[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83805-en":3,"doc-seo-83805-105":30,"detail-sidebar-cat-0-en-105":92},{"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},83805,5909877438554,"Maeve","https://ap-avatar.wpscdn.com/avatar/5600025385ad2bf12a7?_k=1778553567797529272",8,"Research & Report","Memory-Orchestrated Semantic System (MOSS) Auditable Agentic Memory Architecture","Long-term memory remains a core weakness for AI agents, especially when relying on retrieval-augmented generation (RAG) with embedding similarity search, which is hard to audit and limited by vector representation theory. The Memory-Orchestrated Semantic System (MOSS) is an agentic memory architecture where the agent drives retrieval over a structured relational database. MOSS is model-agnostic, storage-agnostic, and API-agnostic, using symbolic, reproducible SQL retrieval with full logging and inspectable steps. Concept vocabulary is derived inductively from the corpus. A year-long deployment indexes an individual scholar’s conversational and document corpus, supporting auditable, sovereign, unbounded memory for long-horizon assistance.","arXiv :2607 .0439 1v 1 [ cs .CL] 5 Jul 2026  \nMemory-Orchestrated Semantic System (MOSS): An Auditable  \nAgentic Memory Architecture  \nSerge Lacasse* , Jérémie Hatier†, Alex Baker†  \n*  \nFaculté de musique, Université Laval, Québec, Canada  \n† Faculté des sciences et de génie, Université Laval, Québec, Canada  \nCorrespondence: serge.lacasse@mus.ulaval. ca  \nAbstract  \nLong-term memory remains a structural weakness of AI agents. The dominant approach, retrieval-augmented generation (RAG), relies on embedding-based similarity search, which is opaque by construction, difficult to audit, and bounded by the theoretical limits of vector representations. We present the Memory-Orchestrated Semantic System (MOSS), an agentic memory architecture in which the agent (not a similarity pipeline) drives retrieval over a structured relational database. MOSS is model-agnostic, storage-agnostic, and API-agnostic:  \nit runs on any relational engine, connects to any LLM provider (or to deterministic non-LLM processes), and deploys on any infrastructure, local or cloud. Its retrieval execution is symbolic and reproducible—once a query is formulated, no LLM participates in the retrieval loop—and every step of the system, from indexing to answer formulation, is logged and inspectable, making MOSS auditable by construction. Rather than imposing an external ontology, MOSS derives its conceptual vocabulary inductively from the corpus itself. We report on a longitudinal deployment unique in the agentic-memory literature: roughly a year of continuous production (since May 2025) over the complete working corpus of an individual scholar—a conversational corpus reaching back to October 2024 (some 44 million tokens, retroactively indexed) comprising  \n110,183 segments, alongside 163,494 catalogued documents, 569 inductively derived concepts, 322,662 concept annotations, and eleven metadata graphs totalling approximately five million relations—across four successive infrastructure generations. While the present case is that of a single researcher, the architecture is in no way specific to one person: it serves a team, an institution, or any entity that accumulates knowledge over time. We argue that auditable, sovereign, structurally unbounded memory is a precondition for AI agents intended to accompany a person or an organization over years rather than sessions.  \nKeywords: agentic memory, long-term memory, AI agents, retrieval-augmented generation, auditability, data sovereignty, knowledge organization, exocortex  \n1 Introduction  \nLarge language models are stateless. Whatever continuity an AI agent exhibits across sessions must be supplied by infrastructure surrounding the model; what recent literature calls the harness [Zhang et al., 2026] . The formula Agent = Model + Harness has become standard. Within that harness, memory is the component that determines whether an agent can accompany, and interact with, a person or an organization over months and years, or whether every conversation begins, cognitively speaking, from zero. This paper centers on a single deployed case—the working corpus of  \nan individual scholar—but the architecture is in no way specific to one person: the same machinery serves a team, an institution, or any entity that accumulates a body of knowledge over time (Section 5.3) . The individual case is our starting point, not our limit; it happens, as we argue, to be the hardest one.  \nThe ambition itself is old. Bush [1945] imagined the memex, a device holding all of a person’s books, records, and communications, navigable by association; Gemmell, Bell and colleagues built MyLifeBits [Gemmell et al., 2002, 2006], storing one researcher’s entire digital life (notably in a SQL database) two decades before large language models existed. What that lineage lacked was not storage or structure but the navigator: an agent intelligent enough to traverse a lifetime of organized memory in the service of ongoing thought.  \nThe industry’s default ans","cbCainXiofUoPFFF","https://ap.wps.com/l/cbCainXiofUoPFFF","pdf",1142832,5,1,22,"English","en",105,"# Abstract\n# 1 Introduction\n## Background and motivation\n## Limits of RAG for long-term memory\n## Prior work and field directions\n# 2 MOSS overview and deployment (implied)","[{\"question\":\"What problem does MOSS target in AI agents’ long-term memory?\",\"answer\":\"It addresses structural weaknesses of agent long-term memory when using retrieval-augmented generation, where embedding-based similarity search is opaque, difficult to validate, and theoretically bounded.\"},{\"question\":\"How does MOSS retrieve information, and why is it auditable?\",\"answer\":\"MOSS runs retrieval over a structured relational database using symbolic, deterministic execution (SQL). Once a query is formulated, no LLM participates in the retrieval loop, and every step is logged for inspection, making the process auditable by design.\"},{\"question\":\"What deployment and data scale does the paper report for MOSS?\",\"answer\":\"It reports about a year of continuous production starting May 2025, indexing a conversational corpus back to October 2024 plus 163,494 catalogued documents. The indexed resources include 110,183 segments, inductively derived concepts and annotations, and multiple metadata graphs totaling roughly five million relations across infrastructure generations.\"}]",1784190524,55,{"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":87,"head_meta":89,"extra_data":91,"updated_unix":28},"memory-orchestrated-semantic-system-moss-auditable-agentic-memory-architecture","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"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":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/memory-orchestrated-semantic-system-moss-auditable-agentic-memory-architecture/83805/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":24,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-07-27","2026-07-16",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What problem does MOSS target in AI agents’ long-term memory?","Question",{"text":76,"@type":77},"It addresses structural weaknesses of agent long-term memory when using retrieval-augmented generation, where embedding-based similarity search is opaque, difficult to validate, and theoretically bounded.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does MOSS retrieve information, and why is it auditable?",{"text":81,"@type":77},"MOSS runs retrieval over a structured relational database using symbolic, deterministic execution (SQL). Once a query is formulated, no LLM participates in the retrieval loop, and every step is logged for inspection, making the process auditable by design.",{"name":83,"@type":74,"acceptedAnswer":84},"What deployment and data scale does the paper report for MOSS?",{"text":85,"@type":77},"It reports about a year of continuous production starting May 2025, indexing a conversational corpus back to October 2024 plus 163,494 catalogued documents. The indexed resources include 110,183 segments, inductively derived concepts and annotations, and multiple metadata graphs totaling roughly five million relations across infrastructure generations.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":93},[94,98,102,106,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},"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":20,"slug":138},19,"General","general"]