Kenya Builds Foundation For AI Agents

Kenya’s next AI project may be less visible than a chatbot but more important for how citizens experience government. The country is working on the shared digital foundations that would allow public institutions to exchange data and services securely before autonomous AI agents are allowed to act across them.

At WSO2Con Africa in Nairobi, ICT Principal Secretary John Tanui said Kenya’s next frontier is the backend. The objective is to connect fragmented government systems through secure application programming interfaces and an enterprise platform that can coordinate services across institutions. ITWeb Africa reported that Tanui described the shift as a move from software that waits for instructions to systems that can pursue authorised outcomes.

That distinction matters. An AI agent cannot deliver a reliable public service if the underlying systems are isolated, the data is incomplete or the permissions are unclear. Giving an agent more autonomy before fixing those conditions would simply make existing administrative fragmentation faster and harder to audit.

The official WSO2Con programme reinforces the technical logic behind the announcement. Its sessions focused on preparing APIs for agentic consumption, securing API traffic, masking personal information, protecting data sovereignty and managing the lifecycle of agents after deployment. Those are not optional features. They are the control layer that determines what an agent is allowed to see, change and explain.

Kenya’s eCitizen platform has already shown the value of bringing public services into a more unified digital experience. The next phase will test whether the state can make the underlying architecture as coherent as the front end. Citizens should not need to understand which ministry owns a database in order to complete a service request.

There is also a governance challenge. Agentic systems can make decisions, call tools and trigger workflows, so government must maintain audit trails, escalation routes and clear human responsibility. Zero-trust security, role-based access and human review should be built into the design rather than added after an incident.

Kenya’s plan is therefore best understood as an infrastructure and accountability project, not simply an AI launch. Its success will depend on whether agencies can share trusted data, whether local teams can maintain the platform and whether the resulting services are simpler for citizens to use.

Sources: ITWeb Africa, WSO2Con Africa agenda, and Kenya Ministry of ICT releases.