AdaptBase

AdaptBase is a research initiative that leverages large language models (LLMs) to organize the world’s knowledge on effective urban climate adaptation. As cities worldwide deploy diverse portfolios of adaptation solutions, the evidence of what works—including costs, actions, and measured outcomes—remains locked in thousands of planning PDFs, where it can’t readily inform planning and financing decisions. AdaptBase uses AI agents guided by a custom open-source ontology, to systematically extract this hidden knowledge and assemble it into a queryable knowledge graph that connects solutions, hazards, jurisdictions, and results across places and projects.

The project has already produced a searchable library of 1,450+ adaptation plans from 600+ cities, a deep-research portal that automatically generates urban adaptation case studies, and the underlying ontology that lets AI interpret these plans consistently. Upcoming work focuses on growing the knowledge graph and developing applied tools, including an AI chatbot to help city planners discover and evaluate solutions used elsewhere, and a benchmarking dashboard for investors and analysts to compare costs, impacts, and outcomes across jurisdictions for resilience underwriting. The team, led by Dr. Anthony Townsend, is seeking pilot partners among cities, investors, and analysts, as well as research collaborators with novel approaches to applied climate adaptation scholarship.