Tuberculosis is a priority for the Unified Health System (SUS) in Brazil and for the Ministry of Health in Mozambique, countries with high TB burden, where treatment faces linguistic barriers, among other aspects. Indigenous peoples in Brazil are especially susceptible to TB, and there are more than 160 indigenous languages in the country. In Mozambique, only 40% of the population speak Portuguese, the country’s official language. Considering this dilemma, the project seeks to implement an information tool for community health agents in Guarani-Kaiowá communities in Brazil and in Nicuadala communities in Mozambique, by studying the practical feasibility of AI, developing a model for translating Portuguese to Guarani-Kaiowá and Portuguese to Echuwabo. The communities will be active participants in validating the translation and adapting the information to the local cultures. The project thus seeks to produce a culturally and linguistically adapted information source on TB, accessible on smartphones, to assist indigenous health agents, community health agents (in the SUS) in Brazil, and polyvalent agents in Mozambique in their work with follow-up of active TB searches and counseling. The project is expected to overcome the linguistic barriers using LLMs, as well as to standardize the process of building the translation model for implementation in other contexts. This is an innovative idea in a South-South scientific collaboration that proposes the use of AI considering the preservation of traditional knowledge and its integration in the biomedical system.