Navigating the AI Reckoning: Practical steps for the public sector

Good afternoon everyone, and thank you for joining today’s webinar. We’ll be exploring key themes from our report on navigating the AI reckoning and what it means specifically for government organizations.

My name is Jake, I’m an Account Director at Squiz, and I’ve worked with public sector clients for around 10 years, with 15 years in total across the sector. I’m joined by Nick Condan, Managing Director at Squiz, who previously spent a decade in digital government as a Chief Digital Officer, leading large-scale web estates and transformation programs.

Before we begin, a few housekeeping notes. This session is being recorded and will be shared afterwards, so there’s no need to take detailed notes. If you have questions, please use the Q&A box rather than chat, and we’ll address them at the end or follow up afterwards. If you experience technical issues, use the chat and we’ll assist you.

While today’s examples focus on government, the principles apply to any organization managing complex digital environments. For those unfamiliar with Squiz, we are a global digital experience platform provider, helping organizations deliver effective digital services for over 25 years. Our AI-powered platform is designed to support both non-technical users and developers, but today’s focus is not the platform itself—it’s your content and how to make it AI-ready.

Our goal today is to move from understanding the problem to identifying practical solutions. AI is now embedded in everyday workflows, and citizen behaviour is changing rapidly. This directly impacts how government services are discovered and delivered.

AI is reshaping how people find information. Citizens increasingly rely on tools like ChatGPT and AI-powered search results rather than navigating traditional government websites. There has been a significant shift toward prompt-based interactions, with hundreds of millions of users adopting these tools in a short time. At the same time, a large proportion of search queries now end without users clicking through to websites, as AI summaries provide immediate answers.

This creates both a challenge and an opportunity. Citizens are often using AI tools before visiting official government sites, meaning your content must be accurate, up-to-date, and easily interpreted by these systems. If not, outdated or third-party information may be surfaced instead, potentially leading to misinformation and reduced trust.

To illustrate this, consider a scenario where a small business owner seeks financial support after flooding. Instead of navigating multiple government websites, they ask an AI tool. The response appears authoritative but is based on outdated information, leading to confusion and a poor experience. Crucially, this interaction happens outside your digital estate, meaning you may not even be aware of the issue.

This lack of visibility is one of the biggest challenges. Traditional analytics show what happens on your website, but not what happens in AI-driven discovery. Meanwhile, content surfaced by AI often drives highly engaged users, meaning errors can have significant impact.

Understanding how large language models work helps explain this. These models are trained on vast amounts of data, breaking content into smaller components and identifying relationships between words and concepts. Once trained, they represent a snapshot in time. To stay current, they use retrieval processes—searching the web in real time, analysing top results, and generating responses based on both stored knowledge and newly retrieved content.

This means your content must be both well-structured and regularly updated. AI does not truly “understand” content—it predicts likely answers based on patterns. If your content is unclear, overly complex, or outdated, those predictions may be inaccurate or your content may be ignored entirely.

AI systems prioritise speed and efficiency. They typically analyse only the top search results and extract the clearest, most structured information. Content that is concise, specific, and easy to interpret is far more likely to be used.

Common challenges include heavy use of jargon or legislative language, reliance on PDFs, fragmented information across multiple pages or departments, and assumptions about user knowledge. For example, policy-heavy language may be technically accurate but difficult for both citizens and AI to interpret. Rewriting content in plain language with clear eligibility criteria and actionable steps significantly improves usability and visibility.

Importantly, this aligns with existing best practices in government digital services. Writing in plain English, structuring content around user needs, and simplifying navigation have long been priorities. AI simply raises the stakes and amplifies the impact of getting this right.

Accessibility is a critical factor. AI systems interpret websites in ways similar to assistive technologies such as screen readers. If your content is not accessible, it is less likely to be discovered or correctly interpreted by AI. This creates a strong alignment between accessibility improvements and AI visibility.

To improve content visibility, several key principles should be applied. Content should be consistent in terminology and structure, specific in detail, complete in answering user questions, consolidated into fewer high-quality pages, and kept up to date. These principles benefit both human users and AI systems.

A shift toward question-and-answer formats, such as FAQs, can be particularly effective, as AI tools are designed to respond to natural language queries. Reducing duplication, removing outdated content, and ensuring clarity all contribute to stronger performance.

However, implementing these changes at scale can be challenging, particularly in government environments with large volumes of content and distributed ownership. Organizational factors such as limited resources, fragmented teams, and governance processes often present barriers.

To address this, structured approaches and supporting tools are essential. Content auditing tools can analyse large volumes of pages, identify issues, and prioritise improvements. A practical framework involves identifying priority areas, generating relevant user questions, updating content, and establishing governance processes to maintain quality over time.

The goal is not to fix everything at once, but to focus on the most impactful areas and demonstrate value quickly. This helps build momentum and secure stakeholder support.

Ultimately, AI is not introducing entirely new requirements—it is reinforcing the importance of good content practices. Clear, accessible, accurate, and well-structured content has always been important. AI simply makes it essential.

Government organizations have a responsibility to ensure that citizens receive accurate and trustworthy information, regardless of how they access it. Whether through official websites or AI tools, the content must be reliable and up to date.

By adopting the right strategies and tools, organizations can reduce risk, improve visibility, and deliver better digital experiences for citizens.