Digital tools and AI can help Congolese SMEs secure finance and boost performance, but only when firms produce reliable, verifiable data and embed tools into management routines; absent skills, infrastructure and institutional trust, AI acts as an assistive technology, not a standalone solution.

Main Finding

Digitalisation and emergent uses of AI can support formalisation, access to finance and firm performance for African SMEs, but only conditionally. Their economic value emerges when digital tools improve the quality, traceability and verifiability of management data and are embedded into organisational routines under adequate infrastructure, skills, affordable finance and institutional trust. AI is an assistive technology (indirect effect) rather than an autonomous determinant of firm performance.

Key Points

• The paper integrates three theoretical lenses: dynamic capabilities (technology must be embedded in routines), information asymmetry (digital traces can reduce opacity), and institutional approaches to formalisation (cost–benefit of registering).
• Proposed causal chain (non‑deterministic): digitalisation → improved traceability/quality of information → greater observability for state and lenders → increased likelihood of formalisation and better credit assessment → improved financing and performance.
• AI’s role is primarily in classification, forecasting and decision support; its effectiveness is mediated by data quality, organisational routines and human supervision.
• Four explicit, testable propositions:
  • P1: Digitalisation supports formalisation when it lowers administrative transaction costs and produces records reusable in official procedures.
  • P2: Digital traceability reduces information asymmetry if data are regular, verifiable and linked to real activity.
  • P3: AI’s effect on performance is mainly indirect and mediated by data quality, organisational capacity and human oversight.
  • P4: Access to finance improves performance only if financing cost, maturity and use are compatible with the SME’s productive capacity and management.
• Congo (RDC) context (Enterprise Surveys 2024 / World Bank):
  • 35.3% of firms report access to finance as the main obstacle.
  • 59.0% face competition from informal/unregistered firms.
  • Only 7.6% use banks to finance investments.
  • GUCE mobile registration caravans recorded 4,613 MPME registered (2025) — improved access to registration but uncertain permanence.
• Important moderators: firm size and age, sector, manager skills, electricity/internet access, cybersecurity, cost of credit and institutional trust.
• Limitations acknowledged: narrative (not systematic) review, no primary firm‑level survey on AI use in RDC, Enterprise Survey covers mainly formal firms with ≥5 employees.

Data & Methods

• Method: structured narrative literature review (analytical synthesis, not causal inference).
• Corpus: peer‑reviewed studies on digital transformation and finance in Africa, OECD/GSMA reports, institutional normative texts (ILO, UNESCO), and official Congo data (Enterprise Surveys 2024; World Bank analyses).
• Timeframe: emphasis on 2021–September 2026, with classic theoretical references retained where relevant.
• Inclusion criteria: sources directly applicable to RDC or SMEs in Africa, empirical SME results, theoretical mechanisms used in the model, or normative frameworks. Excluded non‑verifiable, promotional or context‑irrelevant sources.
• Thematic analysis on six domains: digital capacities & routines; AI uses; formalisation & state relations; economic information & finance; performance steering; ethical/regulatory/infrastructural risks.
• Evidence hierarchy and verification: official Congo statistics cross‑checked; peer‑reviewed empirical studies used for mechanism validation; normative claims tied to institutional texts (ILO, UNESCO).
• No primary data collection; results are conceptual and produce testable analytical relations for future empirical work.

Implications for AI Economics

• Theory and framing
  • Treat AI as complementary capital that yields value only when coupled with data quality and dynamic organisational capabilities. Avoid techno‑deterministic assumptions in economic models of firm productivity.
  • Models of credit markets should incorporate observability from digital traces as an endogenous signal that interacts with lenders’ practices, credit costs and institutions.
• Measurement & empirical agenda
  • Future micro‑surveys should measure: intensity/type of AI use (automation vs decision support), degree of integration with accounting/transactions, data verifiability, managerial routines, cyber practices, and financing terms.
  • Suggested empirical strategies: firm‑level panel data, randomized encouragement to adopt integrated digital bookkeeping, differences‑in‑differences around registration/registry digitisation interventions, instrumental variables for exogenous connectivity or subsidies, and use of digital transactional logs as objective outcomes.
  • Evaluate heterogeneous effects by firm size, sector, urban/rural location and institutional trust.
• Finance & market design
  • Digital traces can enable alternative credit scoring and reduce frictions, but lenders’ adoption depends on perceived reliability, regulation and cost of funds. FinTech solutions should be assessed jointly with changes in lending practices and interest rates.
• Policy and regulation
  • Policies that lower registration costs, digitise public registries, and integrate business digital records with official procedures can strengthen the digitalisation→formalisation channel.
  • Complement subsidies for tools with training to build routines and human supervision skills.
  • Invest in foundational infrastructure (affordable internet, electricity), cybersecurity, and data governance to ensure verifiability and trust.
  • AI governance: require explainability, human‑in‑the‑loop oversight and protections against biased automated credit decisions.
• Practical research priorities for AI economists
  • Quantify how much variance in lenders’ credit decisions can be explained by added digital traceability versus traditional signals.
  • Estimate the marginal productivity of AI tools conditional on measured data quality and managerial routines.
  • Assess welfare/trade‑offs from formalisation induced by digitalisation—e.g., when formalisation raises tax burdens without commensurate access to finance or markets.
  • Explore dynamic feedbacks: whether access to finance enabled by digital traces leads to persistent productivity gains or only short‑term working capital smoothing.

In short: AI and digital tools matter for SME performance and credit access in Africa, but their economic impact is conditional and institutionally mediated. Empirical work should focus on interactions between data quality, organisational routines, financing terms and institutional trust.