The UK Department for Work and Pensions (DWP) is investing up to £23.4 million in a new conversational AI system designed to streamline its benefits call centers. The project aims to implement a “natural language call steering system” that allows citizens to speak naturally when they call, rather than navigating complex touch-tone menus. The AI will identify the caller’s intent and either provide automated self-service options or route the person to the most appropriate human agent. This move comes as the department struggles with a significant increase in claimants, which rose by 11.8% between 2019 and 2023.
A key driver for this investment is the massive volume of wasted call time currently experienced by the department. A National Audit Office report found that over 31 million call minutes could have been avoided in a single year if callers had been directed more efficiently. By using AI to “deflect” routine inquiries—such as checking a payment date—human workers can focus on more complex cases that require manual intervention. The project is expected to officially begin in July 2026 and could run until 2032 if optional extensions are exercised.
To ensure public trust and security, the DWP has mandated that the AI solution must be hosted on a dedicated cloud within the United Kingdom. It must also comply with strict data protection laws, including GDPR and the Data Protection Act, as well as the government’s own security policy frameworks. This localized approach is intended to mitigate risks associated with sensitive personal data being processed by international AI platforms. The government is currently accepting bids for the contract, with a winner expected to be announced by June 1.
This initiative is part of a broader push by the UK government to modernize its technological infrastructure, though officials have recently admitted that a previous goal of securing all systems by 2030 may be out of reach. By focusing on AI for public services, the DWP hopes to create a more resilient system that can handle the growing pressure of an aging population and changing economic conditions. If successful, the conversational AI model could eventually be expanded to other government departments facing similar administrative bottlenecks.
