Amanda Swearngin

Matheus Kunzler Maldaner, Adam Fourney, Amanda Swearngin et al.

AI agents are increasingly asked to carry out work that spans minutes, hours, or longer. Yet the default model of agent behavior is continuous action: issuing tool calls, refreshing pages, searching for alternatives, or otherwise trying to force progress. This is the wrong approach for many long-running tasks, which are better served by a strategy of sustained attention. Instead, agents should monitor an environment, notice when an external event makes progress possible, then respond promptly without wasting resources while waiting. To measure progress on this class of tasks, we introduce SentinelBench, an open-source benchmark for time-evolving monitoring tasks. SentinelBench contains 100 tasks across 10 synthetic web environments, including email, calendars, finance, professional networking, and entertainment. Each environment exposes a live web interface and replays a scripted sequence of events, requiring agents to navigate and reason about web pages whose state shifts underfoot. SentinelBench measures task completion, reaction time, and resource use, exposing the tradeoff between responsiveness and cost. We report results across three models and two browser-agent harnesses, establishing performance baselines for future comparison and demonstrating how agent design choices can dramatically impact key metrics. Together, these results show that SentinelBench distinguishes meaningful differences in agent behavior.

Maryam Taeb, Amanda Swearngin, Eldon Schoop et al.

Developers and quality assurance testers often rely on manual testing to test accessibility features throughout the product lifecycle. Unfortunately, manual testing can be tedious, often has an overwhelming scope, and can be difficult to schedule amongst other development milestones. Recently, Large Language Models (LLMs) have been used for a variety of tasks including automation of UIs, however to our knowledge no one has yet explored their use in controlling assistive technologies for the purposes of supporting accessibility testing. In this paper, we explore the requirements of a natural language based accessibility testing workflow, starting with a formative study. From this we build a system that takes as input a manual accessibility test (e.g., ``Search for a show in VoiceOver'') and uses an LLM combined with pixel-based UI Understanding models to execute the test and produce a chaptered, navigable video. In each video, to help QA testers we apply heuristics to detect and flag accessibility issues (e.g., Text size not increasing with Large Text enabled, VoiceOver navigation loops). We evaluate this system through a 10 participant user study with accessibility QA professionals who indicated that the tool would be very useful in their current work and performed tests similarly to how they would manually test the features. The study also reveals insights for future work on using LLMs for accessibility testing.

Yue Jiang, Eldon Schoop, Amanda Swearngin et al.

Multimodal Vision-Language Models (VLMs) enable powerful applications from their fused understanding of images and language, but many perform poorly on UI tasks due to the lack of UI training data. In this paper, we adapt a recipe for generating paired text-image training data for VLMs to the UI domain by combining existing pixel-based methods with a Large Language Model (LLM). Unlike prior art, our method requires no human-provided annotations, and it can be applied to any dataset of UI screenshots. We generate a dataset of 335K conversational examples paired with UIs that cover Q&A, UI descriptions, and planning, and use it to fine-tune a conversational VLM for UI tasks. To assess the performance of our model, we benchmark it on UI element detection tasks, evaluate response quality, and showcase its applicability to multi-step UI navigation and planning.

Keen You, Haotian Zhang, Eldon Schoop et al.

Recent advancements in multimodal large language models (MLLMs) have been noteworthy, yet, these general-domain MLLMs often fall short in their ability to comprehend and interact effectively with user interface (UI) screens. In this paper, we present Ferret-UI, a new MLLM tailored for enhanced understanding of mobile UI screens, equipped with referring, grounding, and reasoning capabilities. Given that UI screens typically exhibit a more elongated aspect ratio and contain smaller objects of interest (e.g., icons, texts) than natural images, we incorporate "any resolution" on top of Ferret to magnify details and leverage enhanced visual features. Specifically, each screen is divided into 2 sub-images based on the original aspect ratio (i.e., horizontal division for portrait screens and vertical division for landscape screens). Both sub-images are encoded separately before being sent to LLMs. We meticulously gather training samples from an extensive range of elementary UI tasks, such as icon recognition, find text, and widget listing. These samples are formatted for instruction-following with region annotations to facilitate precise referring and grounding. To augment the model's reasoning ability, we further compile a dataset for advanced tasks, including detailed description, perception/interaction conversations, and function inference. After training on the curated datasets, Ferret-UI exhibits outstanding comprehension of UI screens and the capability to execute open-ended instructions. For model evaluation, we establish a comprehensive benchmark encompassing all the aforementioned tasks. Ferret-UI excels not only beyond most open-source UI MLLMs, but also surpasses GPT-4V on all the elementary UI tasks.

Forough Mehralian, Titus Barik, Jeff Nichols et al.

Accessibility is crucial for inclusive app usability, yet developers often struggle to identify and fix app accessibility issues due to a lack of awareness, expertise, and inadequate tools. Current accessibility testing tools can identify accessibility issues but may not always provide guidance on how to address them. We introduce FixAlly, an automated tool designed to suggest source code fixes for accessibility issues detected by automated accessibility scanners. FixAlly employs a multi-agent LLM architecture to generate fix strategies, localize issues within the source code, and propose code modification suggestions to fix the accessibility issue. Our empirical study demonstrates FixAlly's capability in suggesting fixes that resolve issues found by accessibility scanners -- with an effectiveness of 77% in generating plausible fix suggestions -- and our survey of 12 iOS developers finds they would be willing to accept 69.4% of evaluated fix suggestions.

Jason Wu, Yi-Hao Peng, Amanda Li et al.

User interface (UI) design is a difficult yet important task for ensuring the usability, accessibility, and aesthetic qualities of applications. In our paper, we develop a machine-learned model, UIClip, for assessing the design quality and visual relevance of a UI given its screenshot and natural language description. To train UIClip, we used a combination of automated crawling, synthetic augmentation, and human ratings to construct a large-scale dataset of UIs, collated by description and ranked by design quality. Through training on the dataset, UIClip implicitly learns properties of good and bad designs by i) assigning a numerical score that represents a UI design's relevance and quality and ii) providing design suggestions. In an evaluation that compared the outputs of UIClip and other baselines to UIs rated by 12 human designers, we found that UIClip achieved the highest agreement with ground-truth rankings. Finally, we present three example applications that demonstrate how UIClip can facilitate downstream applications that rely on instantaneous assessment of UI design quality: i) UI code generation, ii) UI design tips generation, and iii) quality-aware UI example search.

Gagan Bansal, Wenyue Hua, Zezhou Huang et al. · microsoft-research

As LLM agents advance, they are increasingly mediating economic decisions, ranging from product discovery to transactions, on behalf of users. Such applications promise benefits but also raise many questions about agent accountability and value for users. Addressing these questions requires understanding how agents behave in realistic market conditions. However, previous research has largely evaluated agents in constrained settings, such as single-task marketplaces (e.g., negotiation) or structured two-agent interactions. Real-world markets are fundamentally different: they require agents to handle diverse economic activities and coordinate within large, dynamic ecosystems where multiple agents with opaque behaviors may engage in open-ended dialogues. To bridge this gap, we investigate two-sided agentic marketplaces where Assistant agents represent consumers and Service agents represent competing businesses. To study these interactions safely, we develop Magentic-Marketplace -- a simulated environment where Assistants and Services can operate. This environment enables us to study key market dynamics: the utility agents achieve, behavioral biases, vulnerability to manipulation, and how search mechanisms shape market outcomes. Our experiments show that frontier models can approach optimal welfare -- but only under ideal search conditions. Performance degrades sharply with scale, and all models exhibit severe first-proposal bias, creating 10-30x advantages for response speed over quality. These findings reveal how behaviors emerge across market conditions, informing the design of fair and efficient agentic marketplaces.

Jason Wu, Amanda Swearngin, Arun Krishna Vajjala et al.

Despite being trained on vast amounts of data, most LLMs are unable to reliably generate well-designed UIs. Designer feedback is essential to improving performance on UI generation; however, we find that existing RLHF methods based on ratings or rankings are not well-aligned with designers' workflows and ignore the rich rationale used to critique and improve UI designs. In this paper, we investigate several approaches for designers to give feedback to UI generation models, using familiar interactions such as commenting, sketching and direct manipulation. We first perform a study with 21 designers where they gave feedback using these interactions, which resulted in ~1500 design annotations. We then use this data to finetune a series of LLMs to generate higher quality UIs. Finally, we evaluate these models with human judges, and we find that our designer-aligned approaches outperform models trained with traditional ranking feedback and all tested baselines, including GPT-5.

Xiaoyi Zhang, Lilian de Greef, Amanda Swearngin et al.

Many accessibility features available on mobile platforms require applications (apps) to provide complete and accurate metadata describing user interface (UI) components. Unfortunately, many apps do not provide sufficient metadata for accessibility features to work as expected. In this paper, we explore inferring accessibility metadata for mobile apps from their pixels, as the visual interfaces often best reflect an app's full functionality. We trained a robust, fast, memory-efficient, on-device model to detect UI elements using a dataset of 77,637 screens (from 4,068 iPhone apps) that we collected and annotated. To further improve UI detections and add semantic information, we introduced heuristics (e.g., UI grouping and ordering) and additional models (e.g., recognize UI content, state, interactivity). We built Screen Recognition to generate accessibility metadata to augment iOS VoiceOver. In a study with 9 screen reader users, we validated that our approach improves the accessibility of existing mobile apps, enabling even previously inaccessible apps to be used.

Amanda Swearngin, Chenglong Wang, Alannah Oleson et al.

Although exploring alternatives is fundamental to creating better interface designs, current processes for creating alternatives are generally manual, limiting the alternatives a designer can explore. We present Scout, a system that helps designers rapidly explore alternatives through mixed-initiative interaction with high-level constraints and design feedback. Prior constraint-based layout systems use low-level spatial constraints and generally produce a single design. Tosupport designer exploration of alternatives, Scout introduces high-level constraints based on design concepts (e.g.,~semantic structure, emphasis, order) and formalizes them into low-level spatial constraints that a solver uses to generate potential layouts. In an evaluation with 18 interface designers, we found that Scout: (1) helps designers create more spatially diverse layouts with similar quality to those created with a baseline tool and (2) can help designers avoid a linear design process and quickly ideate layouts they do not believe they would have thought of on their own.

Amanda Swearngin, Yang Li

Tapping is an immensely important gesture in mobile touchscreen interfaces, yet people still frequently are required to learn which elements are tappable through trial and error. Predicting human behavior for this everyday gesture can help mobile app designers understand an important aspect of the usability of their apps without having to run a user study. In this paper, we present an approach for modeling tappability of mobile interfaces at scale. We conducted large-scale data collection of interface tappability over a rich set of mobile apps using crowdsourcing and computationally investigated a variety of signifiers that people use to distinguish tappable versus not-tappable elements. Based on the dataset, we developed and trained a deep neural network that predicts how likely a user will perceive an interface element as tappable versus not tappable. Using the trained tappability model, we developed TapShoe, a tool that automatically diagnoses mismatches between the tappability of each element as perceived by a human user---predicted by our model, and the intended or actual tappable state of the element specified by the developer or designer. Our model achieved reasonable accuracy: mean precision 90.2\% and recall 87.0\%, in matching human perception on identifying tappable UI elements. The tappability model and TapShoe were well received by designers via an informal evaluation with 7 professional interaction designers.