The Central Bank of Kenya has identified nearly 300 artificial‑intelligence use cases spread over 14 of its departments and intends to implement a number of them during 2026, according to the bank’s Head of Research.

Four tiers of AI deployment are being pursued: routine or intelligent automation, machine‑learning models, generative‑AI tools and, later, agentic AI that can perform tasks without direct human input.

Early projects focus on automating repetitive tasks in banking, payments and research, while a pilot chatbot is being tested in the Financial Markets Department.

Machine‑learning techniques are being added to inflation‑forecasting models to help the bank adjust for volatile periods, such as the energy crisis that began in February 2026.

The new approach lets models learn from past forecasting errors, giving greater weight to second‑round effects that were previously underestimated.

Human judgment remains central; economists will interpret AI‑generated outputs and retain responsibility for policy decisions.

To bridge the lag in official statistics, the bank is exploring now‑casting methods that combine high‑frequency indicators such as satellite‑derived night‑light data, electricity consumption and mobile‑payment volumes.

Researchers note that mobile‑payment values in Kenya have historically tracked at roughly 2.5 times GDP, offering a potential proxy for real‑time economic activity between quarterly releases.

Web‑scraping of online supermarket prices is being tested to supplement existing surveys and provide earlier signals of consumer‑price pressures.

The Head of Research emphasized case‑by‑case risk assessments and safeguards against inadvertent data exposure as key elements of the bank’s AI governance framework.