Adaptive Monitoring and Real-World Evaluation of Agentic AI Systems

Adaptive Monitoring and Real-World Evaluation of Agentic AI Systems | 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 Research Article Adaptive Monitoring and Real-World Evaluation of Agentic AI Systems Manish A. Shukla This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7497109/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 02 Feb, 2026 Read the published version in AI and Ethics → Version 1 posted You are reading this latest preprint version Abstract Agentic artificial intelligence (AI) — multi-agent systems that combine large language models with external tools and autonomous planning — are rapidly transitioning from research laboratories into high-stakes domains. Our earlier “Basic” paper introduced a five-axis framework and proposed preliminary metrics such as goal drift and harm reduction but did not provide an algorithmic instantiation or empirical evidence. This “Advanced” sequel fills that gap. First, we revisit recent benchmarks and industrial deployments to show that technical metrics still dominate evaluations: a systematic review of 84 papers from 2023–2025 found that 83% report capability metrics while only 30% consider human-centred or economic axes[2]. Second, we formalise an Adaptive Multi-Dimensional Monitoring (AMDM) algorithm that normalises heterogeneous metrics, applies per-axis exponentially weighted moving-average thresholds and performs joint anomaly detection via the Mahalanobis distance. Third, we conduct simulations and real-world experiments. AMDM cuts anomaly-detection latency from 12.3 s to 5.6 s on simulated goal drift and reduces false-positive rates from 4.5% to 0.9% compared with static thresholds. We present a comparison table and ROC/PR curves, and we reanalyse case studies to surface missing metrics. Code, data and a reproducibility checklist accompany this paper to facilitate replication. Agentic AI Multi-agent systems Evaluation framework Adaptive Multi-Dimensional Monitoring (AMDM) Online anomaly detection Goal drift Safety and ethics Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 02 Feb, 2026 Read the published version in AI and Ethics → 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-7497109","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":513248569,"identity":"99e0818e-390f-4d29-ab3b-aca7794d9c93","order_by":0,"name":"Manish A. Shukla","email":"data:image/png;base64,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","orcid":"","institution":"Independent Researcher","correspondingAuthor":true,"prefix":"","firstName":"Manish","middleName":"A.","lastName":"Shukla","suffix":""}],"badges":[],"createdAt":"2025-08-30 19:23:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7497109/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7497109/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s43681-026-00990-y","type":"published","date":"2026-02-02T15:57:26+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":102234383,"identity":"c07b43f8-ea52-42e1-a223-09d5b0eb7526","added_by":"auto","created_at":"2026-02-09 16:11:11","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":884677,"visible":true,"origin":"","legend":"","description":"","filename":"BlindEval.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7497109/v1_covered_2a9cbc31-23cd-4365-a5cf-77ce776a7fe7.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Adaptive Monitoring and Real-World Evaluation of Agentic AI Systems","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"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":"Agentic AI, Multi-agent systems, Evaluation framework, Adaptive Multi-Dimensional Monitoring (AMDM), Online anomaly detection, Goal drift, Safety and ethics","lastPublishedDoi":"10.21203/rs.3.rs-7497109/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7497109/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Agentic artificial intelligence (AI) — multi-agent systems that combine large language\nmodels with external tools and autonomous planning — are rapidly transitioning from\nresearch laboratories into high-stakes domains. Our earlier “Basic” paper introduced a\nfive-axis framework and proposed preliminary metrics such as goal drift and harm reduction\nbut did not provide an algorithmic instantiation or empirical evidence. This “Advanced”\nsequel fills that gap. First, we revisit recent benchmarks and industrial deployments\nto show that technical metrics still dominate evaluations: a systematic review of 84\npapers from 2023–2025 found that 83% report capability metrics while only 30% consider\nhuman-centred or economic axes[2]. Second, we formalise an Adaptive Multi-Dimensional\nMonitoring (AMDM) algorithm that normalises heterogeneous metrics, applies per-axis\nexponentially weighted moving-average thresholds and performs joint anomaly detection\nvia the Mahalanobis distance. Third, we conduct simulations and real-world experiments.\nAMDM cuts anomaly-detection latency from 12.3 s to 5.6 s on simulated goal drift and reduces\nfalse-positive rates from 4.5% to 0.9% compared with static thresholds. We present\na comparison table and ROC/PR curves, and we reanalyse case studies to surface missing\nmetrics. Code, data and a reproducibility checklist accompany this paper to facilitate\nreplication.","manuscriptTitle":"Adaptive Monitoring and Real-World Evaluation of Agentic AI Systems","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-12 02:39:06","doi":"10.21203/rs.3.rs-7497109/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":"c92a1b4f-b271-46ce-bad1-0da7e17b67f2","owner":[],"postedDate":"September 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-09T16:06:30+00:00","versionOfRecord":{"articleIdentity":"rs-7497109","link":"https://doi.org/10.1007/s43681-026-00990-y","journal":{"identity":"ai-and-ethics","isVorOnly":false,"title":"AI and Ethics"},"publishedOn":"2026-02-02 15:57:26","publishedOnDateReadable":"February 2nd, 2026"},"versionCreatedAt":"2025-09-12 02:39:06","video":"","vorDoi":"10.1007/s43681-026-00990-y","vorDoiUrl":"https://doi.org/10.1007/s43681-026-00990-y","workflowStages":[]},"version":"v1","identity":"rs-7497109","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7497109","identity":"rs-7497109","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}