Combining accounting systems with machine learning can turn MSME transaction data into predictive intelligence that supports business-model innovation and organisational agility, offering a pathway to sustainable competitive advantage — but resource, skills and institutional barriers must be addressed and the framework remains untested in the field.

Main Finding

Firdaus, Soegoto & Wiganarto (2026) develop a literature-grounded conceptual technopreneurship framework showing how Accounting Information Systems (AIS) + Machine Learning (ML) can enable Sustainable Competitive Advantage (SCA) for MSMEs. Using a PRISMA-guided systematic review (2,578 records → 69 eligible articles, Jan 2022–May 2026) and the first three echelons of an echeloned Design Science Research (eDSR) approach, the paper identifies three core MSME problem clusters (capacities & resources; digitalisation paradox; institutional/environmental factors) and positions AIS as the structured data foundation, ML-based predictive analytics as the analytical competence, and AI–ML technopreneurship as the primary mechanism that—via mediators such as Business Model Innovation, Organisational Agility, Digital Absorptive Capacity and Dynamic Capabilities—produces SCA and sustainable performance.

DOI: https://doi.org/10.34010/aisthebest.v11i1.20458

Key Points

• Problem clusters for MSMEs:
  • Capacity & resources constraints (finance, legacy systems, skills).
  • Digitalisation paradox: investment ≠ automatic performance gains because of organizational/environmental frictions.
  • Institutional/environmental barriers (infrastructure gaps, policy-practice mismatch, trust deficits).
• Conceptual architecture:
  • Antecedents/Enablers: internal (digital leadership, entrepreneurial mindset, data strategy, financial readiness) and external (market dynamics, infrastructure, government policy).
  • Data foundation: AIS providing structured transactional, financial, operational and customer data suitable for ML.
  • Analytical competence: ML-driven predictive analytics (cash-flow forecasting, risk prediction, anomaly detection, inventory optimization).
  • Mechanism: AI–ML-based technopreneurship — turning predictive intelligence into entrepreneurial actions.
  • Mediators: Business Model Innovation (BMI), Organizational Agility, Digital Absorptive Capacity, Dynamic Capabilities.
  • Outcomes: Sustainable Competitive Advantage (strategic position) and Sustainable Performance (economic, environmental, social).
• Theoretical integration: Technology–Organization–Environment (TOE), Resource-Based View (RBV), Dynamic Capabilities View (DCV).
• AI capability is framed as multidimensional: tangible (data infra), intangible (culture, knowledge management), and human/skill resources.
• Adoption gap: cited finding that only ~7% small and 15% medium firms have undertaken AI projects versus 59% for large firms—highlighting scale-related adoption inequality.
• Contributions:
  • Theoretical: integrates TOE, RBV, DCV around AIS+ML for MSMEs.
  • Methodological: combines PRISMA-guided SLR with partial eDSR conceptual design.
  • Practical: provides a structured roadmap (conceptual) for MSMEs to leverage AIS+ML for technopreneurship.
• Limitations: conceptual only (no prototype, demonstration or empirical evaluation). Echelons 4–5 (demonstration, evaluation) are left for future work.

Data & Methods

• Literature base: Scopus search covering Jan 2022 – May 2026; final set = 69 peer‑reviewed articles.
• Search query combined terms for technopreneurship/digital entrepreneurship, AI/ML, MSME/SME, competitive advantage/performance, and framework/model.
• PRISMA 2020 procedure used for identification, screening, eligibility, inclusion.
• Thematic synthesis: mixed deductive–inductive coding. Deductive codes derived from TOE, RBV, DCV; inductive coding captured emerging themes.
• Conceptual design: used first three echelons of eDSR (problem analysis → objectives/requirements → conceptual framework design). Did not perform artifact implementation, demonstration, or empirical validation.
• Suggested ML use-cases (from literature synthesis) include cash-flow forecasting, risk prediction, anomaly detection, and inventory optimization using AIS data as training inputs.

Implications for AI Economics

Practical and research implications relevant to AI economics, policy and empirical work:

• Macroeconomic and firm-level productivity
  • AIS+ML can reduce information frictions, improve forecasting and decision-making, and raise firm-level productivity—especially for cash-constrained MSMEs. Economists should quantify these gains via firm-level outcome measures (TFP, value-added, profit margins).
• Heterogeneity and distributional effects
  • Adoption is skewed by firm size, skills and finance; without interventions, AI may widen productivity and market-power gaps between large firms and MSMEs. Models should allow for heterogeneous adoption thresholds and complementarities (human capital, infra).
• Market structure and competition
  • Improved predictive capability and business-model innovation can change competitive dynamics—potentially enabling faster entry/exit, market segmentation, or concentration. Structural models and competition policy analysis should account for AIS+ML as strategic assets (RBV perspective).
• Policy levers and targeting
  • Evidence suggests government incentives can improve adoption links. Policy experiments (subsidies for AIS upgrades, training vouchers, targeted infrastructure investments) are natural candidates for RCTs or quasi-experimental evaluation to estimate take-up elasticities and welfare effects.
• Measurement and data needs
  • Empirical evaluation requires granular microdata: firm accounting/AIS logs, ML adoption status, ML outputs (predictions used), and firm outcomes (sales, employment, margins, survival). Encourage collection of panel AIS-compatible administrative data or survey modules linking to accounting systems.
• Empirical strategies to estimate causal effects
  • Difference-in-differences exploiting staggered rollouts of AIS/ML subsidies or platform access; instrumental variables based on exogenous policy exposure or infrastructure shocks; randomized trials for training/subsidy programs; regression discontinuity around eligibility thresholds for grants; matched-pair comparisons for pilot deployments.
• Economic modeling directions
  • Build diffusion models capturing complementarities (skills, finance, infra) and endogenous dynamic capabilities; quantify returns to scale in AI investments and the role of BMI/organizational agility as mediators.
• Welfare, sustainability and externalities
  • Evaluate non-financial outcomes (environmental and social performance) included in the framework—estimate co-benefits or trade-offs (e.g., employment effects vs productivity gains).
• Measurement of value creation
  • Use intermediate metrics tied to ML outputs (forecast accuracy, reduction in cash-flow volatility, inventory turnover improvements, fraud/anomaly detection rates) as proximate outcomes linking ML use to economic performance.
• Research agenda (high‑priority empirical questions)
  • What is the causal impact of AIS+ML adoption on firm survival, productivity and employment in MSMEs?
  • Which mediators (BMI, agility, absorptive capacity) most strongly transmit AI gains to performance?
  • How do infrastructure, policy, and finance constraints shape adoption thresholds and returns?
  • Do AIS+ML interventions reduce or increase inequality across firms/sectors/regions?
  • Cost–benefit analyses for public subsidies to accelerate adoption versus market-led diffusion.

Short policy recommendations for economists/policymakers - Targeted subsidies for AIS upgrades and training to address the adoption gap; accompany with evaluation designs. - Support data interoperability standards and secure AIS platforms to lower integration costs and trust barriers. - Fund pilot deployments and share performance data (anonymized) to accelerate learning and provide causal evidence to guide scale-up.

Overall, the paper provides a clear conceptual map linking AIS-derived data and ML analytics to technopreneurship and SCA in MSMEs. For AI economists, the framework highlights key mechanisms and measurable intermediate outcomes to prioritize in empirical work and policy evaluation.