UK Department for Transport accelerates public policy insights with Google Cloud AI
For the UK’s Department for Transport (DfT), which manages approximately 55 public consultations each year, analysing citizen feedback is a monumental task. These consultations often generate more than 100,000 free-text responses, creating a staggering data challenge that historically required teams to manually review and classify themes over several months. To help meet the principle to publish consultation responses within 12 weeks and unlock insights faster, the DfT needed a new approach.
To solve this, the DfT’s AI and Data Science Team collaborated with Google Cloud and the Alan Turing Institute to build the Consultation Analysis Tool (CAT). Built on Google’s Vertex AI platform, the CAT system uses Gemini models to identify and categorise themes from massive volumes of public feedback in just a few hours — a process that previously often took months. The evaluated solution has achieved up to 90% accuracy (various measures used) in its analysis, which when combined with a "human-in-the-loop" review process fully captures the previously manual review and classification outputs, enabling the government to respond to citizens faster while saving up to £4 million annually (DfT, 2025). For example, the CAT supported the analysis of public comments responses to the Integrated National Transport Strategy and improving driving test booking rules.
The department’s innovation extends beyond public consultations. Using Google Cloud services like Cloud Run, Cloud CDN and Firestore, DfT also developed a Connectivity Tool to help urban planners make more sustainable infrastructure decisions. Additionally, the AI Correspondence Drafter produces first drafts of responses to public inquiries by using Vertex AI Search to retrieve relevant policy information from secure internal databases (hybrid search) and Gemini for drafting. The potential for AI to transform transport efficiency is immense, but it must be done responsibly. The DfT is utilizing AI to process data at scale while keeping human judgment at the heart of every policy.
By using a "human-in-the-loop" model, they ensure AI outputs are checked for accuracy, fairness and bias, something DfT research with the public also identified as important. Google Cloud provides the processing power, but the vision and final decisions come from DfT policy experts. This transparent approach ensures that technology serves the public interest while contributing to a more efficient future for the nation’s transport system.
To learn more about how Google Cloud is helping public sector organizations accelerate their missions, visit our solutions page.
