Inside the Togo AI Lab: leveraging AI to guide public policy

Around twenty engineers, data scientists, and analysts are hunched over their screens in a large office. Wearing headsets, they process satellite data, build dashboards, and train predictive AI models. "It’s a very dynamic environment with a lot of lively discussion," says Kokouvi Mawudem Koledzi, Manager of the Togo AI Lab. "We're constantly running tests and discovering new things."

The AI Lab is led by the Ministry of Public Service Efficiency and Digital Transformation, in collaboration with the University of Berkeley. It began with the pilot Novissi project in 2020, which used satellite imagery and mobile phone data to identify the most at-risk areas during the Covid-19 pandemic. Since then, the Ministry has built on this initiative by creating a hub where experts collect, process, and analyze data to inform public policy decisions.

"The lack of reliable data was hampering the government's capacity to effectively plan its public spending. The Togo AI Lab has addressed this issue by centralizing and structuring data from wide-ranging sources to provide a more accurate and up-to-date overview of the population's needs."

The team of around forty people, made up of data engineers, data scientists, analysts, geospatial data specialists, and data governance experts, works on the entire data lifecycle. Data is collected, cleaned, structured, and then converted into visual representations for decision-makers in government ministries.

FID is funding this project through its Impact Evidence and Public Policy grant. This support has enabled the Lab to recruit professionals, create open-source training modules, and develop predictive AI models, starting with the agricultural sector. "We are working on models that can forecast annual production yields of maize and sorghum, which will help guide public decisions on how to support farmers."

The Togo AI Lab also uses satellite imaging data to monitor changes in the agricultural landscape, which is then cross-referenced against information from the field and national government databases. The aim is to produce analyses that can be used by government ministries. For example, these tools can target the best locations for healthcare facilities or provide insights into agricultural trends and patterns.

The Lab recently helped select a site for a secondary-level hospital in northern Togo. The team analyzed data on population density, existing facilities, travel distances, and poverty levels to identify the area with the greatest need.

The Lab also assesses satellite imagery to track environmental phenomena in Togo, such as coastal erosion. This data revealed that the coastline has been retreating for several decades, and compared local dynamics between the most eroded areas and sedimentary zones. It is now working on a model to predict how this erosion will develop over the coming years (source : "Coastal Erosion Dynamics in Togo from 1984 to 2025").

According to Kokouvi Mawudem Koledzi, developing this type of laboratory is not easy, particularly in terms of data protection: "The data processed is sensitive and protected by strict protocols, as well as regular security testing. We also prioritize the use of customized open-source software." The Lab team also had to think carefully about how to encourage ministries to use these tools. For each dashboard delivered, a "key user" is designated within the relevant ministry. Training sessions are organized so that decision-makers, and not just technicians, know how to interpret and use these data visualizations.

The Lab does more than simply produce analyses and decision-making tools. Its role is also to embed a data-driven culture across the Togolese administration. In just over a year of operation, its teams have already trained more than 1,500 people in data technology and supported twenty ministries in using these new tools. The AI Lab aims to lastingly change practices and provide decision-makers with a wealth of evidence, indicators, and objective analyses, that will help them develop, guide, and evaluate public policy.