ResearStudio: A Human-Intervenable Framework for Building Controllable Deep-Research Agents

2025-10-14

AI Summary

Existing deep research agents operate under a “fire-and-forget” paradigm, lacking runtime human intervention capabilities. Method: We propose the first deep research agent framework supporting fine-grained, real-time human intervention—built upon a hierarchical Planner-Executor architecture, a runtime “plan-as-document” mechanism, and a high-speed inter-process communication layer. This enables on-the-fly pausing, plan/code editing, custom command injection, and seamless resumption, while dynamically switching between AI- and human-led modes. A novel collaborative workshop design integrates automation efficiency with domain-expert knowledge infusion. Contribution/Results: Our framework achieves state-of-the-art performance on the GAIA benchmark, significantly outperforming OpenAI DeepResearch and Manus. All code, protocols, and evaluation scripts are fully open-sourced.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Multiagent Planning

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Current deep-research agents run in a ''fire-and-forget'' mode: once started, they give users no way to fix errors or add expert knowledge during execution. We present ResearStudio, the first open-source framework that places real-time human control at its core. The system follows a Collaborative Workshop design. A hierarchical Planner-Executor writes every step to a live ''plan-as-document,'' a fast communication layer streams each action, file change, and tool call to a web interface. At any moment, the user can pause the run, edit the plan or code, run custom commands, and resume -- switching smoothly between AI-led, human-assisted and human-led, AI-assisted modes. In fully autonomous mode, ResearStudio achieves state-of-the-art results on the GAIA benchmark, surpassing systems like OpenAI's DeepResearch and Manus. These results show that strong automated performance and fine-grained human control can coexist. The full code, protocol, and evaluation scripts are available at https://github.com/ResearAI/ResearStudio. We will continue to update the repository to encourage further work on safe and controllable research agents. Our live demo is publicly accessible at http://ai-researcher.net:3000/. We support the development of DeepScientist, which can be accessed at https://github.com/ResearAI/DeepScientist.

Problem

• Enabling real-time human intervention in deep-research agent execution
• Providing hierarchical planning with live editable plan-as-document
• Achieving autonomous performance while maintaining fine-grained human control

Innovation

• Collaborative Workshop design enables real-time human control
• Hierarchical Planner-Executor writes steps to live document
• Users can pause, edit plans or code, then resume