The Policy Innovation Lab Catalogue of AI-based tools
We have recently published the June 2026 update of the Policy Innovation Lab’s Catalogue of AI Tools in Government. This catalogue contains a wide range of use cases for AI in public sectors from around the world. It is heavily reliant on the Public Sector Tech Watch catalogue of use cases of emerging technology (including quantum and blockchain) deployed in the European Union. We filter their catalogue for AI use cases relevant to South Africa and extend it to include AI use cases from across Africa and other non-European countries. You can read more about the catalogue filtering methodology and our original findings in our 2024 report. We update the catalogue twice a year. After June’s update, it contains 1,033 real-world examples from across the globe.
Examples in areas of application mentioned by the President
The catalogue is designed to help identify the possibilities of AI in government through user-friendly search and filtering functions (the raw, comprehensive data may also be downloaded). At the inaugural Google Cloud Summit on 1 July 2026, President Cyril Ramaphosa highlighted cloud and AI’s role in the national and continental agenda. He argued that “Africa intends not merely to participate in that future. We intend to help shape it.”
The catalogue contains numerous examples that speak directly to the areas highlighted by the President, who envisioned AI solutions deployed “for disease management and prevention, to manage the national energy grid, by farmers to predict weather patterns, and by scientists to guide our national climate response” alongside cloud-delivered educational content. In the bar chart below, we show the number of examples identified using simple search terms derived from this quote. Some of these examples are already being tested or used in South Africa and neighbouring countries, and we discuss examples of particular interest or relevance.

The President’s vision of educational content delivered through the cloud highlights a central challenge in South Africa: how to use technology to support learning without assuming ideal classroom conditions. South Africa already has a useful proof of concept. Stellenbosch University and Trackosaurus have developed voice AI for play-based maths and language learning. The tool enables spoken interaction for young children in classrooms where reading literacy cannot be assumed and where instruction happens across multiple languages.
Outside the classroom, AI is changing how government listens to citizens. Barcelona’s Decidim is an open-source digital participation platform where residents debate, propose and vote on municipal decisions. In Bowling Green, Kentucky, a month-long digital town hall builds on similar technology that engaged nearly 8,000 residents, roughly a tenth of the city, and surfaced proposals with support across otherwise divided groups. Because tools like Decidim are open source, they offer municipalities an affordable route to structured public participation at scale.
Health also shows how much public sector AI extends beyond the language models most people picture.  Many of the most consequential health tools instead support triage, image-based diagnosis, prognosis and personalised care, especially where access to specialists, facilities and real-time information remains uneven.
Ukraine has deployed AI-assisted analysis of chest X-rays, including mobile screening units, to detect tuberculosis and other lung pathologies with reported accuracy above 90%. Given South Africa’s TB burden, this is among the most directly transferable examples in the catalogue. Greece, in turn, is adding AI -supported triage and care coordination to its national health hotline, helping route around a million citizens to the right level of care.
Closer to South African health-system realities, Piedmont, Italy, is using AI to extend healthcare into its mountainous, underserved areas. Telemedicine cabins let patients undergo exams with a trained operator present while doctors connect remotely, with AI assisting diagnostics, patient monitoring and workflow. The approach speaks directly to South Africa’s own rural health-access gaps, where distance to specialists and facilities remains a major barrier to care.
For South Africa, AI-enabled infrastructure management is not a distant use case. Years of load-shedding have shown how important it is to detect faults earlier, schedule maintenance better and use infrastructure data more effectively.
Norway’s PHM Hydro initiative combines physics-based modelling with sensor data analytics to detect faults and support maintenance across hydropower plants. The project focuses on critical equipment such as turbines and generators, where continuous monitoring can help detect emerging faults, avoid unplanned shutdowns, improve maintenance planning and keep hydropower assets operating efficiently. South Africa’s Ingula Pumped Storage Scheme has a capacity of 1,332 MW, larger than any of Norway’s hydro plants, in which similar AI systems could support predictive maintenance, reduce downtime, optimise pump-turbine performance, and reduce energy and water waste in one of the country’s most important grid-balancing facilities.
France’s national rail operator, SNCF, monitors more than 1,100 transients in real time, analysing thousands of variables per train so that faults can be fixed before they cause failures. Closer to home, Cape Town’s Digital Water Control Room uses telemetry and predictive analytics to reduce false alarms and speed up leak response across the city’s water network. These examples show how AI can support infrastructure reliability before failures become service-delivery crises.
Agriculture is another area where public-sector AI must be designed around real operating conditions, including smallholder farming, basic phones and uneven connectivity. The Hi-SAAI Family Farmers Chatbot provides AI-driven advice on weather, inputs and financial planning to South African family farmers. Austria’s agrifoodTEF takes a complementary approach, giving farmers and agritech developers a national testbed to trial AI and robotics for precision agriculture and food-quality monitoring before committing to full-scale rollout. PlantVillage Nuru offers another relevant African example, using computer vision to support crop disease diagnostics at scale. Together, these tools show that agricultural AI need not be limited to large commercial farms. It can also strengthen advisory services for farmers operating in resource-constrained settings.
The President also linked AI to the national climate response. Climate adaptation increasingly depends on the ability to integrate environmental, infrastructure, and satellite data in ways that enable faster planning. South Africa’s flood risk detection tool, developed by the South African National Space Agency, provides spatial flood inundation scenarios alongside settlement data to support evidence-based policymaking. Internationally, Lisbon’s AI solar installation mapping supports renewable energy planning, while an earth observation initiative in Niger used geospatial AI to identify yield-positive farming practices in dryland conditions. These examples point to the role AI can play in making climate risk more visible and actionable for public institutions.
What the April 2026 update reveals
Beyond the five areas highlighted in the President’s speech, the 129 new entries added since December 2025 reveal several broader shifts in government AI use.
Generative AI continues to move into governments at scale. Chatbots, virtual assistants and voicebots account for 49 of the new entries, while 27 explicitly mention generative AI, LLMs or GPT. Governments are also increasingly procuring rather than building. Eight new entries are Microsoft Copilot rollouts in local government, indicating a preference for adapting tools from trusted partners that integrate easily into known workflows.
Internal use cases are expanding quickly, particularly in support processes, service personalisation and administrative efficiency. Tools that support AI governance are also emerging as a category of their own, with tools such as Z-Inspection and IEEE CertifAIEd used to assess or audit AI systems for ethics, transparency and accountability.
Finally, procurement integrity is becoming a visible use case. Albania’s Diella offers a high-profile example, while Ukraine’s national whistleblower portal presents a more conventional model for AI-supported corruption reporting.
AI usage
This article was drafted using human expertise supported by AI-assisted writing tools.