MindWatcher: Toward Smarter Multimodal Tool-Integrated Reasoning

TL;DR

MindWatcher is a novel multimodal reasoning agent that autonomously invokes diverse tools and manipulates images during reasoning, enabling complex decision-making without human prompts, and is evaluated on a new benchmark.

Contribution

Introduction of MindWatcher, a tool-integrated reasoning agent with interleaved thinking and multimodal chain-of-thought capabilities, along with a new benchmark for evaluation.

Findings

MindWatcher matches or exceeds larger models in performance.

Effective multimodal reasoning with image manipulation.

Insights into agent training dynamics, such as genetic inheritance phenomena.

Abstract

Traditional workflow-based agents exhibit limited intelligence when addressing real-world problems requiring tool invocation. Tool-integrated reasoning (TIR) agents capable of autonomous reasoning and tool invocation are rapidly emerging as a powerful approach for complex decision-making tasks involving multi-step interactions with external environments. In this work, we introduce MindWatcher, a TIR agent integrating interleaved thinking and multimodal chain-of-thought (CoT) reasoning. MindWatcher can autonomously decide whether and how to invoke diverse tools and coordinate their use, without relying on human prompts or workflows. The interleaved thinking paradigm enables the model to switch between thinking and tool calling at any intermediate stage, while its multimodal CoT capability allows manipulation of images during reasoning to yield more precise search results. We implement automated data auditing and evaluation pipelines, complemented by manually curated high-quality datasets for training, and we construct a benchmark, called MindWatcher-Evaluate Bench (MWE-Bench), to evaluate its performance. MindWatcher is equipped with a comprehensive suite of auxiliary reasoning tools, enabling it to address broad-domain multimodal problems. A large-scale, high-quality local image retrieval database, covering eight categories including cars, animals, and plants, endows model with robust object recognition despite its small size. Finally, we design a more efficient training infrastructure for MindWatcher, enhancing training speed and hardware utilization. Experiments not only demonstrate that MindWatcher matches or exceeds the performance of larger or more recent models through superior tool invocation, but also uncover critical insights for agent training, such as the genetic inheritance phenomenon in agentic RL.