Symbolic AI architecture improves goal-driven chat accuracy and conversation flow
An Auditable Symbolic-RAG-Generative AI Architecture for Goal-Oriented Conversation Orchestration
Artificial Intelligence
Summary
Chatbots that help with specific goals, like buying a house, must understand what users say and keep conversations helpful without getting stuck on set questions. The authors developed a system combining symbolic AI and a method that pulls in helpful information to guide such chats better. Their system, called GRACE, keeps track of goals, uses user input carefully, and picks the best next question to move the conversation forward. Tests showed it worked well in both English and Spanish in real situations like real estate and cleaning services.
What this means in practice
- •For customer service developers: Build chatbots that handle complex, goal-driven conversations reliably by using a symbolic and retrieval-based approach to guide and audit dialogs.
- •For business process designers: Design customer interaction flows that ensure business goals are met while respecting user experience constraints using an auditable conversation state model.
Authors
Ramon Gonzalez, Antonio Diaz
Abstract
Goal-oriented conversational systems must answer factual questions, understand visitor-provided information, and advance business objectives without becoming rigid questionnaires. This paper proposes a Symbolic-RAG-Generative architecture centered on the Goal-oriented Retrieval-Augmented Conversation Engine (GRACE). An instruction-constrained Business Goal Compiler transforms business intent into an immutable objective set, normalized priority vector, canonical questions, and initial state vector. At runtime, GRACE receives the complete conversation history, latest visitor message, current state, and grounded answer generated by a separate RAG component. It updates completion only from visitor-authored evidence and selects one contextually modulated follow-up. The core policy maximizes expected business progress subject to a minimum visitor-utility constraint. We formalize the state, monotonic transitions, source separation, question modulation, and constrained policy; present the reference architecture; and define an evaluation comprising 24 English real-estate and 10 Spanish professional-cleaning conversations, totaling 119 protocol-defined visitor turns. Across both domains, GRACE achieves 84.9% exact state-transition accuracy, 91.6% evidence precision, 89.6% evidence recall, 100% monotonicity, and 94.1% terminal-state accuracy. The evaluation establishes compelling symbolic-state performance across standard, multi-goal, RAG-detour, validation, refusal, and robustness scenarios.