DESIGNING RESILIENT ENTERPRISE SYSTEMS THROUGH AI-DRIVEN DEPENDENCY MAPPING AND IMPACT ANALYSIS

Abstract

Enterprise systems are now often spread across a range of hidden technical and organizational dependencies; technical debt can accumulate, configurations can change, and failures can propagate. Redundancy or post-incident recovery alone is not enough for resilience design; such environments require continuously updated dependency knowledge and credible impact analysis before, during, and after change. This review considers peer-reviewed journal literature published between 2015 and 2024 that is relevant to the dependency mapping and impact analysis of resilient enterprise systems using AI. The studies included in the review are grouped around microservice architecture, process mining, digital twins, graph learning, anomaly detection, explainable artificial intelligence and resilience measurement, as well as AIOps. Based on the literature, it is shown that AI-based mapping can increase observability of runtime relations, aid in change-risk prioritization, and aid in failure localization with architectural models and operational telemetry. There are still gaps in causal validation, cross-layer dependency representation, data quality, explainability and evaluation against real enterprise incidents. There is scholarly value in the field, as resilient design increasingly relies on dynamic, evidence-based knowledge of how enterprise elements behave under stress, change and degradation.