Digital Citizen Services – From Browsing to Conversational and Agentic Interactions

Website interactions are shifting from static, keyword-based searches to personalised, conversational AI experiences, but human interfaces will remain crucial.

Conclusion

The evolution of citizen-facing public services websites over the next few years will be profoundly shaped by advancements in artificial intelligence (AI) and natural language processing (NLP). This shift marks the biggest change since the mouse in human-computer interaction.

Observations

Traditional websites, often functioning as static brochures or fixed navigation systems, are becoming insufficient for user demands. Users will increasingly expect instant, personalised answers in their own words, anticipating that websites will listen and talk back. This is not merely a preference but a fundamental redefinition of how humans and, increasingly, AI agents interact with digital content. This will not happen overnight: it will be an evolution over the next few years. Regulatory changes could speed up how quickly new technology is developed and adopted by users. For example, Denmark’s1 recent move to give citizens copyright of their own face, body, and voice might make people more comfortable using voice commands and being monitored by technology2.

However, the notion that websites and other systems of interaction will be entirely replaced by AI agents that users just talk to is strongly rejected by IBRS. Instead, a layered approach is necessary. The future of human interaction with information will involve three distinct forms of interaction3:

  • Verbal Interactions: Ideal for initial discovery tasks, akin to conversational search. This will be common on mobile devices, headphones, and smart assistants.
  • Touch and Explore: Touchscreens and pens will remain crucial for organising thoughts and creative construction.
  • Deep Work Tasks: The combination of keyboard, mouse, and touch will continue to be central for in-depth tasks, though these tasks may start with verbal queries and move through touch for information organisation before deep work begins.

Public Sector Services Websites: Beyond Information Retrieval

For public sector services – and councils in particular – the evolution of websites demands more than just AI-powered information retrieval. They need to be “structured to allow searching through text, voice, and, crucially, with empathic design”4. This is complicated by the need for digital identity. Evolution means moving beyond merely answering questions to guiding users through processes, offering interactive wizards, and helping them accomplish tasks, rather than assuming prior knowledge. The use of AI in providing natural language, in-context advice and assistance as citizens engage in council services will need to be far more nuanced than a mere ‘chatbot’ based on FAQ (frequently asked questions) documents, if all citizens are to benefit from the new technology. Instead, the website (pages) will need to evolve to be responsive.

Natural Language Web Standards

The early signs of this evolution of how websites function are here. In late May, Microsoft revealed its early framework, which allows static web page content to be fed into AI to “recontextualise it for what the person is actually trying to achieve” based on user input. This framework is called natural language interfaces for websites, or NLWeb. Currently, this framework remains solely within the realm of Microsoft and is not supported by other vendors. However, even if Microsoft’s offering does not become an industry standard, IBRS predicts that an industry-wide standard will emerge and mature over the next 3–5 years.

Google as Council Inquiry Agent

Other important players in the age of intelligent websites will be Google and AI vendors that are supplanting traditional search. Google is likely to increasingly regain its role as the gateway to worldwide information, but with AI that already provides not only links to information but also detailed initial answers to user queries. For commercial content creation, this will be a serious problem. However, for governments, particularly local councils, it will be a boon, effectively providing a well-supported, sophisticated AI interface that takes into account citizens’ private details and context, fed by the council’s information and listed services. Councils’ websites will remain responsible for “fulfilling and starting, organising and fulfilling the citizens’ services”, but Google, and possibly other post-search AI vendors, may very well be the citizens’ primary interface into those services.
Public sector service design teams should monitor Google’s migration from search to contextual AI inquiry and identify the most cost-effective ways to pass information to Google, ensuring that Google’s responses to citizens’ requests are sensible, up-to-date, and effective, and that citizens are directly led into empathetic service delivery.

Next Steps

  • Do not rush into implementing a chatbot or virtual agent on websites. Instead, step back and consider the need for empathic design, as detailed in the Australian Digital Citizens 2025 report. The data from this report was conclusive: simply adding an AI assistant does not increase the practical accessibility of services. While an agent may be an option, the empathic design considerations must be the first step.
  • Explore the changing role of Google, from search to contextual advice. Consider how Google’s evolution could benefit your service delivery, and how any risks associated with this evolution should be mitigated.
  • Monitor the introduction and evolution of emerging natural language web standards, such as Microsoft’s NLWeb.
  • Continuously monitor regulatory developments that shape and influence technology advancements related to user rights and safeguards. Ensure contracts with technology vendors provide substantive assurances about user privacy and property rights that will evolve as these advances continue.
  • Engage with your core solution providers regarding their roadmaps for AI-enabled customer and citizen portals. If possible, become involved in their customer-steering groups to help shape new product releases.

Footnotes

  1. Deepfake legislation: Denmark moves to protect digital identity’, World Economic Forum, 2025.
  2. The 2025 AAMI Crash Index analysed over 480,000 insurance claims across Australia and found the most common car safety feature people turned off was lane keeping assist (45 per cent), followed by adaptive cruise control (17 per cent), parking assist (17 per cent), automatic emergency braking (16 per cent), and forward collision warning (11 per cent)
  3. The Future of End-User Computing: The Biggest Change Since the Mouse is Upon Us’, IBRS, 2023.
  4. Australian Digital Citizens 2025’, IBRS, 2025.

Trouble viewing this article?

Search

Register for complimentary membership where you will receive:
  • Complimentary research
  • Free vendor analysis
  • Invitations to events and webinars
Delivered to your inbox each week