Hybrid AI for Responsive Multi-Turn Online Conversations with Novel Dynamic Routing and Feedback Adaptation

TL;DR

This paper presents a hybrid conversational AI framework that combines retrieval-augmented generation with intent-based responses, improving accuracy, latency, and adaptability for enterprise multi-turn interactions.

Contribution

It introduces a novel dynamic routing and feedback mechanism that enhances response quality and system scalability in enterprise conversational AI.

Findings

Achieves 95% accuracy in diverse queries

Maintains low latency of 180ms

Outperforms standalone RAG and intent systems

Abstract

Retrieval-Augmented Generation (RAG) systems and large language model (LLM)-powered chatbots have significantly advanced conversational AI by combining generative capabilities with external knowledge retrieval. Despite their success, enterprise-scale deployments face critical challenges, including diverse user queries, high latency, hallucinations, and difficulty integrating frequently updated domain-specific knowledge. This paper introduces a novel hybrid framework that integrates RAG with intent-based canned responses, leveraging predefined high-confidence responses for efficiency while dynamically routing complex or ambiguous queries to the RAG pipeline. Our framework employs a dialogue context manager to ensure coherence in multi-turn interactions and incorporates a feedback loop to refine intents, dynamically adjust confidence thresholds, and expand response coverage over time. Experimental results demonstrate that the proposed framework achieves a balance of high accuracy (95\%) and low latency (180ms), outperforming RAG and intent-based systems across diverse query types, positioning it as a scalable and adaptive solution for enterprise conversational AI applications.