How AI and LLMs Will Transform Government Websites—Again

How AI and LLMs will change government websites

One of the first jobs I ever had in government technology was leading the complete redesign of the hospital quality reporting system. This was over 7 years ago, but strangely enough, I still see some of the same challenges—and opportunities—playing out today as we look ahead to the AI-powered future of government websites.

It was 2017, and the United States Digital Service was pivotal in wanting to update its first-generation, fairly clunky CCSQ process that paid hospitals for Medicare and Medicaid. Back then, most government websites were still organized like an org chart. If you didn’t know which agency was responsible for what you needed, good luck finding it.

We took a different approach. Instead of organizing content by agency, we organized it by user need. Whether you were a resident, a business owner, a state employee, or someone else, we built pathways that helped you get to the right information, without needing an insider’s map of state government.

Putting Human Needs First—Before It Was Obvious

We also made a bold choice at the time: we put human needs front and center. This was new. That work anticipated a major shift in how people would come to find and consume government information—not by starting on a homepage, but by typing questions into search engines.

As Google and other platforms became the de facto starting point for citizens looking for services or answers, governments had to pivot. They started investing in search engine optimization (SEO), adding sitemaps, refining content, and rewriting web pages to be more easily indexed. They built shared services and even partnered with search companies to make sure their content showed up when it mattered most.

Now, AI and LLMs Are Reshaping the Landscape Again

Today, we’re at the start of a new shift—and it feels eerily familiar.

Just like search engines crawled and indexed web content to return relevant results, large language models (LLMs) like ChatGPT are now trained on that same public information to generate human-like responses. And just like with early search, there’s both excitement and concern—especially around accuracy.

Let’s face it: when someone asks a model like ChatGPT for a tax deadline or vaccine requirement and it gets it wrong, the consequences can be serious. Governments are authoritative sources, and yet their websites aren’t always optimized for how AI consumes and interprets information.

But here’s the opportunity: government agencies can lead the way in making content LLM-ready. By thinking about language, formatting, metadata, and retrieval-friendly architecture, agencies can start to shape the quality of the responses LLMs give. Practices like document question answering and structured prompts can dramatically improve accuracy and ensure the right information reaches the right people.

We’ve Been Here Before—We Know How to Adapt

We’ve already adapted once, from brochureware to search-optimized content. Now we need to do it again—this time for LLMs.

This shift isn’t just technical. It’s about rethinking how people interact with government. Instead of clicking through menus or scanning dozens of PDFs, they’ll be able to ask a question in plain language and get a trustworthy, contextualized answer. It’s a simpler, more human interface—but it only works if the content behind it is ready.

I’ve been working hands-on with LLMs to understand how they work, where they fall short, and how government can use them responsibly. If you’re thinking about what your agency’s digital future looks like, now is the time to start planning for this shift.

Want help making your content AI- and LLM-ready? Let’s talk. The next generation of government websites is already on the horizon—and we can shape it together.