With the government seeking to make public services more inclusive, could Voice AI be part of the answer?
That was the question explored in our workshop “Serving Every Postcode: Voice AI, Agents And The New Agenda For Citizen Services”, bringing together public sector digital leaders to learn from the NHS, ElevenLabs, and the Ministry of Justice.
Voice AI agents are can understand and respond to spoken language, enabling them to hold conversations, answer questions and carry out tasks on behalf of users.
As the technology handles increasingly complex interactions, participants discussed how it could improve accessibility, streamline interactions with public services, and reduce the administrative burden for frontline professionals.
Speaking at the event, Matthew Hills, GTM Lead for Government and Public Sector UKI at Voice AI company ElevenLabs, highlighted the potential applications of Voice AI agents, which include 24/7 voice assistance, multilingual service delivery, faster routing and resolution of enquiries, and automated guidance and triage for benefits and eligibility services.
Voice AI, therefore, potentially has the ability to modernise public services. However, while recognising these potential benefits, workshop participants emphasised that “the tool is only as good as the deployment”.
A recurring theme throughout the discussion was the importance of putting the principles of trust and inclusion at the core of building AI-enabled services. Indeed, Hills argued that trust and observability are what an organisation should begin with.
He cited the need to build guardrails into technology from the very start, as well as implementing ongoing monitoring to understand how tools perform in practice. At a time in which, as ElevenLabs emphasised, there is a lot of fear around AI, being transparent about how systems are designed and secured is crucial. Equally crucial is consulting end users, whose input is necessary to ensure that tools achieve their stated aim.
In a similar vein, attendees warned against introducing AI for its own sake, arguing that its deployment should seek to meet specific needs and achieve clearly defined outcomes. These outcomes, one facilitator argued, should fundamentally centre around giving time back to staff and to improving user experience.
Presenters spoke of how digital transformation initiatives that fail to prioritise inclusivity risk creating more barriers than they remove both for staff and for user groups such as elderly and immigrant communities. For example, moving the appointment service for a clinic entirely online could reduce accessibility for those who lack digital skills or familiarity with English.
Participants stressed that to improve inclusion, it is crucial to rethink the entire user journey when moving from analogue to digital, rather than simply focusing on digitising existing processes. Voice AI could help address some of the barriers created by unfamiliarity with digital processes or with English, for example, by offering more natural, conversational ways to interact with services, including in different languages. However, its deployment must form part of a wider approach to service design.
Ultimately, presenters emphasised how Voice AI is not intended to replace human interaction, but rather to help professionals make better use of it by allowing them to focus on the complex cases and personal interactions where their expertise is most valuable, and to simplify citizens’ experience with public services.
The challenge is not simply to identify where Voice AI can be deployed, but to understand where it can genuinely improve outcomes for citizens and staff, while ensuring that those who need human support can still access it.