TL;DR
This work offers project stakeholders a standards-aligned perspective on where NTD tends to appear in AI-enabled systems, highlighting processes that may systematically contribute to the accumulation of NTD.
Large language models (LLMs) are increasingly relevant to Business Process Management (BPM), particularly when process knowledge is dispersed across documents, conversations, and other unstructured sources. Their probabilistic outputs, however, raise questions about validation, traceability, and accountability. This paper develops a lifecycle-based conceptual framework for allocating and governing LLM use across the six stages of the BPM lifecycle. The framework separates generative interpretation from formal, empirical, and expert validation. It comprises five interdependent layers and six operational principles, implemented through a stage-risk-validation matrix, a principle-intensity map, and four evaluation dimensions. Governance requirements increase as outputs approach live execution or decisions that are difficult to reverse, with controls aligned with the NIST AI Risk Management Framework, the EU AI Act, and the GDPR. A customer complaint-handling scenario demonstrates how the framework can be applied. An illustrative stress test using the BPI Challenge 2017 event log and ten independent LLM generations instantiates the validation layer under information-asymmetric conditions. Although all generated models were structurally valid, the event log revealed incomplete activity coverage and control-flow mismatch. This illustrates the value of an external referent but does not establish comparative performance or a general difference in error detectability between LLM-generated and process-mining artefacts. The framework therefore positions LLMs as tools for turning unstructured information into preliminary process knowledge, while established BPM methods and human expertise remain responsible for validating consequential outputs.
The article identifies three results: master data defects move through reporting and AI pipelines, S/4HANA transformations expose tolerated legacy errors, and AI readiness requires governed data quality gates before model deployment.
Digital transformation in Knowledge-intensive Processes is shifting toward Agentic Business Process Management to overcome challenges posed by unstructured data complexity. This research-in-progress evaluates the structural tension between operational efficiency and compliance in Knowledge-intensive Processes automation by examining Generative AI extraction, database architectures, and performance trade-offs between Python and Low-Code/No-Code platforms under AI governance frameworks. Using Design Science Research, this study proposes a five-pillar conceptual framework and establishes a qualitative baseline through domain expert interviews. Initial findings reveal structural fragmentation as core barriers, driving a 20 to 40% waste premium. This paper provides the architectural foundation and experimental protocol for future quantitative benchmarking.
A process-centric reference architecture designed for technical implementability, traceability, auditability, and human-supervised enterprise-scale GAI adoption is contributed.
This study examines the main operational business activities in a local government unit of the Philippines (the Tuguegarao City Assessor’s Office). It presents how Robotic Process Automation (RPA) can improve its property assessment workflow. A qualitative, process-based approach was used. Data were gathered through structured questionnaires, process walkthroughs, and document reviews across the key business CAO subprocesses. A Phase-1 prototype was also developed in UiPath Studio to test automation logic for data extraction, validation, and computation within the FAAS generation process. The Appraisal and Assessment stage, where the Field Appraisal and Assessment Sheet (FAAS) is prepared, contains the most repetitive and error-prone tasks. Manual transcription from property titles/eTitles, calculation of values, and multiple rounds of data encoding and cross-checking make this the most time-consuming part of the process. Preliminary simulations indicate an estimated 75–85% reduction in FAAS processing time, translating to approximately 3,600 man-hours saved annually across key appraisal and encoding tasks. The UiPath prototype proved that automating property data extraction, valuation computation, and structured data encoding can produce standardized, audit-ready valuation records from the start. The RPA solution helps reduce human error, shorten processing time, and improve the accuracy and timeliness of the city’s revenue forecasts. These improvements support better fiscal planning and enhance the city’s ability to fund local public services. This study provides a practical and governance-aligned model that shows how an LGU can use RPA not only to make processes faster but also to connect legacy systems with new national data standards.
Software delivery in large organisations is shaped by two different needs. Management must maintain control over scope, cost, commitments, risk and compliance, while delivery teams must respond when technical findings, user feedback or external dependencies change the working assumptions. This paper develops a conceptual framework that connects traditional project governance, agile execution and artificial intelligence (AI)-supported decision-making within one operating model. Five applications are considered: effort and capacity estimation, backlog prioritisation, delivery-risk sensing, project knowledge capture and software quality analytics. The framework also treats bias in historical records, data quality, explainability and accountability as project-management concerns. Four propositions link the framework with schedule adherence, scope control, rework, throughput, release quality and stakeholder satisfaction. A staged route for empirical validation and a practical readiness assessment are proposed. The study is conceptual and draws on practitioner-informed observations. It argues that AI creates value only when its outputs are connected to an existing decision process at the point where action is still possible; responsibility for interpreting, accepting or rejecting the recommendation remains with the accountable project participants.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
