• Home
  • Search
  • WebPilot: A Versatile and Autonomous Multi-Agent System for Web Task Execution with Strategic Exploration
  • Cite Icon9
  • https://doi.org/10.1609/aaai.v39i22.34505Copy DOI Icon

WebPilot: A Versatile and Autonomous Multi-Agent System for Web Task Execution with Strategic Exploration

  • Abstract
  • Literature Map
  • Citations
  • Similar Papers
Abstract

LLM-based autonomous agents often fail to execute complex web tasks that require dynamic interaction, largely due to the inherent uncertainty and complexity of these environments. Existing LLM-based web agents typically rely on rigid, expert-designed policies specific to certain states and actions, lacking the flexibility and generalizability needed to adapt to unseen tasks. In contrast, humans excel by exploring unknowns, continuously adapting strategies based on new observations, and resolving ambiguities through exploration. To emulate human-like adaptability, web agents need strategic exploration and complex decision-making. Monte Carlo Tree Search (MCTS) is well-suited for this, but classical MCTS struggles with vast action spaces, unpredictable state transitions, and incomplete information in web tasks. In light of this, we develop WebPilot, a multi-agent system with a dual optimization strategy that improves MCTS to better handle complex web environments. Specifically, the Global Optimization phase involves generating a high-level plan by breaking down tasks into manageable subtasks, continuously refining this plan through reflective analysis of new observations and previous subtask attempts, thereby focusing the search process and mitigating challenges posed by vast action spaces in classical MCTS. Subsequently, the Local Optimization phase executes each subtask using a tailored MCTS designed for complex environments, effectively addressing uncertainties and managing incomplete information by iteratively refining decisions based on new observations. Experimental results on WebArena and MiniWoB++ demonstrate the effectiveness of WebPilot. Notably, on WebArena, WebPilot achieves SOTA performance with GPT-4, achieving a 93% relative increase in success rate over the concurrent tree search-based method. WebPilot advances autonomous agents, enabling more reliable decision-making in practical environments.

Similar Papers
  • PDF
  • Research Article
  • Citations328

Monte Carlo Tree Search: a review of recent modifications and applications

  • Jul 19, 2022
  • Artificial Intelligence Review
  • Maciej Świechowski +3
  • Conference Article
  • Citations7

Decentralized Communication-less Multi-Agent Task Assignment with Cooperative Monte-Carlo Tree Search

  • Apr 01, 2020
  • Mohammadreza Daneshvaramoli +6
  • Research Article
  • Citations1

Using Response Functions for Strategy Training and Evaluation

  • Jan 01, 2015
  • University of Alberta Library
  • Trevor Davis
  • Research Article
  • Citations4

A Dynamic Programming Algorithm for Finding an Optimal Sequence of Informative Measurements.

  • Jan 30, 2023
  • Entropy
  • Peter N Loxley +1
  • Research Article
  • Citations5

Generative design for complex floorplans in high-rise residential buildings: A Monte Carlo tree search-based self-organizing multi-agent system (MCTS-MAS) solution

  • Aug 22, 2024
  • Expert Systems With Applications
  • Peiyang Su +5
  • Research Article

A decision-making framework using MCTS as a hierarchical task network and deep learning connector

  • Oct 01, 2025
  • Science Progress
  • Tianhao Shao +3
  • Conference Article
  • Citations25

A monte carlo tree search framework for quantum circuit transformation

  • Nov 02, 2020
  • arXiv (Cornell University)
  • Xiangzhen Zhou +2
  • Book Chapter

Cost-Awareness in Multi-Agent Active Search

  • Sep 28, 2023
  • Frontiers in artificial intelligence and applications
  • Arundhati Banerjee +2
  • Conference Article
  • Citations4

Intention-Aware Multiagent Scheduling

  • May 05, 2020
  • Michael Dann +3
  • Book Chapter
  • Citations3

The Multiple Uses of Monte-Carlo Tree Search

  • Apr 20, 2022
  • Richard Senington
  • Research Article
  • Citations8

Measurement Maximizing Adaptive Sampling with Risk Bounding Functions

  • Jul 17, 2019
  • Proceedings of the AAAI Conference on Artificial Intelligence
  • Benjamin Ayton +2
  • Research Article

ToC: Tree-of-Claims Search with Multi-Agent Language Models

  • Mar 14, 2026
  • Shuyang Yu +2
  • Supplementary Content

Web Agent Agentic Reinforcement Learning Decision Model Under Multi-Cost and Failure Risk Constraints

  • Feb 02, 2026
  • Preprints.org
  • Qianli Ma +4
  • Conference Article

Design and Implementation of Multi-Agent Systems: A Lang Graph-Based Approach

  • Jul 04, 2025
  • Moupiya Das +6
  • Research Article
  • Citations50

A deep learning-based multi-agent system for intrusion detection

  • Mar 17, 2020
  • SN Applied Sciences
  • Faten Louati +1
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.