DuCCAE: A Hybrid Engine for Immersive Conversation via Collaboration, Augmentation, and Evolution
arXiv:2603.19248v1 Announce Type: cross Abstract: Immersive conversational systems in production face a persistent trade-off between responsiveness and long-horizon task capability. Real-time interaction is achievable for lightweight turns, but requests involving planning and tool invocation (e.g., search and media generation) produce heavy-tail execution latency that degrades turn-taking, persona consistency, and user trust. To address this challenge, we propose DuCCAE (Conversation while Collaboration with Augmentation and Evolution), a hybrid engine for immersive conversation deployed within Baidu Search, serving millions of users. DuCCAE decouples real-time response generation from asynchronous agentic execution and synchronizes them via a shared state that maintains session context and execution traces, enabling asynchronous results to be integrated back into the ongoing dialogue. The system orchestrates five subsystems-Info, Conversation, Collaboration, Augmentation, and Evoluti
arXiv:2603.19248v1 Announce Type: cross Abstract: Immersive conversational systems in production face a persistent trade-off between responsiveness and long-horizon task capability. Real-time interaction is achievable for lightweight turns, but requests involving planning and tool invocation (e.g., search and media generation) produce heavy-tail execution latency that degrades turn-taking, persona consistency, and user trust. To address this challenge, we propose DuCCAE (Conversation while Collaboration with Augmentation and Evolution), a hybrid engine for immersive conversation deployed within Baidu Search, serving millions of users. DuCCAE decouples real-time response generation from asynchronous agentic execution and synchronizes them via a shared state that maintains session context and execution traces, enabling asynchronous results to be integrated back into the ongoing dialogue. The system orchestrates five subsystems-Info, Conversation, Collaboration, Augmentation, and Evolution-to support multi-agent collaboration and continuous improvement. We evaluate DuCCAE through a comprehensive framework that combines offline benchmarking on the Du-Interact dataset and large-scale production evaluation within Baidu Search. Experimental results demonstrate that DuCCAE outperforms strong baselines in agentic execution reliability and dialogue quality while reducing latency to fit strict real-time budgets. Crucially, deployment metrics since June 2025 confirm substantial real-world effectiveness, evidenced by a tripling of Day-7 user retention to 34.2% and a surge in the complex task completion rate to 65.2%. Our hybrid architecture successfully preserves conversational continuity while enabling reliable agentic execution, offering practical guidelines for deploying scalable agentic systems in industrial settings.
Executive Summary
The article proposes DuCCAE, a hybrid engine for immersive conversation that addresses the trade-off between responsiveness and long-horizon task capability in conversational systems. DuCCAE decouples real-time response generation from asynchronous agentic execution, enabling the integration of asynchronous results into ongoing dialogue. The system consists of five subsystems that support multi-agent collaboration and continuous improvement. The authors evaluate DuCCAE through benchmarking and large-scale production evaluation, demonstrating its effectiveness in reducing latency while preserving conversational continuity. The deployment metrics since June 2025 confirm substantial real-world effectiveness, including a tripling of user retention and a surge in complex task completion rate. This innovative architecture offers practical guidelines for deploying scalable agentic systems in industrial settings.
Key Points
- ▸ DuCCAE addresses the trade-off between responsiveness and long-horizon task capability
- ▸ Decouples real-time response generation from asynchronous agentic execution
- ▸ Consists of five subsystems that support multi-agent collaboration and continuous improvement
- ▸ Demonstrates effectiveness in reducing latency while preserving conversational continuity
Merits
Strength in Scalability
DuCCAE's hybrid architecture enables the integration of asynchronous results into ongoing dialogue, making it suitable for large-scale production environments.
Practical Guidelines
The authors provide practical guidelines for deploying scalable agentic systems in industrial settings, making the research more accessible and applicable to practitioners.
Real-World Effectiveness
The deployment metrics since June 2025 confirm substantial real-world effectiveness, including a tripling of user retention and a surge in complex task completion rate.
Demerits
Complexity in Implementation
The hybrid architecture of DuCCAE may be challenging to implement, requiring significant expertise and resources.
Limited Generalizability
The research is primarily focused on conversational systems deployed within Baidu Search, and its generalizability to other domains and platforms is unclear.
Expert Commentary
The article makes a significant contribution to the field of conversational AI by proposing a hybrid architecture that addresses the trade-off between responsiveness and long-horizon task capability. The evaluation of DuCCAE through benchmarking and large-scale production evaluation demonstrates its effectiveness in reducing latency while preserving conversational continuity. However, the complexity of implementation and limited generalizability of the research are notable limitations. The implications of the research are significant, particularly in the context of agentic systems and conversational AI. The development of hybrid architectures like DuCCAE can improve the responsiveness and effectiveness of conversational systems, leading to better user experience and outcomes. Furthermore, the research highlights the importance of designing conversational systems that can adapt to changing user needs and contexts, which can inform policy decisions regarding the development and deployment of conversational AI in various sectors.
Recommendations
- ✓ Recommendation 1: Researchers should investigate the applicability of DuCCAE to other domains and platforms, including those outside of the Baidu Search environment.
- ✓ Recommendation 2: Developers should consider the complexity of implementation and the resources required to deploy hybrid architectures like DuCCAE, and develop strategies to mitigate these challenges.
Sources
Original: arXiv - cs.AI
