DVLA’s AI phone system halves call navigation time

TL;DR:

• The DVLA has replaced its keypad phone menus with an AI voice system that lets callers describe enquiries in their own words.
• Average call-navigation time has fallen from around three minutes to roughly 90 seconds.
• About 20,000 calls a month are now routed automatically to the right adviser.

The DVLA has moved a slice of its vast contact-centre operation onto AI, deploying a natural-language voice system that interprets what callers want rather than making them work through layered touchtone menus. Handling roughly 900,000 calls a month, the agency had described its old system as a “clunky customer journey” — and the early figures suggest the change has measurably sped things up.

A working public-sector case study

The new interactive voice response system, built on Google AI technology, analyses a caller’s intent and then either answers automatically, sends a gov.uk link by text, offers self-service or connects the person to the right adviser. According to the agency’s published transparency records, average navigation time has dropped from about three minutes to 90 seconds, around 20,000 calls a month are now routed automatically, and the DVLA has gained clearer visibility into what people are actually calling about.

What makes the rollout notable is that it is live and measurable, rather than a pilot or a promise. UK public-sector AI announcements often describe intentions; this one comes with before-and-after numbers attached. The work also feeds into wider cross-government efforts on conversational technology alongside the Government Digital Service and the Department for Science, Innovation and Technology — a sign the DVLA’s experience is meant to inform other departments, not stay siloed.

Looking forward

The DVLA says it will keep exploring how AI and automation can support customer service across government. The harder questions sit beyond call-routing: how the system handles vulnerable callers, edge cases and complaints, and whether efficiency gains translate into genuinely better service rather than simply fewer humans to reach. For now, it stands as one of the cleaner examples of UK government AI delivering a concrete operational result — a useful counterweight to projects where the technology was deployed despite known flaws.