Search papers, labs, and topics across Lattice.
This paper explores scalable behavior cloning for browser agents through skill distillation, arguing that the primary challenge lies in decision-making under incomplete information rather than low-level operations. By converting user interaction trajectories into compact, natural-language skills, the authors enable agents to read, retrieve, reuse, and compose these skills effectively. The study also introduces a skill graph to facilitate the organized growth of skills, highlighting that the scalability of browser agents can leverage the collective skills demonstrated by internet users rather than relying solely on manually designed tasks.
Browser agents can achieve unprecedented scalability by harnessing the collective skills of internet users through skill distillation.
Internet users collectively perform an enormous range of skilled work through web browsers, from software development and document editing to search, forms, and enterprise workflows, making human browsing a highly scalable but under-exploited source of reusable browser skills. We argue that the bottleneck for browser agents is decision-making under incomplete information rather than low-level operation, and that the priors agents lack are already implicit in human interaction traces. We therefore study scalable behavior cloning for browser agents via skill distillation, converting user interaction trajectories into compact natural-language skills that agents can read, retrieve, reuse, and compose directly. We further organize the distilled skills into a skill graph so that growth proceeds through consolidation rather than unbounded accumulation. This suggests that the scalability of browser agents may come less from manually designed tasks and more from the collective skills already expressed by internet users. Our project is available at: https://lab.einsia.ai/browserbc/.
