Smart Document Management for Modern Secretarial Practice with AI Facilitation

Thesis Abstract

The rapid digitization of administrative workflows and the expanding role of AI in document-centric tasks have transformed the secretarial function, yet many organizations struggle with inconsistent adoption, data silos, and inefficiencies in document lifecycle management. This study investigates how smart document management systems (SDMS) augmented with AI facilitation can enhance efficiency, accuracy, and strategic support in modern secretarial practice. The aim is to evaluate the effectiveness, adoption determinants, and contextual benefits of AI-enabled SDMS in corporate secretarial settings. Specific objectives are (1) to assess the impact of AI-assisted features (autoclassification, semantic search, and automated workflow routing) on task completion time and error rates; (2) to examine user acceptance, perceived usefulness, and perceived ease of use using the Technology Acceptance Model (TAM) extended with perceived risk and trust in automation; (3) to analyze changes in compliance quality and audit readiness resulting from standardized metadata and version control; (4) to identify organizational factors (training, governance, and data governance maturity) that mediate SDMS deployment outcomes; and (5) to develop a framework for aligning SDMS capabilities with secretarial practice scenarios across professional service firms. A mixed-methods research design is adopted, integrating a quasi-experimental component with a cross-sectional survey and in-depth interviews. The study will be conducted in three professional service firms of comparable size (approximately 350–500 employees) that have implemented AI-enabled SDMS in the past 12–18 months. A purposive sample of 60 secretarial and administrative staff will participate in a controlled time-motion study comparing baseline manual/document-centric workflows with AI-assisted workflows over a eight-week period. For the survey phase, a stratified sample of 180 employees across the three organizations will be invited to provide responses, with a target response rate of at least 70%. Qualitative data will be collected from 18 semi-structured interviews with senior secretarial managers, IT stewards, and compliance officers. Data collection instruments include a validated multi-item TAM questionnaire, a workflow efficiency dashboard capturing task duration and error rates, a compliance and audit-readiness checklist, and interview guides designed to elucidate user experiences and governance concerns. Validity and reliability will be ensured through pilot testing, triangulation of quantitative and qualitative data, and interview coding checks with intercoder reliability (Cohen’s kappa) targeting at least 0. Eight. Analytical techniques will include descriptive statistics and inferential analyses for quantitative data. Regression analysis will test the effects of AI features on task performance, while structural equation modeling (SEM) will examine relationships among perceived usefulness, perceived ease of use, trust in automation, and behavioral intention to use SDMS. Time-motion data will be analyzed using paired t-tests and repeated-measures ANOVA to identify efficiency gains. Qualitative data will undergo thematic analysis following Braun and Clarke’s methodology, with coding to identify themes related to adoption, perceived risks, and governance. A composite index for compliance quality will be constructed from the audit-readiness checklist and metadata completeness measures, with multilevel modeling employed to assess organizational-level moderators. Expected findings include significant reductions in document processing time (25–40%), lower misclassification rates (10–25%), and improved compliance audit outcomes (notably in metadata completeness and version control traceability) in AI-facilitated workflows. The study anticipates that perceived usefulness and trust in automation will robustly predict intention to use, with training quality and governance maturity strengthening effects. Theoretical contribution will extend TAM by integrating trust and risk constructs within the context of AI-assisted document management in professional secretarial practice, and enrich the literature on technology-enabled administrative workflows with a practical framework for SDMS deployment. Practical implications include a validated measurement framework for evaluating AI-enabled SDMS, guidelines for training and governance, and a deployment blueprint tailored to secretarial roles in professional services. The study will conclude that AI-facilitated SDMS can meaningfully augment secretarial productivity, accuracy, and strategic value when combined with robust governance, targeted training, and careful change management. Recommendations emphasize staged implementation, continuous monitoring of metadata standards, and ongoing user engagement to sustain gains and manage systemic risks.