DuMate-DeepResearch: An Auditable Multi-Agent System with Recursive Search and Rubric-Grounded Reasoning

2026-06-05

AI Summary

This work addresses key challenges in deep research tasks—such as ambiguous long-term planning, bottlenecks in single-agent task decomposition, hallucination in long-form text generation, and lack of auditability—by proposing a multi-agent deep research framework. The framework decouples task comprehension and scheduling from an extensible tool ecosystem and integrates graph-driven dynamic planning, a recursive two-level sub-agent execution mechanism, and scoring-criterion-guided reasoning optimization at test time, enabling fully traceable and collaborative reasoning throughout the research process. Implemented on the Qianfan Agent Foundry platform, the system achieves state-of-the-art composite scores of 58.03% and 61.95% on DeepResearch Bench and DeepResearch Bench II, respectively, ranking first in information recall and analytical capability, and substantially enhancing the controllability, stability, and auditability of the research workflow.

Abstract

Deep Research (DR) has emerged as a new agentic paradigm to tackle complex, open-ended research tasks, demanding systems that can iteratively frame problems, acquire evidence, verify sources, and synthesize long-form reports. In practice, however, current DR systems are constrained by four interrelated limitations: long-horizon planning over an underspecified scope, the bottleneck of decomposing and scheduling such tasks within a single agent, hallucination risk in long-form synthesis, and limited process auditability. This technical report presents DuMate-DeepResearch, a multi-agent DR framework built on the Qianfan Agent Foundry. The framework decouples the Agent Core, which handles task understanding, planning, and scheduling, from an extensible Tool Ecosystem for retrieval, evidence acquisition, and report rendering, making every intermediate decision and tool invocation explicitly traceable. Building on this infrastructure, DuMate-DeepResearch further introduces three mechanisms: (i) a graph-based dynamic planning strategy expands the research roadmap coarse-to-fine and continuously revises it through reflection, re-planning, backtracking, and parallel branching; (ii) a recursive two-level execution design delegates each complex search sub-task to an inner Search Agent that runs its own planning loop, isolating noisy retrieval and stabilizing long-horizon execution; (iii) a rubric-based test-time optimization mechanism dynamically generates task-specific quality criteria and uses them as live reasoning scaffolds for evidence-grounded synthesis and adaptive stopping. Across two deep research benchmarks, DuMate-DeepResearch establishes new state-of-the-art results: the best overall score (58.03%) on DeepResearch Bench, and the best overall score (61.95%) on DeepResearch Bench II while ranking first in information recall and analysis.

Problem

• Deep Research
• multi-agent system
• long-horizon planning
• hallucination
• auditability

Innovation

• multi-agent system
• recursive search
• rubric-grounded reasoning
• dynamic planning
• auditable AI