News

DWP posts £19.5 million tender for conversational AI and natural language call steering solution

The Department for Work and Pensions (DWP) has posted a contract with a total value of £19.5 million for a conversational AI and natural language call steering solution to help direct citizens to the correct services and support self-service.

The new solution will aim to allow callers to speak naturally to explain why they are calling, before routing them to the correct agent or self-service offering, integrating with the department’s existing contact centre to improve how it currently handles inbound calls.

Interested suppliers should be able to demonstrate their solution’s existence as “UK-shored in a dedicated cloud environment”, compliant with applicable policies, standards, and frameworks, the DWP notes. These include NIST and NCSC standards.

A maximum of five suppliers will be invited to tender, with suppliers required to submit information, confirm adherence to minimum standards, and outline technical ability as part of the first stage of procurement. This will be assessed through a procurement specific questionnaire, with the scored component focusing on technical and professional ability.

Suppliers will be required to submit a signed Non-Disclosure Agreement, along with a Financial Viability Risk Assessment Tool, templates for which are available on the procurement portal. A draft contract containing service specifications, performance levels, and an outline implementation plan, can also be accessed.

The five suppliers scoring highest on the initial stage will be invited to tender, which will see the submission of pricing in the form of a financial model.

Contract dates are given as the 6 July 2026 t0 5 July 2030, with a possible extension of two 12-month periods to July 2032.

Wider trend: Digital transformation in government services

The UK government has published its roadmap for modern digital government, an action plan bringing together integral products, platforms, and transformation initiatives to demonstrate how digitalisation across the government is improving public services, increasing accessibility, and promoting value for money. The roadmap aims to deliver on five themes: easier lives, faster growth, firmer foundations, smarter organisations, and higher productivity and efficiency, with progress being tracked through activity levels, delivery of projects and commitments, and outcomes such as user satisfaction. For citizens, public services will be redesigned to be more efficient and accessible. Public sector workers will also benefit from access to modern digital tools, data and skills, including AI tools to help complete tasks more quickly, better access to joined-up, high quality data, and opportunities to build digital, data, and AI skills.

The UK Government has updated its Cyber Action Plan, to tackle “critically high” cyber risk as part of the Roadmap for Modern Digital Government, looking to move toward proactive action, clear accountability, mandatory requirements, and comprehensive central support. £210 million has been invested in forming a new Government Cyber Unit, to provide direction and expert support. The government shares findings from the first year of GovAssure, its cyber security scheme for assessing government critical systems, noting “significant gaps” in departments’ cyber security and resilience, and levels of low maturity with asset management, protective monitoring, and response planning.

The Department for Science, Innovation and Technology (DSIT) is partnering with Anthropic, an AI safety and research company, to build and pilot an AI-powered assistant for the GOV.UK website following DSIT’s “scan, pilot, scale” framework supporting testing and iteration prior to wider roll-out. The AI assistant is said to help people visiting the government website to navigate services, with an initial focus on employment, finding work, accessing training, and understanding available resources. According to Anthropic, the assistant will be capable of maintaining context across interactions, allowing people to pick up from where they left off when returning to the website. Users will also have “full control” over their data, it states, including what is remembered about their interactions.