WebFactory: Automated Compression of Foundational Language Intelligence into Grounded Web Agents

TL;DR

WebFactory introduces an automated reinforcement learning pipeline that compresses large language models' internet knowledge into efficient GUI agents, achieving high performance with minimal data from only 10 websites.

Contribution

The paper presents a novel, fully automated pipeline for training grounded GUI agents that efficiently compress LLM knowledge into actionable behaviors, reducing data requirements.

Findings

Agent trained on synthetic data from 10 websites matches performance of models trained on larger datasets.

Our agent outperforms the base foundation model on internal offline and online transfer benchmarks.

WebFactory demonstrates high data efficiency and strong generalization capabilities.

Abstract

Current paradigms for training GUI agents are fundamentally limited by a reliance on either unsafe, non-reproducible live web interactions or costly, scarce human-crafted data and environments. We argue this focus on data volume overlooks a more critical factor: the efficiency of compressing a large language model's (LLM) latent knowledge into actionable agent behavior. We introduce WebFactory, a novel, fully automated closed-loop reinforcement learning pipeline for GUI agents, systematically compressing LLM-encoded internet intelligence into efficient, grounded actions. Our pipeline features a process of scalable environment synthesis, knowledge-aware task generation, LLM-powered trajectory collection, decomposed reward RL training, and systematic agent evaluation. Remarkably, our agent demonstrates exceptional data efficiency and generalization. Trained on synthetic data from only 10 websites within WebFactory, it achieves performance comparable to GUI agents trained on the same amount of human-annotated data from a much larger set of environments. This superior performance is consistent across our internal offline and online transfer benchmarks, where our agent also significantly outperforms the base foundation model. We further provide critical insights into the "embodiment potential" of different LLM foundations, offering a new axis for model evaluation. This work presents a scalable and cost-effective paradigm for transforming passive internet knowledge into active, grounded intelligence, marking a critical step towards general-purpose interactive agents.