Search papers, labs, and topics across Lattice.
This paper introduces AdaLens, an interactive system designed to enhance the observability and steerability of long-running agentic data analysis workflows. By integrating a storyline-based representation that consolidates analytical plans, execution progress, and findings, AdaLens allows analysts to effectively monitor and guide the analysis process in real-time. The evaluation through case studies and user feedback demonstrates that AdaLens significantly improves analysts' ability to manage complex, evolving data analyses compared to traditional interfaces.
Real-time monitoring and steering of long-running data analyses can now be achieved through a unified storyline interface, transforming how analysts interact with autonomous workflows.
Large language models are pushing data science toward increasingly autonomous and agentic workflows, with recent systems already supporting multi-step and long-running analyses. As these workflows become more autonomous, conventional interfaces no longer provide adequate support for two critical requirements: observability for understanding an agent's evolving reasoning and evidence, and steerability for redirecting low-value directions or deepening promising ones during execution. Existing interactive approaches improve process visibility and open intervention points, but they remain largely designed for discrete, turn-by-turn exchanges rather than the parallel branches and evolving decision structures of long-running agentic analysis. We study this need as interactive oversight in long-running agentic data analysis and present AdaLens, an interactive system for monitoring and steering ongoing runs. AdaLens combines a storyline-based representation that unifies analytical plans, execution progress, intermediate findings, and data-column involvement with steering interactions grounded in these analytical elements for directional guidance and execution control. We evaluate AdaLens through two case studies and a user study, examining how it supports analysts in monitoring and steering long-running agentic data analysis.
