רונן ברפמן

The Trust-CUA workshop brings the fast-growing area of computer-using agents (CUAs)– agents that operate across GUIs, browsers, APIs, and CLIs– to the core concerns of the IUI community: human-centered design, trust, and interactive control. We focus on methods, interfaces, and evaluations that make CUAs trustworthy-by-design: predictable, steerable, auditable, and safe for deployment in real organizations. While agent benchmarks have advanced, the intelligent user interface layer that mediates agency, oversight, and learning remains under-specified. This workshop centers the IUI perspective: mixed-initiative interaction, explanation and sensemaking, failure-aware UX, risk communication, and interfaces for policy compliance and recovery. We aim to produce (1) an IUI-oriented checklist for trustworthy CUA interaction design, (2) a shared evaluation profile (“CUBench-IUI”) with user-centered metrics (e.g., time-to-recovery, predictability, oversight effort, and policy-aligned success), and (3) a curated set of design patterns and open research challenges to guide the next year of IUI research on agentic systems.