AdaptBase

Knowledge graph of urban climate adaptation plans from the Urban Tech Hub at Cornell Tech; AI agents read city plans and extract actions, hazards, outcomes and costs, each traced to its source sentence.

The Urban Tech Hub at Cornell Tech (Jacobs Technion-Cornell Institute, New York) runs it as a research project led by Anthony Townsend.³ The aim is to organise what cities have written about adapting to climate hazards.^1,3 Agents find adaptation plans and code them as structured facts against a team-built ontology, a defined vocabulary of 21 entity types and more than 60 relationship types.^1,3 Each fact links back to the plan, page and sentence it came from.¹ Heat-action codings from Cornell's Heat Plan Database are folded in.^1,2

A practitioner can ask which cities use cool pavements, or how flood defences are being paid for.² Answers come back as linked nodes: places, plans, actions, solutions, hazards, outcomes, stakeholders and financing.² In October 2026 the site reported 115 plans from 86 cities read in full, 13,204 actions and 238,142 evidence records.¹ The Urban Tech Hub's May 2026 project page gives a wider plan library of more than 1,450 plans from 600 or more cities.³ The two counts describe different stages of the pipeline.

The hosted site has a search interface, an ontology viewer, a plan archive of the original PDFs, and an API and MCP endpoint for other tools.¹ (MCP, the Model Context Protocol, is a standard way for AI assistants to call outside tools.) The plan-archive pipeline is public on GitHub.⁴ It uses Supabase, OpenAI embeddings and an Astro static site.⁴ No licence is given for the data or the main project.⁴

AdaptBase is in alpha, and the site warns of rough edges and thin areas of the graph.¹