Uruguay’s partnership with Microsoft has produced eight artificial intelligence projects for public bodies, placing state services and public data in a programme built alongside a corporate partner. Microsoft and the state-backed Technological Laboratory of Uruguay (LATU) presented the results on Monday, 5 October, at LATU’s Innovation Park in Montevideo.
LATU president Lucila Arboleya set a target of 16 projects a year, and the tally has reached eight. “When we opened this laboratory, we set out to have 16 projects a year, and now, six months later, we have reached eight,” she told the Montevideo daily la diaria. Three projects are finished; five should finish in October. Neither Microsoft nor LATU named the three that are complete.
Public services, corporate-built tools
The lab received 47 applications from 32 public bodies and approved 12 projects. It trained 120 people and granted US$200,000 in cloud credits. Microsoft engineers work with staff from each institution to build the tools. Microsoft also matches every dollar Uruguay invests with credits for its Azure cloud, said Elianne Elbaum, who runs the lab for Microsoft in Uruguay.
Six projects target state services. The General Registry Directorate of the Ministry of Education and Culture is developing automated data extraction from electronic documents to speed up filings. Ceibal, the state education technology programme, built a chat assistant for its internal wiki, which holds more than 2,000 documents. The state power utility UTE is automating the drafting of tender documents.
The Hospital de Clínicas, a university hospital in Montevideo, is digitizing handwritten kidney dialysis patient records. The Tribunal de Cuentas, the state audit court, is getting a legal assistant to help draft reports. Agesic, the e-government agency under the presidency, is building an AI agent to help people find the correct online procedure. No launch date has been announced for the agent.
Research accounts for two projects. The public Technological University of Uruguay (UTEC) is using drone images to spot palm trees that may face attack by the red palm weevil, a destructive beetle. Another project, run with LATU, analyzes satellite images to tell fields of soy, sorghum and maize apart.
Who controls the data—and the bill?
Each institution remains responsible for its data and for putting the tools to work. Arboleya told la diaria that each body signed a confidentiality agreement with Microsoft and that the Uruguayan state retains control of the data. She said early results show shorter processing times and fewer errors, but offered no figures.
The programme has about US$400,000 a year for three years, or about US$25,000 per project, Arboleya told Ámbito. Cloud credits have a limit, and the partners haven’t said who will pay for cloud computing once they run out. They also haven’t provided figures for time or money saved, or said how many prototypes will enter regular use. Arboleya said the institutions, not the lab, put the pilots into practice.
The industry ministry, LATU, innovation agency Uruguay Innova, state telecom company Antel, Agesic, research agency ANII and development agency ANDE also take part in the partnership. Arboleya said requests for complex equipment can take two years to result in delivery. “Any day you can cut from that process is extremely important,” she told Ámbito.
President Yamandú Orsi of the centre-left Broad Front coalition presented Uruguay as “a trusted regional hub” to develop, test and scale AI at an artificial intelligence dialogue hosted by the Inter-American Development Bank in New York on 22 September. His office said the remarks formed part of a wider plan. The account describes government agencies, public institutions and Microsoft as the programme’s actors; it reports no community-led or mutual-aid response.
A third group is due to begin in November with ANDE, ANII, Antel and the postal service Correos. Microsoft said four more projects will start in February 2027; la diaria reported March. The lab hasn’t explained how it will reach 16 projects in its first year. Public institutions may own the data and carry responsibility for implementation, but the project’s ongoing costs and routine use remain unanswered.