By The Weekly Vision Business Reporter
Artificial intelligence is no longer confined to pilot projects in Kenya’s boardrooms and research labs. It is beginning to appear in barbershops, small farms and clinics, yet the distance between experimentation and measurable commercial returns remains wide.
A growing number of micro, small and medium-sized enterprises (MSMEs) and agricultural operators are testing AI tools that convert routine data into actionable business insights. One Nairobi barbershop discovered it served roughly 180 loyal customers a month only after an AI platform began analysing its transaction records, an insight that subsequently guided its expansion plans. The same system has been integrated with M-PESA Business, allowing owners to transform raw mobile-money statements into structured dashboards without needing specialised data skills, a meaningful advantage for operators who cannot afford a dedicated analyst.
Early uptake has been brisk: one platform reported more than 600 MSME sign-ups in its first month and over 3,500 within three months of launch, figures that suggest genuine demand rather than a marketing-driven trial.
In agriculture, tools such as FarmSawa’s GFI Agri platform claim to serve more than 10,000 farmers across all 47 counties, offering disease-scanning capabilities with a reported 98 per cent accuracy and services available in three languages, addressing both the technical and accessibility barriers that have historically limited digital tools’ reach among smallholders.
The commercial case is strongest where results have been independently measured. A clinical study involving nearly 40,000 patient visits across fifteen clinics found that clinicians using an AI decision-support tool recorded a 16 per cent relative reduction in diagnostic errors and a 13 per cent reduction in treatment errors compared with those who did not, a result with direct implications for healthcare cost and liability exposure, not just clinical quality.
Larger corporates have reported tangible gains as well. Safaricom has cited AI-driven customer segmentation as one factor behind sustained growth in its mature voice business, an example of AI supporting incremental revenue in a sector often assumed to have plateaued.
Despite these examples, the broader picture remains one of uneven progress. A 2025 survey of Kenyan business professionals found that 96 per cent of organisations had begun an AI journey, yet only 35.2 per cent had achieved widespread or advanced implementation, a gap that points to stalled pilots rather than outright failure.
Investment levels help explain the shortfall. Across Africa, organisations invest an average of just 2 per cent of revenue in AI, compared with 5 per cent among global leaders, and only a minority report that their investment is sufficient to meet their ambitions. For smaller Kenyan firms in particular, the gap between running pilots and embedding tools that demonstrably lift revenue, cut costs or improve yields remains substantial, constrained by data infrastructure, technical talent and access to patient capital.
Coverage of AI in Kenya has tended to oscillate between technological hype and isolated success stories, leaving investors, lenders and policymakers with a thin evidence base for where the returns are real and repeatable. The examples that have produced measurable outcomes- the barbershop’s data-driven expansion, the clinics’ error-reduction figures, the agri-platform’s farmer reach- share a common feature: they solve a specific, quantifiable operational problem rather than promising broad transformation.
For businesses and investors assessing where to commit capital, the near-term opportunity appears concentrated in tools that plug directly into existing transaction or operational data, such as mobile-money records, patient visits or crop-health images, rather than in open-ended AI adoption. Cost, connectivity, digital literacy and data quality remain the practical barriers separating the next wave of MSMEs and smallholder farmers from the same kind of measurable returns already visible in these early cases.
