Virtual Agent Evaluation With Synthetic Customer Conversations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network sites face challenges in returning up-to-date results for complex queries due to computational delays, leading to inaccurate results and high resource consumption, while current virtual agents are not sophisticated enough to provide meaningful resolutions, requiring users to wait for live human agents, wasting time and resources.
Innovation Solution
A system that uses a first machine learning model to simulate customer interactions with virtual agents, generating and scoring simulated conversations to improve the virtual agent's response quality and diversity, reducing the need for extensive training data and resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex queries are processed to return accurate results, then result accuracy is improved, but computational resource consumption increases and processing time increases
Solution Approach 1:
The patent creates a virtual agent that simulates customer interactions and generates synthetic training data copies instead of using real customer conversations. This copying approach allows the system to train virtual agents on diverse customer scenarios without consuming computational resources for actual customer service operations, thereby reducing overall computational resource consumption while maintaining training accuracy.
Solution Approach 2:
The system performs preliminary training of virtual agents using simulated conversations before they are deployed to handle actual customer queries. By pre-training on synthetic data generated from simulated customer interactions, the virtual agents are prepared to handle complex queries efficiently during production, reducing the computational burden on real-time processing.
2Reliability
If virtual agents are trained with extensive real customer interaction data, then response quality is improved, but training time and resource consumption increase
Solution Approach 1:
The patent generates synthetic training data by having machine learning models simulate customer interactions with virtual agents. This creates copies of real customer conversations that can be used for training without requiring actual customer time or resources. The synthetic data captures diverse customer scenarios, personalities, and query patterns, enabling comprehensive training without the time constraints of real customer interactions.
Solution Approach 2:
The system uses machine learning models to automatically generate and score simulated conversations for training virtual agents without human intervention. The machine learning models self-evaluate the quality of simulated conversations and select appropriate training data, eliminating the need for human annotators to review and label customer interactions, thereby significantly reducing training time and resources.
3Reliability
If live human agents are used to resolve customer issues, then issue resolution effectiveness is improved, but time consumption and resource consumption increase
Solution Approach 1:
The patent implements virtual agents that autonomously handle customer queries and issues without requiring human intervention. The virtual agents use machine learning models to understand customer problems, generate appropriate responses, and resolve issues independently. This self-service capability eliminates customer waiting times while maintaining effective issue resolution through continuously improving AI-powered responses.
Solution Approach 2:
The system incorporates feedback mechanisms where simulated conversations are scored by machine learning models to continuously improve virtual agent performance. The feedback loop allows virtual agents to learn from simulated customer interactions and refine their response quality over time, ensuring they provide effective resolutions comparable to human agents but without the time delay.
Data Source
AI summary
A system is described for training a virtual agent of a listing network platform using a machine learning model. The system establishes a communication session with a virtual agent of a listing network platform and generates, by a first machine learning model, conversation data representing a customer support issue associated with the listing network platform. The system transmits, by the first machine learning model, at least a portion of the conversation data to the virtual agent via the communication session. The system receives one or more responses to the at least the portion of the conversation data from the virtual agent in the communication session and stores a simulated conversation comprising the at least the portion of the conversation data generated by the first machine learning model and the one or more responses received from the virtual agent.