LLM Intent Detection for Unstructured Data Transfer

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data transfer processes are cumbersome, time-consuming, and inaccessible to users lacking computer literacy, particularly affecting vulnerable social groups, requiring manual navigation and accurate entry of login details and recipient information.

Innovation Solution

A computer system utilizing a Large Language Model (LLM) to identify the intent to transfer data from unstructured text, generating a standardized transfer message with elements like primary and recipient account numbers and transfer amounts, facilitated by a prompt engine and LLM API, with optional client confirmation and additional data requests.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual navigation and data entry processes are used for data transfer, then data transfer can be effected, but the process becomes time-consuming and onerous

Engineering Contradiction:

Improvedata transfer speedVSAvoidtime required for data transfer

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-storing user account information, login credentials, and recipient details in a database. When a user expresses transfer intent through natural language, the system retrieves pre-stored information to automatically populate transfer fields, eliminating the need for manual navigation and data entry during the transfer process

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system enables self-service by allowing users to declare transfer intent through natural language communication (voice or text). The system then autonomously retrieves stored information, populates transfer fields, and executes the transfer without requiring user navigation through application interfaces or manual data entry, making the process accessible to users with varying computer literacy levels

Inventive Principle:

Principle #25Self-service

2Adaptability or versatility

If manual navigation and data entry are required, then data transfer can be completed, but accessibility is reduced for users lacking computer literacy

Engineering Contradiction:

Improveaccessibility to vulnerable social groupsVSAvoiduser effort required

Core Design Contradiction:

Adaptability or versatilityVSEase of operation

Solution Approach 1:

The system replaces the mechanical interaction model (manual navigation through graphical interfaces, clicking buttons, typing in fields) with a natural language processing model. Users can express transfer intent through voice calls, teleconferencing sessions, or text chats, and the system interprets these unstructured inputs to execute transfers, making the system accessible to users with limited computer literacy, illiteracy, or disabilities

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system achieves universality by supporting multiple input modalities (voice, text chat, teleconferencing) and serving diverse user groups including those with varying computer literacy levels, language capabilities, and physical abilities. The natural language processing capability allows the same system to serve users with ADHD, avolition, illiteracy, or limited English proficiency through their preferred communication method

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250321988A1Systems and methods for generating transfer messages based on unstructured data

Publication Date: 2025.10.16 THE TORONTO DOMINION BANK

AI summary

Systems and methods for sending a transfer message based on unstructured text data are disclosed. A method may receive unstructured text data associated with an account, and based on the unstructured text data, identify an intent to transfer data. The unstructured text data may then be sent to a Large Language Model (LLM) via a first prompt engine and an LLM Application Programming Interface (API). First LLM output data may then be received form the LLM, and the transfer message may be sent based on the first LLM output data. The LLM may be a type of artificial intelligence model designed to understand and generate natural-language input.