[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84399-en":3,"doc-seo-84399-105":30,"detail-sidebar-cat-0-en-105":91},{"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":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},84399,7971461741311,"Ophelia","https://ap-avatar.wpscdn.com/avatar/74000253aff267980c6?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779345379180704826",8,"Research & Report","Workflow as Knowledge: Semantic Persistence for LLM-Mediated Workflows","Large language model (LLM) applications increasingly rely on explicit workflows for tool use, retrieval, branching, checkpointing, and human approval. This paper proposes a Lisp-inspired, language-independent conceptual model that represents workflow definitions, instances, inference records, context snapshots, and dependency relations as persistent knowledge objects. It distinguishes derive (deterministic computation over available state) from infer (LLM-mediated judgment under declared context and an executor capability policy), enabling inspectable, resumable, and reviewable semantic workflow artifacts.","Workflow as Knowledge: Semantic Persistence for LLM-Mediated Workflows  \nEmanuele Quinto  \nUNHCR København, Denmark [emanuele.quinto@protonmail.ch](emanuele.quinto@protonmail.ch)  \nCarlo Andrea Rozzi  \nCNR—Istituto Nanoscienze Modena, Italy  \nFrancesco Zanitti  \nZeLe & F ApSKøbenhavn, Denmark  \narXiv :2607 .08740v 1 [ cs .AI] 9 Jul 2026  \nAbstract  \nLarge language model (LLM) applications increasingly use explicit workflows for tool use, retrieval, branching, checkpointing, and human approval. Existing workflow systems already address many execution concerns. This paper proposes a Lisp-inspired but language-independent conceptual model: symbolic forms, object identity, and live-image thinking are used as explanatory lenses, not implementation commitments. In this model, workflow definitions, workflow instances, inference records, context snapshots, and dependency relations are represented as persistent knowledge objects in a shared knowledge substrate. Its central semantic distinction is between derive and infer: derive is deterministic computation over available state; infer is mediated LLM judgment under declared context and executor-controlled capability policy.  \nThe result is a preliminary conceptual account of semantic persistence: workflows do not merely produce knowledge and leave traces, but can themselves be represented as inspectable, resumable, and reviewable knowledge objects, while formal transition semantics remain future work.  \n1. Introduction  \nLLM systems have moved from single-turn prompting toward structured workflows. Agent frameworks now expose graphs, typed steps, loops, tool calls, checkpointing, human approval gates, andreusable submodules (AgentSPEX 2026 ; LangGraph 2026; Khattab et al. 2023 ; Josifoski et al. 2023 ; WorkflowLLM 2024) . This shift is necessary: unconstrained reactive prompting leaves control flow implicit, makes intermediate state difficult to inspect, and complicates resumption after interruption.  \nHowever, making execution explicit does not by itself solve a deeper representational problem. A workflow definition may live as source code, declarative configuration, or a database-backed representation. A running occurrence is managed by a runtime layer. As of 2026, practitioner discussions often describe parts of this surrounding control-and-runtime machinery as an agent harness (O’Reilly 2026a) . Intermediate model outputs may be retained as logs, traces, rows, or chat history. Such artifacts can be correlated by infrastructure or provenance mechanisms, but correlation is not the same as assigning them explicit roles in the shared semantic object model developed here.  \nOur proposal first separates three conceptual layers. A lower runtime service layer supplies  \nmodel adapters, tools, external processes, and persistence/indexing facilities.  \nA middle control layer contains the domain-specific language (DSL) machine and its executor: it interprets declared objects, assembles context, mediates model and tool calls, validates results, applies permitted transitions, and exposes the knowledge-substrate interface.  \nA higher semantic layer contains workflow definitions, workflow instances, and their linked inference, approval, and panel records. Contemporary agent harness implementations may package portions of the middle and lower layers together; that packaging is not itself the definition of aworkflow.  \nFigure 1 summarizes the three conceptual layers and the mediation boundary between semantic objects, the DSL-machine control layer, and runtime services.  \nFigure 1 . Semantic workflow objects are interpreted by the DSL-machine control layer, which coordinates  \nruntime services and writes back workflow instances, mediated effects, and records of inference, approval, and panel activity. The bidirectional relation indicates that the control layer both reads semantic objects and writes back persistent semantic objects or relations.  \nWithin this architecture, a workflow-definition is a ","cbCaioFGB1W44m5W","https://ap.wps.com/l/cbCaioFGB1W44m5W","pdf",1306517,3,1,39,"English","en",105,"# Abstract\n# Introduction\n## From single-turn prompting to structured workflows\n## Three conceptual layers: runtime, control, and semantic\n## Semantic persistence and Lisp-inspired lenses","[{\"question\":\"What problem does the paper address in LLM workflow systems?\",\"answer\":\"It targets the deeper representational gap: making execution explicit is not enough, because workflow artifacts are often treated as logs or traces without explicit roles in a shared semantic object model.\"},{\"question\":\"How does the paper define the difference between derive and infer?\",\"answer\":\"Derive is deterministic computation over available state, while infer is mediated LLM judgment under declared context and executor-controlled capability policy.\"},{\"question\":\"What does semantic persistence mean in this model?\",\"answer\":\"Semantic persistence means preserving typed workflow objects and linked inference records—along with identities, relations, and required context—so they remain inspectable beyond a single execution run.\"}]",1784195324,98,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"workflow-as-knowledge-semantic-persistence-for-llm-mediated-workflows","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":20},"https://docshare.wps.com/document/research-report/",{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/workflow-as-knowledge-semantic-persistence-for-llm-mediated-workflows/84399/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-24","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the paper address in LLM workflow systems?","Question",{"text":75,"@type":76},"It targets the deeper representational gap: making execution explicit is not enough, because workflow artifacts are often treated as logs or traces without explicit roles in a shared semantic object model.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the paper define the difference between derive and infer?",{"text":80,"@type":76},"Derive is deterministic computation over available state, while infer is mediated LLM judgment under declared context and executor-controlled capability policy.",{"name":82,"@type":73,"acceptedAnswer":83},"What does semantic persistence mean in this model?",{"text":84,"@type":76},"Semantic persistence means preserving typed workflow objects and linked inference records—along with identities, relations, and required context—so they remain inspectable beyond a single execution run.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]