[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-86059-en":3,"doc-seo-86059-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},86059,687197207057,"Sage","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","How Do Practitioners Build SE Agents? Insights from a Mixed-Methods Study","The rise of software engineering (SE) agents—LLM-based agents that can understand large codebases and perform engineering tasks with limited human intervention—has outpaced knowledge about how developers build them. A mixed-methods study combines semistructured interviews with 20 practitioners from 12 organizations and an online survey of 80 practitioners. Results show bottlenecks shift as implementation costs drop: requirements, coordination, review, and deployment become more prominent, while reviewing and evaluating agent output becomes central.","How Do Practitioners Build SE Agents? Insights from a Mixed-Methods Study  \nYunbo Lyu 1 , David Williams2 , Jieke Shi 1 , Zhensu Sun 1,* , Chao Peng3 , Zhou Yang4,5 , Federica Sarro2 , David Lo 1  \n1 Singapore Management University, Singapore 2University College London, UK  \n3Tencent, China 4University of Alberta, Canada 5Alberta Machine Intelligence Institute, Canada  \n{yunbolyu, jiekeshi, zssun, [davidlo](davidlo}@smu.edu.sg)[}](davidlo}@smu.edu.sg)[@smu.edu.sg](davidlo}@smu.edu.sg), {david.williams.22, [f.sarro](f.sarro}@ucl.ac.uk)[}](f.sarro}@ucl.ac.uk)[@ucl.ac.uk](f.sarro}@ucl.ac.uk), [chao.peng@acm.org](chao.peng@acm.org), [zy25@ualberta.ca](zy25@ualberta.ca) * Corresponding author  \narXiv :2607 . 10856v 1 [ cs . SE] 12 Jul 2026  \nAbstract—The rise of Software Engineering (SE) agents, i.e., LLM-based agents that can understand large codebases and carry out engineering tasks with limited human intervention, has been marked by rapid advance and adoption, but little is known about how developers build these systems in practice: existing studies mine repositories or examine deployment, but few investigate how SE agents are constructed. Through semistructured interviews with 20 practitioners from 12 organizations and an online survey of 80 practitioners, this paper is the first to study how SE processes are changing in the development of SE agents and what challenges developers face. We find that as implementation becomes cheaper, bottlenecks shift rather than disappear: long-standing non-coding work such as requirements, coordination, review, and deployment becomes more visible, while reviewing and evaluating agent output becomes new and central. We characterize a seven-stage workflow and a shift toward evaluation-driven development, in which evaluation steers iteration and specifications become versioned artifacts read by both humans and agents. We further identify six challenges that teams face, together with the practices they adopt to address them, including unreliable evaluation signals, comprehension debt as code outpaces understanding, and behavioral changes introduced by provider-side model updates.  \nIndex Terms—Software Engineering, Agents, AI4SE, AI coding assistants, SE agents, AI, Software Process.  \nI. INTRODUCTION  \nSoftware Engineering (SE) is being reshaped by agents built on Large Language Models (LLMs) [1], which we refer to as SE agents. Exemplified by tools such as Claude Code [2] and Codex [3], these agents can carry out complex SE tasks, such as feature development, debugging, and code review, with minimal human oversight. Their adoption is already substantial: Anthropic reported that Claude Code authored more than 80% of the code merged into its own codebase as of May 2026 [4], while Cursor, one of the earliest companies to commercialize SE agents, surpassed $2 billion in annualized revenue within four years [5] . These trends suggest that SE agents are becoming routine actors in software development rather than experimental tools.  \nMeanwhile, it is easy to overlook that SE agents are themselves software systems, yet ones whose behavior depends not only on conventional code, but also emerges from the underlying model, prompts, tools, context, and execution environment. As software, they should in principle be amenable to an effective development process; however, because they differ fundamentally from conventional software, that process  \ncannot simply be carried over. This difference is most apparent in debugging: when an SE agent misbehaves, there may beno single faulty statement to repair, since the failure can originate in the model, the harness, 1 a tool interface, or their interaction. The familiar edit-run-debug loop no longer offers a concrete point of intervention. More broadly, when behavior is determined by a model rather than by the code one wrote, the traditional anchors of software development shift, raising the question of how SE agents should be specified, developed, evaluated, deploy","cbCaih8CagqxF6TA","https://ap.wps.com/l/cbCaih8CagqxF6TA","pdf",939338,1,13,"English","en",105,"# Introduction\n## Motivation and Research Gap\n## Study Design and Methodology\n## Key Findings and Workflow Overview","[{\"question\":\"What does the study investigate about SE agents?\",\"answer\":\"How practitioners build SE agents in practice, and how the SE development process changes, including key challenges developers face.\"},{\"question\":\"What research methods were used?\",\"answer\":\"The study uses semistructured interviews with 20 practitioners across 12 organizations and an online survey of 80 practitioners experienced in building SE agents.\"},{\"question\":\"What major shift occurs as implementing SE agents becomes cheaper?\",\"answer\":\"Bottlenecks shift rather than disappear: non-coding work such as requirements, coordination, review, and deployment becomes more visible, while reviewing and evaluating agent output becomes new and central.\"}]",1784208156,33,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"how-do-practitioners-build-se-agents-insights-from-a-mixed-methods-study","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/how-do-practitioners-build-se-agents-insights-from-a-mixed-methods-study/86059/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What does the study investigate about SE agents?","Question",{"text":74,"@type":75},"How practitioners build SE agents in practice, and how the SE development process changes, including key challenges developers face.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What research methods were used?",{"text":79,"@type":75},"The study uses semistructured interviews with 20 practitioners across 12 organizations and an online survey of 80 practitioners experienced in building SE agents.",{"name":81,"@type":72,"acceptedAnswer":82},"What major shift occurs as implementing SE agents becomes cheaper?",{"text":83,"@type":75},"Bottlenecks shift rather than disappear: non-coding work such as requirements, coordination, review, and deployment becomes more visible, while reviewing and evaluating agent output becomes new and central.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":105,"slug":137},19,"General","general"]