LLM Prompting for API Call Selection in Automated Support
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated communications systems, such as interactive voice response systems, are tedious for users and complex to maintain, making them inefficient for updating and supporting changes in communications sessions.
Innovation Solution
Utilizing large language models (LLMs) to automate customer support by creating prompts that include session text, available API calls, and instructions to select actions, allowing the LLM to determine appropriate API calls and responsive communications, thereby eliminating the need for graph-based navigation and simplifying maintenance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If graph-based interactive voice response systems are used to automate communications sessions, then automation capability is achieved, but device complexity and difficulty of maintenance increase
Solution Approach 1:
The patent replaces the mechanical graph-based navigation system with a language model that processes natural language directly. Instead of following predefined graph paths and nodes, the language model understands user intent and generates appropriate responses or API calls, eliminating the complexity of graph maintenance while preserving automation capability.
Solution Approach 2:
The system changes the fundamental parameter of how automation is achieved: from structured graph navigation to unstructured language processing. By transforming the input from graph node transitions to natural language prompts, the system maintains automation while simplifying the underlying structure to just the language model and available tool definitions.
2Extent of automation
If graph-based interactive voice response systems are used, then automation capability is achieved, but ease of operation deteriorates due to tedious step-by-step processes
Solution Approach 1:
The patent replaces the mechanical step-by-step graph navigation with direct language-based interaction. Users can express their needs in natural language, and the language model interprets these to directly invoke appropriate tools or provide responses, eliminating the tedious sequential steps required by graph-based systems.
3Loss of energy
If traditional automated communications techniques are used, then cost reduction is achieved, but adaptability to changes deteriorates
Solution Approach 1:
The system introduces dynamics by allowing the language model to adapt to new topics and requirements through updated prompts and tool definitions, rather than requiring structural changes to a fixed graph. The available tools and their descriptions can be dynamically added or modified without reconfiguring the entire automation system.
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.