Our process began with grounding the project in scientific reality. In collaboration with METER Group, we installed environmental sensors in forests across Czechia, Brazil, and Korea, collecting six months of continuous data on local climate and a wide range of environmental parameters. These readings were then cross-referenced with open-source Copernicus datasets to reveal cycles, anomalies, and environmental stress patterns. Working alongside environmental scientists, we transformed these raw measurements into meaningful insights on each tree's health, resilience, and adaptation to changing climates.

To translate these findings into narrative form, we developed a custom Large Language Model (LLM) trained on the environmental data and enriched with academic research, ecological literature, and historical records. Each tree was given a persona and a distinct narrative voice, carefully crafted from its local ecosystem, physical characteristics, and contextual surroundings — allowing data to evolve into expression.

The resulting "tree reports" were published through a dedicated digital platform that documented every stage of the process and offered transparent access to the datasets and sources used by the LLM in generating each article. By merging scientific precision with emotional storytelling, Tree Correspondents transformed environmental data into lived experience. We didn't give trees a voice — we created a way for people to finally listen.