US-12627475-B2 - Encrypted context-based prompt system

US12627475B2US 12627475 B2US12627475 B2US 12627475B2US-12627475-B2

Abstract

Disclosed are various embodiments for an encrypted context-based prompt system for interactions between multiple entities. In one embodiment, a system comprising a computing device configured to receive a request to activate a prompt session. The request comprises a template identifier representing a type of interaction between a first device and a second device. A decrypted term is determined by decrypting an encrypted term associated with the request. A large language model prompt is generated based at least in part on the template identifier and the decrypted term. A proposed term is generated by executing a large language model application based at least in part on the large language model prompt. An encrypted proposed term is transmitted to the first client device or the second client device.

Inventors

Alaric M. Eby
ANDRAS L. FERENCZI
Hilary Packer

Assignees

AMERICAN EXPRESS TRAVEL RELATED SERVICES COMPANY, INC.

Dates

Publication Date: 20260512
Application Date: 20240530

Claims (20)

1 . A system, comprising: a computing device comprising a processor and a memory; and machine-readable instructions stored in the memory that, when executed by the processor, cause the computing device to at least: receive a request to activate a prompt session for a first client device or a second client device, the request comprising a template identifier representing a type of interaction between the first client device and the second client device; determine a decrypted term by decrypting an encrypted term associated with the request based at least in part on a symmetric key associated with the first client device or the second client device; generate a large language model prompt based at least in part on the template identifier and the decrypted term; generate a proposed term by executing a large language model application based at least in part on the large language model prompt, the proposed term being encrypted using a shared encryption key to generate an encrypted proposed term; and transmit the encrypted proposed term to the first client device or the second client device.
2 . The system of claim 1 , wherein the machine-readable instructions, when executed by the processor, cause the computing device to at least: identify a registration request for the first client device and the second client device; receive the symmetric key from the first client device or the second client device; and associate the symmetric key with a user profile for decrypting data associated with the first client device or the second client device.
3 . The system of claim 1 , wherein the shared encryption key is generated using a Diffie-Hellman Key Exchange Protocol.
4 . The system of claim 1 , wherein generating the large language model prompt further comprises inserting the decrypted term into a placeholder for a text prompt associated with the template identifier.
5 . The system of claim 1 , wherein the template identifier is associated with a data structure that comprises a first entity term, a second entity term, and a text prompt, the text prompt comprising a first instruction for concealing the first entity term from the second client device and a second instruction for concealing the second entity term from the first client device.
6 . The system of claim 1 , wherein the machine-readable instructions, when executed by the processor, cause the computing device to at least: receive a rejection of the encrypted proposed term from the first client device or the second client device; and transmit a request for an alternative term to the first client device or the second client device.
7 . The system of claim 1 , wherein the encrypted term comprises a first encrypted term and a second encrypted term, and the machine-readable instructions, when executed by the processor, cause the computing device to at least: identify a request to initiate a context interaction for the first client device and the second client device; and transmit a request for a public key to the first client device or the second client device based at least in part on the request to initiate the context interaction.
8 . A method, comprising: identifying, by a computer device, a request to active a prompt session, the requesting comprising a template identifier and an encrypted term; determining, by the computing device, a decrypted term by decrypting the encrypted term based at least in part on a symmetric key associated with a first client device or a second client device; generating, by the computing device, a large language model prompt based at least in part on the decrypted term and the template identifier; generating, by the computing device, an encrypted proposed term based at least in part on inputting the large language model prompt to the large language mode to generate a proposed term, the proposed term being encrypted using a shared encryption key to generate the encrypted proposed term; and transmitting, by the computing device, the encrypted proposed term to the first client device or the second client device.
9 . The method of claim 8 , further comprising: identifying, by the computing device, a registration request for the first client device and the second client device; receiving, by the computing device, the symmetric key from the first client device or the second client device; and associating, by the computing device, the symmetric key with a user profile for decrypting data associated with the first client device or the second client device.
10 . The method of claim 8 , wherein the shared encryption key is generated using a Diffie-Hellman Key Exchange Protocol.
11 . The method of claim 8 , wherein generating the large language model prompt further comprises inserting the decrypted term into a text prompt associated with the template identifier.
12 . The method of claim 8 , wherein the template identifier is associated with a data structure that comprises a first entity term, a second entity term, and a text prompt, the text prompt comprising a first instruction for concealing the first entity term from the second client device and a second instruction for concealing the second entity term from the first client device.
13 . The method of claim 8 , further comprising: receiving, by the computing device, a rejection of the encrypted proposed term from the first client device or the second client device; and transmitting, by the computing device, a request for an alternative term to the first client device or the second client device.
14 . The method of claim 8 , wherein the encrypted term comprises a first encrypted term and a second encrypted term, and further comprising: identifying, by the computing device, a request to initiate a context interaction for the first client device and the second client device; and transmitting, by the computing device, a request for a public key to the first client device or the second client device based at least in part on the request to initiate the context interaction.
15 . A non-transitory, computer-readable medium, comprising machine-readable instructions that, when executed by a processor of a first client device, cause the first client device to at least: identify a request to initiate a context interaction with a second client device, the request being associated with a template identifier that represents a type of interaction between the first client device and the second client device; generate an encrypted term for the template identifier based at least in part on an input received from a user interface and a symmetric key associated with the first client device; generate a shared encryption key based at least in part on a private key associated with the first client device and a second public key associated with the second client device; and transmit a prompt package to a large language model application for generating a proposed term for the first client device and the second client device, the prompt package comprising the shared encryption key, the template identifier, the encrypted term, and an identifier for the second client device.
16 . The non-transitory, computer-readable medium of claim 15 , wherein the machine-readable instructions that, when executed by the processor of the first client device, cause the first client device to at least: generate the private key and a first public key associated with the first client device in response to receiving the request to initiate the context interaction; and transmit the first public key to the second client device.
17 . The non-transitory, computer-readable medium of claim 15 , wherein generating the encrypted term for the template identifier further causes the first client device to encrypt the input using the symmetric key associated with the first client device, the input comprising a first party term for the context interaction.
18 . The non-transitory, computer-readable medium of claim 15 , wherein the template identifier is associated with a data structure that comprises a first placeholder, a second placeholder, and a text prompt, the text prompt comprising a first instruction for concealing data entered for the first placeholder from the second client device.
19 . The non-transitory, computer-readable medium of claim 18 , wherein the template identifier further comprises a second instruction for concealing data entered for the second placeholder from the first client device.
20 . The non-transitory, computer-readable medium of claim 15 , wherein the machine-readable instructions, when executed by the processor, further cause the first client device to at least: display a user interface for selecting the context interaction and the identifier for the second client device, wherein the selection of the context interaction and the identifier for the second client device generates the request to initiate the context interaction with the second client device.

Description

BACKGROUND During a negotiation, opposing parties can spend a significant amount of time presenting offers and counter offers. Various factors can affect the amount of time taken to negotiate terms for an agreement between parties, such as the time available for the parties, the methods used for communicating the offers, the number of terms that need to be agreed upon, and other suitable factors. BRIEF DESCRIPTION OF THE DRAWINGS Many aspects of the present disclosure can be better understood with reference to the following drawings. The components in the drawings are not necessarily to scale, with emphasis instead being placed upon clearly illustrating the principles of the disclosure. Moreover, in the drawings, like reference numerals designate corresponding parts throughout the several views. FIG. 1 is a drawing depicting a scenario of an application being used to negotiate the terms for a car sale according to various embodiments of the present disclosure. FIG. 2 is a drawing of a network environment according to various embodiments of the present disclosure. FIGS. 3A and 3B are drawings of a user interface illustrating a large language model application being used to generate terms for a context interaction according to various embodiments of the present disclosure. FIG. 4 is a sequence diagram illustrating one example of functionality implemented as portions of an application executed in a computing environment in the network environment of FIG. 2 according to various embodiments of the present disclosure. FIG. 5 is a flowchart illustrating one example of functionality implemented as portions of a service executed in a computing environment in the network environment of FIG. 2 according to various embodiments of the present disclosure. FIGS. 6A and 6B are flowcharts illustrating one example of functionality implemented as portions of an application executed in a client device in the network environment of FIG. 2 according to various embodiments of the present disclosure. DETAILED DESCRIPTION The various embodiments of the present disclosure are directed to an encrypted context-based prompt system for determining terms between multiple parties for a context-based interaction. Often, during a negotiation, opposing parties spend a significant amount of time presenting offers and counter offers in order to obtain favorable terms. Various circumstances and factors can affect the amount of time taken to negotiate terms between parties. These factors can include the availability of participants, the methods used for communicating the offers, the location of the participants, the complexity of the terms, the quantity of terms, and other suitable factors. The embodiments of the present disclosure include systems and methods for automatically determining negotiated terms between multiple parties without revealing each party's constraints and/or expectations for the interaction. For example, a car salesperson and a car buyer can use a software application for negotiating the terms of a car purchase. The car salesperson and the car buyer can enter their constraints within the application. Some illustrative and non-limiting constraints in this example can include the car buyer's maximum walk away price, the car seller's lowest available sales price, and/or the car seller's available financing options for the car buyer, etc. Without revealing each party's constraints to the opposing side, the software application can automatically determine one or more terms that would satisfy the constraints of both parties. The software application can be used for a wide variety of scenarios because the software application can identify a context interaction or negotiation scenario between the parties. From the identification of the context interaction, the software application can identify applicable rules, input data needed for placeholders associated with an interaction template for the context interaction, and other suitable data. In some examples, these identified components can be used to generate a large language model prompt. A large language model application can receive the generated large language model prompt in order to negotiate terms. As such, different identified context interactions can cause the generation of different large language model prompts. One or more encryption layers can be used to conceal data associated with the constraints or private terms for each party and to conceal the proposed terms generated by the large language model for the interaction from unauthorized users. Accordingly, various embodiments of the present disclosure provide various advantages. For example, various embodiments can include an implementation of an application protocol for discretely determining an agreement of terms and/or conditions between multiple client devices, which each have distinct constraints that need to be satisfied. The application protocol can improve a user's experience because the application protocol can re