Language Model Prompts for Maintainable API-Driven Communications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated communications systems, such as interactive voice response systems, are tedious for end users and difficult for companies to maintain and update, especially when changes occur in the subject of communications sessions.
Innovation Solution
A computer-implemented method using a language model to automate communications sessions by creating prompts that include session text, available API calls, and instructions to select actions, determining the next action, and performing API calls or transmitting responsive communications, without relying on graphs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If an interactive voice response system uses a graph to obtain information from an end user and provide responses, then the system can automate communications sessions, but the graph becomes difficult to update and maintain when the subject of communications sessions changes
Solution Approach 1:
The patent transitions from a static graph structure to a dynamic prompt-based system using large language models. The prompts can be easily modified and updated without reconfiguring the entire system architecture, allowing the automation to adapt to changing subjects and topics while maintaining ease of updates.
Solution Approach 2:
The system changes the fundamental parameter of how automation is achieved - from fixed graph-based routing to flexible language model-based decision making. This allows the same automation framework to handle different subjects by simply changing the prompt content rather than restructuring the system.
2Manufacturing precision
If an interactive voice response system requires users to go through each step of a long process, then the system can provide structured information gathering, but the process becomes slower compared to obtaining assistance from a human
Solution Approach 1:
The system performs preliminary actions by having the large language model analyze the entire conversation context and predict the most relevant information needs in advance. This allows the system to jump to key questions rather than following a rigid step-by-step sequence, reducing time while maintaining structured information gathering.
Solution Approach 2:
The system uses partial action by selectively engaging only the necessary portions of the information gathering process based on real-time conversation analysis. The large language model determines which information is truly needed rather than systematically collecting all possible data points, reducing overall session time while maintaining quality.
3Loss of energy
If a company uses traditional automated communications techniques, then operational costs can be reduced, but the system becomes tedious for end users
Solution Approach 1:
The patent introduces a large language model as an intermediary between the automated system and the user. This intermediary provides human-like understanding and response generation, making the automated system less tedious for users while maintaining the cost benefits of automation. The LLM acts as a bridge that preserves operational efficiency while improving user experience.
Data Source
AI summary
A communications session with a user may be automated using a language model. The language model may be instructed to select a next action to be performed where the next action may include transmitting a responsive communication to the user or performing an API call. The prompt used to query the language model may include one or more of the following: a representation of text of the communications session, a list of available API calls, instructions to select a next action, a representation of API calls performed, or a representation of API call responses received. The language model may be sequentially queried to continue the communications session by transmitting responsive communications or performing API calls. In some implementations, a prompt template may be used to generate the prompt and a prompt template may be selected using text of the communications session.