Geo-Llama, the first LLM for Geographic Data

Short description of the innovation study

 Geospatial data remains an underutilised source of knowledge. One reason for this is that working with geodata is complex, time-consuming, and requires many manual steps. Additionally, geodata is highly dependent on external data sources – e.g. infrastructure or weather data – which companies often need to integrate from external sources. These factors limit insights into the world around us to specialised experts, such as geoscientists or data scientists. As part of the FFplus innovation study, DataMonkey will train its own AI model for geodata. To achieve this, the team will combine both proprietary datasets and open-source LLMs. The resulting model (W1) will enable users to query geodata, e.g. from OpenStreetMap, via the DataMonkey platform. This will facilitate fast geospatial insights for all types of users, regardless of their expertise in data or IT.

Organisations involved: 

End User: DataMonkey