LLM-Graph-of-Thoughts leveraging unified data access across diverse GraphQL APIs

Inventors: Murali Krishna Akula (Hyderabad Telangana, IN); Uttam Dey (Charlotte, NC); Durga Prasad Kutthumolu (Hyderabad Telangana, IN)

Assignee: Bank of America Corporation

G06F16/215 G06F16/24522

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12,386,799
App. No.: 18/748,241
Granted: Aug 12, 2025
Kind: B1

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

Methods and systems for synergistic use of a large language model (“LLM”) and Graph-of-Thoughts (“GoT”) for unified data access across GraphQL application protocol interfaces (“APIs”) via a single API. Methods and systems may include requesting to identify and retrieve data concerning GraphQL APIs from a GraphQL API schema hub. Methods and systems may include identifying key relationships between data elements in the GraphQL APIs using relationship semantic analysis and transporting a text prompt via a cache to an LLM-GOT synergistic processor. Methods and systems may include modeling information generated by the LLM-GoT synergistic processor as an arbitrary graph, generating a unified GraphQL API based on synergistic outcomes, and storing the unified GraphQL API in the GraphQL API schema hub.

Claims (62)

1. One or more non-transitory computer-readable media storing computer-executable instructions which, when executed by a processor on a computer system, perform a method for multi-cloud synergistic GraphQL generative artificial intelligence (“AI”) that eliminates a need for manual editing of queries, the method comprising:

receiving a request from an Open Banking application programming interface (“API”) to identify GraphQL APIs;

retrieving data concerning the GraphQL APIs from a GraphQL API schema hub;

transmitting a response to the Open Banking API concerning an accessibility of the GraphQL APIs;

after the transmitting, receiving a request to generate and build a unified GraphQL API based on the retrieved GraphQL APIs;

in response to receiving the request, building the unified GraphQL API by:

searching the Open Banking API to identify data, stored on the Open Banking API, that concerns GraphQL APIs;

retrieving, from the GraphQL API schema hub, Open Banking and GraphQL API schemas related to the GraphQL APIs;

using relationship semantic analysis to identify key relationships between data elements in the GraphQL APIs;

based on the key relationships, forming a unified GraphQL API prompt by generating a text prompt;

transporting the text prompt via a cache to a large language model and Graph-of-Thoughts (“LLM-GoT”) synergistic processor, the LLM-GOT synergistic processor generating information concerning the GraphQL APIs, wherein the information concerning the GraphQL APIs generated by the LLM-GOT synergistic processor includes LLM thoughts concerning relationships between account information and transaction history, each unit of the information concerning the GraphQL APIs being a vertex in a GoT, and each edge in the GoT being a dependency between two vertices in the GoT;

modeling the information concerning the GraphQL APIs generated by the LLM-GOT synergistic processor as an arbitrary graph of nodes that includes connections between the key relationships;

combining, via the LLM-GoT synergistic processor:

distilling networks of the LLM thoughts;

enhancing the LLM thoughts using feedback loops; and

after the distilling and the enhancing, modeling the LLM thoughts, using the arbitrary graph of nodes, into synergistic outcomes;

generating the unified GraphQL API based on the synergistic outcomes; and

storing the unified GraphQL API in the GraphQL API schema hub;

receiving, from a user interface (“UI”), a query;

feeding the query to the unified GraphQL API;

receiving, from the unified GraphQL API, instructions to edit the query; and

executing the instructions to edit the query, eliminating a need to manually edit the query.

2. The media of claim 1 , wherein the GraphQL API schema hub is connected to a GraphQL database and a GraphQL server in a multi-cloud.

3. The media of claim 2 , wherein the multi-cloud is connected to financial institutions via the GraphQL server, the GraphQL server being a GraphQL API gateway.

4. The media of claim 1 , wherein the text prompt is in JavaScript Object Notation (“JSON”) format.

5. The media of claim 1 , wherein the key relationships include a relationship not explicitly defined in the GraphQL APIs.

6. The media of claim 1 , wherein the key relationships include a relationship between a financial account number and a routing number.

7. The media of claim 1 , wherein the GraphQL APIs is subjected to local traffic management.

8. The media of claim 1 , wherein the unified GraphQL API includes a zip code.

9. The media of claim 1 , wherein the unified GraphQL API replaces the GraphQL APIs.

10. The media of claim 1 , wherein the eliminating a need to manually edit the query includes reducing a quantity of API developers required.

11. A method for performing multi-cloud synergistic GraphQL generative artificial intelligence (“AI”) that eliminates a need for manual editing of queries, the method comprising: