DHIS2 is deployed as the national health information system in more than 70 countries, covering health data for over 40% of the world's population. Underpinning these implementations is a vast knowledge ecosystem that has grown organically over decades: user documentation, online training, in-person academies, conferences, an online community of practice, videos, presentations and more.
AI is already impacting how DHIS2 users interact with this knowledge base, but there remain many unanswered questions about what this impact will look like and how it can be shaped and directed.
Exploring and addressing these gaps is what this project is about.
Example problems worth exploring
• Are agents quietly replacing real community? A verified analysis already shows DHIS2's new AI assistant cut technical support activity on the community support forum by roughly 40%. Is that healthy? It can provide faster answers with less duplicated effort but is it hollowing out the peer exchange that has long helped new country teams get unstuck and build relationships? What happens when there are harder questions the AI can't answer and the forum has been abandoned?
• When training can't be adapted, who gets left behind? DHIS2 training material is provided generically with the intention of it being localised to fit a country’s language, administrative environment and specific implementation. Currently this is a very large undertaking for country teams. Can AI be systematically utilised to provide localised training with lower demand on country teams to develop and maintain? When content can be translated increasingly accurately and quickly, what stand out as the most important considerations for localisation?
• Is DHIS2's knowledge base ready for the AI-agent era? As more implementers turn to AI agents to answer configuration questions, build custom apps and manage data flows, how legible are DHIS2's own docs and APIs to those tools today? What is the gap between 'human-readable' and ‘agent-ready’, can and should both audiences be served the same way?
The HISP network and DHIS2's community span dozens of countries, each with implementers, trainers, and users who live with these problems daily. Any of the questions above can be tested against real practice by talking to the people actually running DHIS2 on the ground, from Sri Lanka to Tanzania to Vietnam. These problems are surfacing and being tackled in vastly different contexts, providing a rich research base.