Newest papers

LiveUIBench: Evaluating Agent as App Workflows

Tong Zhang, Yuxuan Cheng, Ruimin Zhang, Yun Peng, Tao Xie · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.27977 · cited by 0 works (OpenAlex) · open-access PDF

German

Anwendungen bieten praktische Funktionen für Benutzer, um tägliche Aufgaben zu erledigen. Jedoch können verschiedene Benutzer unterschiedliche oder sogar widersprüchliche Anforderungen für die gleiche Anwendung haben, während sich die Anforderungen des gleichen Benutzers im Laufe der Zeit ändern können. Die Personalisierung von Anwendungen, die durch Agent-zu-App oder UI-zu-Code-Ansätze erzeugt werden, kann Änderungen an der erzeugten Implementierung erfordern. Wir schlagen Agent als App vor, eine Aufgabe, bei der Benutzer ihre Anforderungen in natürlicher Sprache ausdrücken und ein Agent eine personalisierte Anwendung erstellt, während er sein Verhalten während des Gebrauchs liefert. Wir implementieren LiveUIAgent, das eine Aufgabenorchestrierung erzeugt, die die Benchmark

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

Abstract

Applications provide convenient features for users to complete daily tasks. However, different users may have different or even conflicting requirements for the same application, while the requirements of the same user may change over time. Personalizing applications produced by agent-to-app or UI-to-code approaches can require changes to the generated implementation. We propose Agent as App, a task in which users express their requirements in natural language and an agent creates a personalized application while continuing to supply its behavior during use. We implement LiveUIAgent, which generates a task-specific orchestration connecting the user interface, saved information, and service actions. A shared agent runtime renders and updates the interface while carrying out the requested operations. Revising the orchestration and interface while reusing that runtime provides a lightweight basis for personalization as requirements evolve. We introduce LiveUIBench, a benchmark with diverse requirements and ordered interactions across personal services. The empirical findings show that the evaluated models struggle to complete these workflows reliably. Applications can satisfy many local checks while leaving the overall task unfinished, and greater partial coverage can coincide with fewer completed workflows.

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

Topics: Computer science, Orchestration, Workflow, Personalization, Task (project management), Reuse

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.27977, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

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