more non-transitory computer-readable media of claim 17 , wherein the instructions, when executed, further cause the computing platform to:

identify, based on identifying that the status of the query response indicates that the query response does not satisfy the user query, the alternative query response,

wherein identifying the alternative query response comprises selecting, from one or more historical query responses stored at the long-term memory component, a historical query response corresponding to the user query.

Assignments (1)

References Cited (29)

Cited By (1)

US 12,664,376