Dialog Flow Modification via Agent Response Pattern Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing agent-assist systems in contact centers are not effective in handling complex conversations and require extensive training and manual configuration, providing limited feedback and insights for performance improvement.
Innovation Solution
A method and system that monitor conversations between customers and human agents, using a classification model to identify changes between automated response recommendations and agent responses, associating tags with agent responses, and providing flow modification recommendations to improve dialog flows based on identified patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If extensive training and manual configuration are provided to improve agent-assist system performance, then system effectiveness improves, but time and resource consumption increases
Solution Approach 1:
The system automatically monitors conversation data, executes dialog flows, and generates flow modification recommendations without requiring manual configuration or extensive training. The automated system serves itself by continuously learning from conversation patterns and providing actionable insights to improve dialog flows
Solution Approach 2:
The system implements a feedback loop where agent responses are compared with automated recommendations, patterns are identified from discrepancies, and flow modification recommendations are generated. This continuous feedback mechanism enables the system to self-improve without external training input
2Measurement precision
If manual review and updating of dialog flows is performed to improve recommendations, then recommendation quality improves, but productivity decreases
Solution Approach 1:
The system automatically performs the review and updating process by monitoring conversation data, identifying patterns in agent modifications, and generating flow modification recommendations. This eliminates the need for manual review while maintaining high recommendation quality through automated pattern recognition
Solution Approach 2:
The manual mechanical process of reviewing and updating dialog flows is replaced with an automated computational system that uses machine learning models to analyze conversation data and generate recommendations, significantly improving productivity while maintaining precision
3Device complexity
If limited feedback mechanisms are used (accept/reject options), then system complexity is reduced, but information quality for performance improvement deteriorates
Solution Approach 1:
The system implements a comprehensive feedback mechanism that captures detailed information about agent responses, including modifications made to automated recommendations, patterns identified across multiple conversations, and specific flow modification suggestions. This rich feedback provides high-quality performance insights while maintaining manageable system complexity
Data Source
AI summary
A server manages conversation data for a plurality of conversations between agent devices and customer devices and provides to the agent devices, one or more automated response recommendations to one or more customer messages by executing one or more dialog flows. The server, using a classification model, determines for each of the messages when there are one or more changes between the corresponding one or more automated response recommendations and one or more agent responses. Further, the server associates one or more tags to the one or more agent responses when the one or more changes are determined. Subsequently, the server identifies one or more patterns in the one or more tags associated to the one or more agent responses and provides one or more modification recommendations to the one or more dialog flows to one or more enterprise user devices based on the identified one or more patterns.