This report presents an AI learning and stocktaking exercise jointly commissioned and conducted by the evaluation offices of the Adaptation Fund, the Climate Investment Funds, the Green Climate Fund, and the Global Environment Facility. Building on a previous study carried out by evaluators on forestry work across the funds, this study used AI to explore existing evaluation material on the four funds’ forestry and sustainable forest management interventions. It demonstrates how AI can be applied to extract and organise findings from a large and varied body of documents.

As a first attempt of its kind, the exercise is primarily an exploration of method. Its value lies in showing what AI can do (and can’t do) with a large and diverse evidence base, and in laying groundwork for future evaluative work. It was not designed to answer a defined evaluative question, or to synthesize the recommendations of the underlying evaluations. Nor does it assess or compare the funds’ performance. 

The evidence brought together here opens promising avenues for future evaluative work. Sustainability stands out as an area worth deeper examination, particularly what distinguishes interventions whose results are sustained. Likely factors include financing, institutional capacity, country ownership, exit strategies, and mechanisms for replication and scaling. Recurring factors such as implementation timeframes, stakeholder engagement, coordination and project design could also be synthesized systematically, together with the recommendations of the underlying evaluations. Such work could help answer a broader question of shared interest: when, why and under what conditions do climate-financed forestry interventions achieve and sustain results?