The Thesis
There is a version of artificial intelligence in African finance that replaces bankers. Jean-Luc Konan, founder and CEO of the COFINA Group, is building a different one. In an interview published by Financial Afrik on July 26, Konan described AI as “an accelerator of discernement” — a technology that sharpens the judgment of loan officers rather than supplanting them. In markets where formal credit data is sparse, and where the difference between a viable SME and a bad loan depends on contextual knowledge no algorithm yet carries, that framing matters more than it sounds.
The Missing Middle COFINA Fills
COFINA was founded in 2013 with a specific mandate: finance the “missing middle.” This is a term of art in development finance for businesses that have outgrown microfinance but remain too informal for a commercial bank. An SME needing a 50 million FCFA working capital line in Conakry or Dakar cannot get a meaningful audience at most banks. COFINA was built for exactly that business.
Thirteen years later, the group operates in eight countries — Côte d’Ivoire, Guinea, Senegal, Gabon, Congo, Mali, Burkina Faso, and Togo — employs 2,000 people, and has financed more than 85,000 business projects. Several of those markets are among Africa’s most Muslim-majority economies: Senegal is approximately 95% Muslim, Mali and Guinea around 90%. For entrepreneurs in those countries, COFINA has become one of the few scaled lenders that actually shows up.
The tension COFINA has always navigated is the same one that constrains all mesofinance: creditworthiness is difficult to assess when borrowers lack formal accounts, registered payrolls, or audited financials. Decisions have relied heavily on relationship banking — loan officers who know the local market, understand informal cash flows, and can read a business’s viability beyond its paperwork.
The AI Bet
Konan’s argument is that AI does not disrupt this dynamic so much as it amplifies it. “AI provides unprecedented analytical power, but it does not replace what makes a decision good: discernement,” he told Financial Afrik. The model COFINA is building uses AI to process larger datasets, identify patterns in repayment behavior, and flag risks that a loan officer might miss — while keeping a human in the final decision chair.
This is a more conservative approach than AI-first lending models, which automate credit decisions end-to-end. It is also, arguably, a more appropriate one for markets where a mis-calibrated algorithm can ripple through an already fragile SME ecosystem. COFINA’s existing partnership with Proparco — the private sector arm of France’s AFD development agency — targeting SMEs in Côte d’Ivoire and Senegal suggests a lender that does not take credit risk lightly.
The Headwinds
The argument for human-augmented AI is sound in principle. The implementation risks are real. African SME data is heterogeneous, often informal, and inconsistently collected across COFINA’s eight markets. A system trained primarily on Ivorian repayment data may perform poorly when deployed in Guinea or Congo, where economic conditions and business culture differ materially. COFINA has not published technical specifications for its AI approach, which makes it difficult to assess how rigorously these cross-market calibration challenges are being addressed.
There is also the question of whether “discernement accelerator” is a useful product frame or an aspirational one. Many institutions across Africa have described AI as a tool to improve credit decisions without specifying what those systems actually do — or what happens when they get it wrong.
What to Watch
Konan’s 2030 vision is for COFINA to be “the natural partner of businesses growing in Africa.” That vision depends on whether its AI-augmented model actually improves credit outcomes at scale — not just whether it processes applications faster. The metric to follow is not loan volume. It is non-performing loan rates, disaggregated by market. If AI is genuinely improving discernement, the signal will appear in repayment quality. Does COFINA’s expanded AI use narrow the gap between how it assesses creditworthiness and how that creditworthiness actually performs? That answer is still outstanding.