Cross-Agent Memory Architecture with Contextual Coherence and Factually Grounded Multi-Agent System

Cross-Agent Memory Architecture with Contextual Coherence and Factually Grounded Multi-Agent System | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Cross-Agent Memory Architecture with Contextual Coherence and Factually Grounded Multi-Agent System Aaryan Agrawal, Pavan Y.D.G, Sathwik B.C., Shreyas D K, Vishwanath Pethri Kamath, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9180420/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The growth of Large Language Models (LLMs) as universal reasoning tool has led to the creation of intelligentagent systems. However, ensuring smooth communication between multiple agents, maintaining contextual consistencyduring tasks and reducing hallucinations which attributes to the model generating false or misleading information is still achallenge. This work presents a comprehensive Agentic AI (AAI) framework that provides organized and context-awarecooperation amongst different agents using modular memory techniques. The framework integrates both intra-agent andcross-agent communication and memory using methods like Retrieval Augmented Generation (RAG), vector memory (e.g.,Qdrant) and context linking for episodic memory. The implementation is for a specific use case of an intelligent shopping assistant by making use of existing platform to coordinate research area of agents focusing on tasks of preference extraction, memory management, searching of products and recommendations. The agents actively access and share knowledge through semantic indexed memory and labelling of metadata. The evaluations of both synthetic and real-world tasks show a 28% reduction in hallucination rates and improvement of 35% in task completion accuracy compared to agents that lack memory. The system enhances coherence, relevance and factual accuracy by grounding it in enduring and shared memory. The work lays a strong foundation for developing dependable, adaptable and scalable Agentic AI systems that could be applied in areas of support for decision making, virtual assistants and self-planning. Physical sciences/Mathematics and computing Biological sciences/Psychology Social science/Psychology Full Text Additional Declarations No competing interests reported. Supplementary Files Supplimentaryinformationethicsapproval.pdf Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9180420","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":629215040,"identity":"2787461b-a086-4e3e-9be2-77953a01ff2d","order_by":0,"name":"Aaryan Agrawal","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Aaryan","middleName":"","lastName":"Agrawal","suffix":""},{"id":629215043,"identity":"ddac8eba-16d9-4420-9e43-86acf705d045","order_by":1,"name":"Pavan Y.D.G","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Pavan","middleName":"","lastName":"Y.D.G","suffix":""},{"id":629215047,"identity":"803cda22-4382-440a-87f1-df30c95e844c","order_by":2,"name":"Sathwik B.C.","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Sathwik","middleName":"","lastName":"B.C.","suffix":""},{"id":629215048,"identity":"1952339e-df2d-437c-8a08-a667f50d24e5","order_by":3,"name":"Shreyas D K","email":"data:image/png;base64,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","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":true,"prefix":"","firstName":"Shreyas","middleName":"D","lastName":"K","suffix":""},{"id":629215053,"identity":"a7216338-8ade-48de-8e1b-92edf658946c","order_by":4,"name":"Vishwanath Pethri Kamath","email":"","orcid":"","institution":"Samsung R\u0026D Institute","correspondingAuthor":false,"prefix":"","firstName":"Vishwanath","middleName":"Pethri","lastName":"Kamath","suffix":""},{"id":629215058,"identity":"e370dfc9-887d-48e7-aaa4-041135948062","order_by":5,"name":"Dr. Nalina V.","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"Dr.","firstName":"Nalina","middleName":"","lastName":"V.","suffix":""},{"id":629215063,"identity":"97b87faf-1577-4011-926a-d5377dcfb888","order_by":6,"name":"Dr. Nalini M.K.","email":"","orcid":"","institution":"B.M.S. College of Engineering","correspondingAuthor":false,"prefix":"Dr.","firstName":"Nalini","middleName":"","lastName":"M.K.","suffix":""}],"badges":[],"createdAt":"2026-03-20 15:39:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9180420/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9180420/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108016067,"identity":"283fd242-5bbd-43fe-ae76-7d646005e89e","added_by":"auto","created_at":"2026-04-28 13:41:16","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":330739,"visible":true,"origin":"","legend":"","description":"","filename":"Samsungprism.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9180420/v1_covered_d84b554a-8f1c-4794-956e-3e6cdd26e0bb.pdf"},{"id":107947534,"identity":"81d00130-63a6-45cd-8980-559c086235b3","added_by":"auto","created_at":"2026-04-27 23:51:55","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":636077,"visible":true,"origin":"","legend":"","description":"","filename":"Supplimentaryinformationethicsapproval.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9180420/v1/5a41ae67039081d6c0871738.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eCross-Agent Memory Architecture with Contextual Coherence and Factually Grounded Multi-Agent System\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-9180420/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9180420/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The growth of Large Language Models (LLMs) as universal reasoning tool has led to the creation of intelligentagent systems. However, ensuring smooth communication between multiple agents, maintaining contextual consistencyduring tasks and reducing hallucinations which attributes to the model generating false or misleading information is still achallenge. This work presents a comprehensive Agentic AI (AAI) framework that provides organized and context-awarecooperation amongst different agents using modular memory techniques. The framework integrates both intra-agent andcross-agent communication and memory using methods like Retrieval Augmented Generation (RAG), vector memory (e.g.,Qdrant) and context linking for episodic memory. The implementation is for a specific use case of an intelligent shopping assistant by making use of existing platform to coordinate research area of agents focusing on tasks of preference extraction, memory management, searching of products and recommendations. The agents actively access and share knowledge through semantic indexed memory and labelling of metadata. The evaluations of both synthetic and real-world tasks show a 28% reduction in hallucination rates and improvement of 35% in task completion accuracy compared to agents that lack memory. The system enhances coherence, relevance and factual accuracy by grounding it in enduring and shared memory. The work lays a strong foundation for developing dependable, adaptable and scalable Agentic AI systems that could be applied in areas of support for decision making, virtual assistants and self-planning.","manuscriptTitle":"Cross-Agent Memory Architecture with Contextual Coherence and Factually Grounded Multi-Agent System","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-27 23:51:50","doi":"10.21203/rs.3.rs-9180420/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"501d92c8-6e4b-4baa-aea3-dbd4cc46f223","owner":[],"postedDate":"April 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67121272,"name":"Physical sciences/Mathematics and computing"},{"id":67121273,"name":"Biological sciences/Psychology"},{"id":67121274,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2026-04-28T13:40:26+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-27 23:51:50","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9180420","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9180420","identity":"rs-9180420","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}