Conversational Bot State Machine for Adaptive Flow Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conversational marketing bots struggle to effectively engage users in sales and marketing conversations, as they lack the proactive and reactive capabilities to adapt conversation flows based on user inputs, leading to inefficient lead qualification and customer experience.
Innovation Solution
A directed acyclic graph (DAG) is used to pre-configure a desired conversation flow, with nodes containing synthetic conversation transcripts and event expressions, allowing the bot to determine whether user inputs extend or interrupt current events, enabling flexible and adaptive conversation management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a rules-based chatbot with decision trees is used to control conversation flow, then the bot can proactively lead the conversation by asking questions, but the system becomes brittle and does not scale well
Solution Approach 1:
The patent transitions from static decision trees to a dynamic state machine model where the conversation state is represented by a current state and transitions between states based on user input. This dynamic approach allows the bot to adapt conversation flow proactively while maintaining scalability, as the state machine can handle complex conversations without requiring manual decision tree construction for each scenario.
Solution Approach 2:
The state machine framework serves multiple functions: it manages conversation flow, tracks conversation state, determines when to transfer to human agents, and handles various conversation scenarios uniformly. This universal approach replaces multiple specialized decision trees with a single flexible mechanism that scales across different conversation types and complexities.
2Reliability
If an intent-based chatbot using machine learning is used, then the bot becomes more robust, but it is not flexible to context switches once an intent is activated
Solution Approach 1:
The patent implements dynamic state transitions within the conversation management system. Instead of being locked into a single intent, the system can transition between different conversation states based on user input, allowing flexible context switching while maintaining the robustness of intent-based classification. The state machine evaluates transitions based on current state and user input, enabling adaptive conversation flow.
Solution Approach 2:
The patent segments the conversation into distinct states that can be independently managed and transitioned between. Each state represents a specific conversation context or phase, and the system can move between states based on user input. This segmentation allows the bot to maintain robust intent recognition while enabling flexible context switches by transitioning to appropriate states.
3Productivity
If customer support bot approach is used with prescribed remediation steps, then the conversation can be ended quickly, but it does not work well for sales and marketing conversations where engagement is the goal
Solution Approach 1:
The patent implements dynamic conversation management where the system can adapt its approach based on the conversation type and user behavior. For sales and marketing conversations, the state machine can transition to states that prioritize engagement and information gathering rather than quick resolution. The system dynamically adjusts conversation pacing and direction based on user responses and conversation state.
Solution Approach 2:
The patent changes key conversation parameters such as the goal state and transition criteria based on conversation type. For sales conversations, the system prioritizes states that maximize engagement and lead qualification, whereas for support conversations, it prioritizes resolution states. This parameter adaptation allows the same underlying system to serve multiple conversation types effectively.
4Productivity
If the bot follows a fixed conversation flow, then it can efficiently qualify leads, but it cannot gracefully handle user interruptions or redirect the conversation when needed
Solution Approach 1:
The patent implements a dynamic state machine that continuously evaluates user input against possible state transitions. When a user interrupts or redirects the conversation, the system can transition to appropriate states that handle the new direction while maintaining lead qualification objectives. The dynamic nature allows the bot to efficiently follow prescribed flows when appropriate while gracefully handling interruptions through state transitions.
Data Source
AI summary
Method and system to control a conversational bot uses a directed acyclic graph to specify a desired conversation flow. A graph node has synthetic conversation transcripts annotated with events, wherein an event in a synthetic conversation transcript has preconfigured event expressions that represent ways in which dialogue at the node can unfold. During an on-going conversation with an actor, the system provides a data model uniquely associated with the conversation and that specifies a linear sequence of observations. The data model includes events representing semantically-related conversation fragments located in annotated historical conversation transcripts. In response to receipt of an input in association with a current graph node, the system determines whether the input extends an event in the synthetic conversation transcript associated with the node. If so, a response that continues a current conversation flow in the graph is provided. If not, a response that interrupts a current conversation flow in the graph is provided.