Optimizing LLM-based workflows is typically formulated as a global search, where candidate workflows are evaluated based on a scalar metric. This paradigm, however, suffers from a critical flaw: information collapse. By reducing rich, multi-step execution traces to simple success/failure signals, existing methods are rendered blind to the underlying structure of failures, fundamentally preventing them from modeling the workflow's failure distribution. We reconceptualize this challenge as a distributional problem. We propose a new paradigm where the optimization goal is not to maximize a scalar score, but to directly minimize a workflow's Expected Failure Mass, i.e., the integral of its failure probability density function defined over a high-dimensional Failure Signature Space (FSS). This distributional lens allows us to move from inefficient, zero-order optimization to a principled, gradient-like descent on the failure landscape itself. We introduce CE-Graph, a framework that operationalizes this paradigm through a novel, failure-driven refinement process. CE-Graph approximates the failure distribution from a pool of counterexamples, identifies its densest regions as recurring failure modes, and applies targeted, operator-constrained graph edits via a Propose-and-Verify mechanism to greedily reduce the failure mass. On math, code, and QA benchmarks, our CE-Graph achieves higher robustness at a significantly lower cost than strong baselines. This suggests that a system's reliability emerges not from avoiding failures, but from systematically learning and reshaping the geometric structure of its failure distributions.
Do Language Models Converge to Themselves? Recursive Self-Refinement as Textual Relaxation
Pfizer(辉瑞公司);University of California San Diego(加利福尼亚大学圣地亚哥分校);Institute for Medical Philosophy and Future Artificial Intelligence(医学哲学与未来人工智能研究所);Tongji University(同济大学);Shanghai Jiao Tong University(上海交通大学);Institute of Humanities and Social Science Data, Fudan University(复旦大学人文社会科学数据研究所)
RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent
Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing(未来区块链与隐私计算北京高精尖创新中心);School of Artificial Intelligence(人工智能学院);Beihang University(北京航空航天大学);Tsinghua University(清华大学);Beijing Academy of Artificial Intelligence (BAAI)(北京人工智能研究院)
Text-to-Layout: A Generative Workflow for Drafting Architectural Floor Plans Using LLMs
Department of Civil and Environmental Engineering, University of Houston(土木与环境工程系,休斯顿大学);Center for Transportation Research, The University of Texas at Austin(交通研究中心,德克萨斯大学奥斯汀分校);Department of Information Science Technology, University of Houston(信息科学与技术系,休斯顿大学)
Complexity and Scale in AI-Assisted Workflow Management: A Federated Learning Case Study
RENCI, University of North Carolina at Chapel Hill, NC, USA(RENCI,北卡罗来纳大学教堂山分校);Information Sciences Institute, University of Southern California, Marina del Rey, CA, USA(信息科学研究所,南加州大学马里纳德尔雷耶斯分校)
ATWL: A Formal Language for Representing, Comparing, and Reusing Visual Analytics Workflows
Fraunhofer Institute IAIS(弗劳恩霍夫研究所IAIS);Lamarr Institute for Machine Learning and Artificial Intelligence(拉马尔机器学习与人工智能研究所);City St George’s, University of London(伦敦大学城市圣乔治学院);University of Zurich(苏黎世大学);University of Stuttgart(斯图加特大学)
Plausible but Wrong: A case study on Agentic Failures in Astrophysical Workflows
Bonn-Aachen International Center for Information Technology, University of Bonn, Germany(波恩-阿森国际信息科技中心,波恩大学,德国);Lamarr Institute for Machine Learning and Artificial Intelligence, Germany(拉马尔机器学习与人工智能研究所,德国)
University of Sydney(悉尼大学);University of Adelaide(阿德莱德大学);Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学);Carnegie Mellon University(卡内基梅隆大学)
ProfiliTable: Profiling-Driven Tabular Data Processing via Agentic Workflows
School of CS & Key Lab of High Confidence Software Technologies (MOE), Peking University(计算机科学学院及高可信软件技术重点实验室(教育部),北京大学);Academy for Advanced Interdisciplinary Studies, Peking University(先进跨学科研究学院,北京大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);School of Software & Microelectronics, Peking University(软件与微电子学院,北京大学);School of CS & Beijing Key Laboratory of Software and Hardware Cooperative Artificial Intelligence Systems, Peking University(计算机科学学院及北京软件与硬件协同人工智能系统重点实验室,北京大学);Center for Machine Learning Research, Peking University(机器学习研究中心,北京大学)
ClawForge: Generating Executable Interactive Benchmarks for Command-Line Agents
University of North Carolina at Chapel Hill(北卡罗来纳大学切里波因特分校);Stanford University(斯坦福大学);University of California, Berkeley(加州大学伯克利分校);University of Southern California(南加州大学);University of California, Santa Cruz(加州大学圣克鲁兹分校)