PlanAI — Greater Cambridge's AI Tool Cuts Planning-Consultation Analysis From 18.5 Hours to 16 Minutes
United Kingdom
Greater Cambridge Shared Planning and the University of Liverpool built PlanAI, a large language model trained on 15 years of …
United States of America · Bowling Green · See the United States of America profile
Evidence: Descriptive / self-reported Top 47% 60/100 · Ask Evidence Copilot about this practice
Bowling Green, Kentucky used Google Jigsaw's Sensemaker AI to analyse input from ~8,000 residents (about 10% of the population) for its BG2050 city plan, surfacing 2,370 ideas with 80%+ cross-partisan agreement and saving an est. 28 staff work-days vs. manual synthesis.
Bowling Green and Warren County, Kentucky ran a month-long online public-input campaign, 'What Could BG Be?', to shape BG2050, a 25-year comprehensive city plan, built with Google's technology incubator Jigsaw and local firm Innovation Engine.
The campaign used Jigsaw's Sensemaker, an AI tool that categorises large volumes of open-ended public comment into topics and measures where agreement and disagreement lie across the respondent population. About 8,000 residents — roughly 10% of the city's population — submitted close to 4,000 distinct ideas and weighed in on each other's submissions over a million times combined.
Sensemaker distilled the input into 2,370 discrete ideas that at least 80% of respondents agreed with, cutting across the usual partisan and demographic divides on topics such as parks, housing, mental health and transportation. City and county staff estimated the tool saved an average of 28 work-days of data analysis, synthesis, visualisation and report-writing compared with a manual process.
Officials say findings have already fed into zoning-reform discussions and plan to repeat the exercise annually; this was reported as the first large-scale, real-world proof of concept for Sensemaker. The process still relies on self-reported staff time-savings estimates rather than an independent audit, and there is limited public detail on how AI-clustered ideas were checked against the raw comments before being adopted into planning decisions.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
United Kingdom
Greater Cambridge Shared Planning and the University of Liverpool built PlanAI, a large language model trained on 15 years of …
United Kingdom
University of Liverpool and Greater Cambridge Shared Planning built PlanAI to summarise consultation responses. On 3 trial consultations it summarised …
Germany
Hamburg's DIPAS platform uses fine-tuned LLaMA and RoBERTa models to automatically cluster and summarise thousands of citizen comments on urban-planning …
Japan
GovTech Tokyo used AI 'Broad Listening' and the open-source Talk to the City tool to cluster ~27,000-30,000 public comments into …
Open full copilot Grounded in cited practices — always check the sources.