API Metadata Extraction for Intelligent Agent Automation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current process of registering services with intelligent agents and manually training natural language processors is laborious and time-consuming, requiring separate steps for each agent, making it difficult for developers to make services accessible via multiple intelligent agents.
Innovation Solution
An intelligent agent automatically discovers APIs exposed by services and uses descriptive metadata to train natural language processors and function call builders, enabling automatic recognition and interaction with services without manual registration or training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual registration and training processes are used for each intelligent agent, then services can be made accessible to intelligent agents, but the process becomes laborious and time-consuming
Solution Approach 1:
The patent extracts natural language processing information from existing service descriptions, user inputs, and entity mappings to create training data for intelligent agents. This copying approach eliminates manual training by replicating relevant information patterns across multiple agents, significantly reducing the time and labor required to deploy services to each new intelligent agent platform.
Solution Approach 2:
The system enables services to automatically generate and provide their own training data to intelligent agents through automated extraction of service descriptions, user inputs, and entity mappings. This self-service mechanism eliminates the need for manual registration and training processes, allowing services to be automatically made accessible to multiple intelligent agents without human intervention.
2Adaptability or versatility
If separate manual training is performed for each intelligent agent, then each agent can recognize service invocations, but the complexity of service deployment increases
Solution Approach 1:
The patent creates a universal natural language processing information structure that can be used across multiple intelligent agent platforms. By extracting standardized service descriptions, user inputs, and entity mappings that can be replicated across different agents, the system enables a single service to be adapted to multiple intelligent agents without requiring separate customization for each platform, thereby reducing deployment complexity while maintaining broad compatibility.
3Speed
If automated API discovery is implemented, then service discovery speed increases, but the system must process and interpret metadata automatically
Solution Approach 1:
The patent replaces manual mechanical processes of service registration and training with automated computational processes. The intelligent agent automatically discovers services by processing API metadata, service descriptions, and user inputs through natural language processing algorithms. This substitution of manual operations with automated information extraction and pattern recognition systems enables rapid service discovery while the system manages the processing complexity through standardized data structures and automated extraction mechanisms.
Data Source
AI summary
Systems, methods and computer program products are described herein that enable an intelligent agent to automatically discover an application programming interface (API) exposed by a service, to obtain descriptive metadata associated with the API, and to generate, based on the metadata, natural language processing information and/or function call building information associated with at least one functionality of the service. In addition, the intelligent agent is enabled to use the natural language processing information to determine that a user intends to interact with the service based on at least one user input of the user and to use the function call building information to build a function call for interacting with the service based on the at least one user input of the user, and to place the function call to the service on behalf of the user.