AI Workflow Selection for Electronic Conversation Agents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The large number of workflows and tools available to customer service agents in electronic conversation systems results in a steep learning curve and request latency, leading to a poor user experience due to the time required to find appropriate workflows.
Innovation Solution
An AI system that monitors conversations in real-time, identifies user intents, and automatically recommends or selects suitable workflows, forms, and widgets to simplify the agent's task, reducing the need for manual selection and improving workflow selection over time based on usage data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the number of workflows and tools available to agents is increased to handle diverse customer requests, then the system's capability to fulfill user requests is improved, but the time required for agents to find appropriate workflows increases
Solution Approach 1:
The patent replaces the manual mechanical search process with an AI-based automated recommendation system. The system uses natural language processing to analyze conversation context and automatically suggests relevant workflows, forms, and widgets to agents, eliminating the need for manual searching through large catalogs of available tools.
Solution Approach 2:
The patent introduces an AI system as an intermediary between the conversation context and the workflow selection. This intermediary analyzes the conversation in real-time and bridges the gap by presenting only the most relevant workflows to agents, rather than requiring direct access to the entire workflow library.
2Adaptability or versatility
If the number of workflows and tools available to agents is increased to handle diverse customer requests, then the system's capability to fulfill user requests is improved, but the learning curve for agents increases
Solution Approach 1:
The patent replaces the manual process of learning and remembering numerous workflows with an AI-based recommendation system. The system automatically analyzes conversation context and suggests appropriate workflows, allowing agents to focus on customer interaction rather than memorizing tool catalogs.
Solution Approach 2:
The system provides self-service by automatically identifying and presenting relevant workflows based on conversation context. Agents don't need to manually search or remember workflow names; the system serves them the appropriate tools based on real-time analysis of the customer's needs.
3Ease of operation
If agents manually search for and select appropriate workflows from a large catalog, then workflow selection flexibility is maintained, but request latency increases leading to poor customer experience
Solution Approach 1:
The patent performs preliminary action by pre-analyzing conversation context and proactively suggesting relevant workflows before agents need to search for them. The AI system continuously monitors the conversation and prepares workflow recommendations in advance, so agents receive suggestions at the optimal moment rather than having to search manually.
Solution Approach 2:
The patent replaces the manual search and selection mechanism with an automated AI-based recommendation system that maintains flexibility while improving speed. The system analyzes conversation context in real-time and presents relevant workflow options, preserving agent choice while eliminating manual searching.
Data Source
AI summary
Within an environment in which users converse at least partly with human agents to accomplish a desired task, a server assists the agents by identifying workflows that are most applicable to the current conversation. Workflow selection functionality identifies one or more candidate workflows based on techniques such as user intent inference, conversation state tracking, or search, according to various embodiments. The identified candidate workflows are either automatically selected on behalf of the agent, or are presented to the agent for manual selection.