How can artificial intelligence (AI) and digital public infrastructure (DPI) work together to improve public services in South Africa? And what would it take to move promising innovations across institutional boundaries and embed them within government systems?
These questions were central to the AI and DPI Partner Convening, hosted by the Gates Foundation in Cape Town on 21 and 22 October 2026. The Policy Innovation Lab at Stellenbosch University worked alongside the Foundation to deliver the event, with Prof Willem Fourie, Chair of Policy Innovation and lead of the Lab, as programme director. The Policy Innovation Lab contributed to panel discussions, facilitated thematic working sessions and supported the coordination of the convening.
The event brought together government representatives, researchers and organisations working on AI and DPI initiatives to exchange implementation experiences, identify opportunities for collaboration and explore how innovations could be adapted and scaled across different institutional settings. Participants included representatives from the Digital Services Unit (DSU) in the Presidency, alongside organisations such as Genesis Analytics, Cenfri, Jembi Health Systems, mDoc, Sebenza.taxi, Health Systems Trust, EMGuidance, Wits-VIDA, Glassbox AI Advisory and the University of Cape Town.
While digital public infrastructure provides the foundational systems for identity verification, payments, and data exchange, AI offers opportunities to make services more accessible, responsive, and efficient. Realising these opportunities, however, requires more than technological innovation. It also depends on effective institutional coordination, appropriate governance arrangements, public-sector capabilities and sustainable implementation models. These considerations shaped the discussions over the two days, as participants explored both the opportunities presented by AI and DPI and the practical challenges of implementing them within government.
These organisations work on different aspects of a shared challenge: improving the accessibility and effectiveness of public services. The consequences of fragmented government systems are practical. When departments cannot exchange information, people seeking social grants or health care may have to provide the same information repeatedly, costing time, transport, and lost income. Officials must navigate disconnected systems, while implementing partners devote resources to overcoming institutional and technical barriers.
The first day focussed on introducing what AI and DPI initiatives each of the partners work on and what lessons could be shared with other participants. Zameer Brey of Glassbox AI Advisory and Nanjira Sambuli of the Gates Foundation explored the complementary roles of DPI and AI. For example, an AI assistant may help someone complete an application, but verifying their identity, accessing relevant records and processing payments still depend on functioning government systems. A more accessible interface cannot, on its own, overcome fragmented public infrastructure.
Dr Manicom from the Lab, along with Cenfri and Genesis Analytics, participated in a panel discussion of DPI in South Africa with a focus on how DPI enables access to services. The discussion focussed on building trust in the central DPI technologies. The Lab’s contributions focussed on how coordination and alignment across departments is essential to providing fair and equitable access to multiple services, while good governance is required to establish trust. Legal questions resolve who can verify citizens for various services and what happens if the credentials fail. Finally, the strategic development of long-term data sovereignty will determine the sustainability and stability of services.
Later, Dr Manicom facilitated a discussion with members from the Policy and Research Services Branch of the South African Presidency and Genesis Analytics on the operational realities of DPI in South Africa. After describing progress and the current state on matters of DPI governance and deployment, the conversation developed into discussions around incentives. Key questions asked were how to incentivise both citizens and frontline workers to adopt these tools, and how to incentivise departments to share their data so that citizens can access their services in a platform they do not control (as opposed to maintaining the status quo). A key insight is that by generating momentum with a few services that work well, citizens will begin to expect the same quality and level of services from the staff they engage with and across other government services.
The second day focused on the practical conditions needed to implement, adapt, and sustain AI and DPI initiatives. Mixed groups of government officials, implementing partners and Foundation representatives explored four themes: rules and institutions, systems and adaptation, people and capability, and funding and durability. The discussions emphasised the importance of practical, workable solutions, a sentiment captured by one participant’s observation: “Something working is way better than perfect architecture.”
Five points stood out:

1. Coordination is the main constraint. Systems often work well inside one institution and stall when they depend on another department or sphere of government.
2. Data sharing needs clearer rules and more consistent agreements. Agreements are often rewritten for every partnership; common templates and clear consent rules would save time and build trust.
3. Government needs the capability to run systems, not only buy them. Participants discussed building a core of skills inside government, working alongside private partners and universities.
4. Plan for life after the grant from the start. Teams should identify early what evidence, financing, and institutional capacity government would need to adopt and sustain a system.
5. Sustainability does not always mean handover. Some systems may be best run through long-term partnerships, provided ownership and accountability are clear.

For South Africa, the building blocks are taking shape, from large-scale social grant payments to a growing set of digital government services and capable implementing partners. The challenge is ensuring that these systems work effectively together. The next phase of digital transformation depends not only on developing new technologies, but also on embedding them within government institutions: clear ownership, skilled people inside government, and sustainable financing arrangements.
These considerations are closely aligned with the work of the Policy Innovation Lab. Working with the Presidency, the Lab builds AI capabilities among policymakers, contributes to the evidence base for regulating DPI, and supports policy advice and dialogue at the centre of government, so that useful tools and practices become part of how government works. The convening reinforced the importance of collaboration between government, researchers and implementing partners in addressing these challenges. The Lab looks forward to continuing this work with the Presidency, the Gates Foundation and the partners who took part.