Subscribe to Our Newsletter

Success! Now Check Your Email

To complete Subscribe, click the confirmation link in your inbox. If it doesn’t arrive within 3 minutes, check your spam folder.

Ok, Thanks

UK transport department uses Google Cloud AI to cut consultation analysis from months to hours

The tool has achieved up to 90% accuracy processing free-text responses and could save the department up to £4 million a year

Defused News Writer profile image
by Defused News Writer
UK transport department uses Google Cloud AI to cut consultation analysis from months to hours

The Department for Transport (DfT) has deployed Google Cloud artificial intelligence tools to analyse its public consultations, replacing a manual review process that previously took months with a system capable of categorising responses in hours.

The department runs around 55 consultation exercises a year, which can generate more than 100,000 free-text responses, and has a stated commitment to publish consultation outcomes within 12 weeks.

To meet that target, the DfT's AI and Data Science Team built the Consultation Analysis Tool (CAT) in collaboration with Google Cloud and the Alan Turing Institute, running on Google's Vertex AI platform.

The system uses Gemini large language models to identify and categorise themes across large volumes of feedback, and has achieved up to 90% accuracy in evaluated testing.

Google Cloud said the tool could save the department up to £4 million annually and enable faster responses to citizens.

CAT has already been used to support analysis for the Integrated National Transport Strategy and a review of driving test booking rules.

The DfT has also built two further AI tools on Google Cloud infrastructure: a Connectivity Tool for urban planning work, built using Cloud Run, Cloud CDN and Firestore, and an AI Correspondence Drafter that combines Vertex AI Search for document retrieval with Gemini for drafting responses.

Across all three tools, the department said it retains human oversight through a human-in-the-loop model, under which outputs are checked for accuracy, fairness and bias before use, and final policy decisions remain with departmental experts.

The deployment reflects growing interest among UK government departments in using large language models to process high-volume, unstructured text, an area where manual approaches have historically created significant backlogs.

The recap

  • DfT deploys Consultation Analysis Tool built on Vertex AI.
  • Tool reaches up to 90% accuracy and saves £4 million.
  • Google Cloud said organisations can learn more on solutions page.
Defused News Writer profile image
by Defused News Writer

Explore stories