Newest papers

Evolving Agent Interfaces

Saaket Agashe, Yuzhe Yang, Gonzalo Gonzalez-Pumariega, Bhuvan Chandra Koduru, Chih-Lun Li, Jiachen Yang, Ang Li, Xin Eric Wang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28601 · cited by 0 works (OpenAlex) · open-access PDF

French

Les agents d'utilisation de l'ordinateur sont limités par les interfaces que les humains leur donnent. Un agent peut taper une cellule de feuille de calcul par cellule lorsque une seule ligne de code le ferait, ou imprimer toutes les deux mille lignes d'un fichier à travers une coque pour trouver une fonction. Les deux interfaces et les connaissances de l'utilisation sont construites à la main, qui ne peuvent pas suivre le rythme avec les nouvelles applications, les flux de travail et les modèles. Nous affirmons que la conception de l'interface devrait être elle-même une capacité d'agent. Nous introduisons Agent S4, un scientifique d'utilisation de l'ordinateur autonome qui construit des interfaces d'agent évoluant à partir de l'expérience de tâche. Un méta-agent examine

Machine translation (Meta M2M100); the English original follows.

Abstract

Computer-use agents are limited by the interfaces humans give them. An agent may type a spreadsheet cell by cell when a single line of code would do, or print all two thousand lines of a file through a shell to find one function. Both the interfaces and the knowledge of when to use each are built by hand, which cannot keep pace with new applications, workflows, and models. We argue that interface design should itself be an agent capability. We introduce Agent S4, an autonomous computer-use scientist that builds Evolving Agent Interfaces from task experience. A meta-agent examines task agents' trajectories on synthetic training tasks, without success labels, and constructs and revises interfaces and usage knowledge in a repository. Population-based evolution then selects which repositories to develop further by their performance on validation tasks the meta-agent never sees, favoring changes that generalize beyond the seen tasks. On OSWorld V2, Agent S4, with Qwen3.8-27B and the evolved repository of interfaces, sets a new state of the art in open-source computer-use agents, reaching 21.3% binary score, 3.8 times the same model with fixed interfaces. This is above both GPT-5.5 (13.0%) and Claude Opus 4.8 (20.6%), at about 4.3% and 1.5% of their respective costs. On ScienceBoard, the evolved interfaces improve the success rate from 41.4% to 58.6% with the same model. These results suggest that useful interfaces can be evolved from experience, making autonomous interface design a source of improvement complementary to advances in models and harnesses. We open-source our code at https://github.com/simular-ai/agent-s.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Task (project management), Interface (matter), Code (set theory), Human–computer interaction, Autonomous agent

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange