Dynamic LLM Prompt Retraining for Open Banking Data Aggregation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing open banking services face challenges in efficiently aggregating and managing data due to the lack of automated, dynamic, and adaptable artificial intelligence solutions, leading to manual intervention and inefficiencies in data management and decision-making.

Innovation Solution

A computer-implemented method and system using a dynamic large language model (LLM) to automatically adapt to open banking system objectives, involving generating prompts, evaluating outputs, and iteratively retraining the LLM to improve performance characteristics, thereby enhancing data intake, aggregation, and analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If manual intervention is used for data aggregation and decision-making in open banking services, then flexibility and adaptability are maintained, but productivity and efficiency deteriorate due to time-consuming manual processes

Engineering Contradiction:

ImproveadaptabilityVSAvoidproductivity

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system enables self-service through automated LLM agents that independently perform data aggregation, evaluation, and decision-making tasks without requiring manual human intervention. The agents autonomously navigate web pages, extract financial data, evaluate outputs against performance characteristics, and iteratively improve through self-directed retraining, thereby maintaining adaptability while significantly improving productivity

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical processes with an automated computational system based on large language models. The LLM-based agents substitute human operators in performing data aggregation, evaluation, and decision-making tasks, transforming manual adaptive processes into automated intelligent systems that maintain flexibility while eliminating time-consuming manual operations

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Extent of automation

If existing AI solutions are used for open banking services, then automation is partially achieved, but adaptability to open banking data structures and vernacular deteriorates because existing AI solutions are not adapted for such uses

Engineering Contradiction:

ImproveautomationVSAvoidadaptability

Core Design Contradiction:

Extent of automationVSAdaptability or versatility

Solution Approach 1:

The system implements dynamics through iterative retraining mechanisms where LLM agents continuously adapt to open banking data structures and vernacular. The agents evaluate their outputs against performance characteristics and automatically retrain on new data, creating a dynamic system that evolves and improves its adaptability to domain-specific requirements while maintaining high automation levels

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent applies parameter changes by modifying the LLM agents through iterative retraining processes. The system changes the internal parameters and knowledge structures of the LLMs by training them on open banking-specific data, thereby adapting general-purpose AI solutions to domain-specific requirements while preserving the automated operation of the agents

Inventive Principle:

Principle #35Parameter changes

3Productivity

If data aggregation is performed with high efficiency, then productivity improves, but measurement precision deteriorates due to the inherent structure and lack of structure in open banking data

Engineering Contradiction:

ImproveproductivityVSAvoidmeasurement precision

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system implements feedback mechanisms where LLM agents evaluate their data aggregation outputs against defined performance characteristics. This feedback loop allows the system to identify precision issues in aggregated data and trigger iterative retraining of the LLM agents, thereby maintaining high productivity while continuously improving measurement precision through automated correction of data structure challenges

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20260050970A1Computer-implemented methods, systems comprising computer-readable media, and electronic devices for providing aggregation large language model dynamic open banking services

Publication Date: 2026.02.19 MASTERCARD INT INC

AI summary

A computer-implemented method for providing dynamic LLM open banking services that includes: generating a predefined training action and a predefined prompt modification for FI feed data prompts and storing with respective metadata configured for matching against values for performance characteristics of the LLM; generating an output based on a first prompt for FI feed data to the LLM; evaluating the output against the predefined performance characteristics to generate values; matching the values to the predefined training action and prompt modification using the metadata; based on the predefined training action, curating a training data set and retraining the LLM thereon to generate a retrained LLM; based on the predefined prompt modification, generating a second prompt seeking FI feed data; and generating a second output based on the second prompt to the retrained LLM.