Conversation Orchestration Using Trained Large Language Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current conversational AI systems require extensive development and deployment efforts, involving large teams and lengthy timelines, as they struggle to efficiently handle complex use cases that demand multiple language models and systems working collaboratively, lacking an effective mechanism for orchestration.
Innovation Solution
A conversation management framework within a virtual assistant server that receives conversational inputs, generates data records, and communicates with software modules to provide responses, utilizing a hub and spoke architecture with orchestrator large language models to coordinate and execute sub-tasks, enabling efficient orchestration of customer conversations across multiple models and systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple language models and systems are used to handle complex use cases, then the capability to fulfill diverse customer needs is improved, but the system complexity and orchestration difficulty increase
Solution Approach 1:
The patent introduces a dialog flow engine as an intermediary component that coordinates between multiple language models and systems. This engine manages the orchestration of specialized models (NLU, response generation, information retrieval) and external systems, reducing the complexity burden on individual components while maintaining the ability to handle diverse customer needs through coordinated collaboration.
Solution Approach 2:
The system segments the virtual assistant functionality into specialized language models and software modules, each handling specific sub-tasks (understanding customer intent, generating responses, retrieving information). This segmentation allows each component to be optimized for its specific function while the dialog flow engine integrates them, resolving the contradiction between versatility and complexity.
2Reliability
If rigorous development and testing are performed to ensure satisfactory customer interaction, then the quality and reliability of the virtual assistant is improved, but the development time and resource requirements increase
Solution Approach 1:
The system employs automated testing frameworks and evaluation mechanisms that enable self-assessment of the virtual assistant's performance. This self-service approach to quality assurance reduces the need for extensive manual testing by human teams, maintaining high reliability standards while significantly reducing development time and resource requirements.
Solution Approach 2:
The patent implements feedback loops where the virtual assistant's performance is continuously evaluated against predefined criteria and customer interactions are analyzed. This automated feedback mechanism enables iterative improvement and quality assurance without requiring extensive manual intervention, resolving the contradiction between reliability and development time.
3Manufacturing precision
If custom virtual assistants are developed with specialized expertise for specific use cases, then the precision and effectiveness of customer service is improved, but the development cost and timeline increase
Solution Approach 1:
The dialog flow engine serves as a universal platform that can orchestrate multiple specialized language models across different use cases. This multi-functional architecture allows the same core infrastructure to support various customer service scenarios (technical support, sales, customer care) with specialized models, reducing development costs and timelines by avoiding redundant infrastructure for each use case while maintaining precision through specialization.
Data Source
AI summary
A virtual assistant server receives conversational inputs as part of a conversation from a customer device and generates a version of a data record of the conversation upon: the receiving of each of the conversational inputs, or receiving each output generated by one of a plurality of software modules when the one of the software modules receives a system input from the conversation management framework. The virtual assistant server provides each of the generated versions of the data record to a communication orchestrator and receives for each of the generated versions of the data record, execution instructions from the communication orchestrator. Further, the virtual assistant server communicates with one or more of the software modules based on the received execution instructions, and provides based on the communicating, one or more of the outputs of the software modules to the customer device, when the one or more of the outputs comprise one or more responses to one or more of the conversational inputs.