Permit me to make a very big deal out of a very, very small slice of the history of journalism.

In the mid-2000s, when newspapers were still quite large and fully panicking about what to do about the digital product, a movement formed around what ended up being called news apps. Now, young reader, today we think of “apps” meaning “phone” but this is the pre-iPhone days. What news apps developers made in the mid to late 2000s were websites. But not just websites — they were big, data-driven, searchable, clickable things. You explored them.

It took a blending of skills that didn’t exist in newsrooms natively — programming and design and engineering and scaling. It also rode the wave of an innovation of the moment: cloud servers. Don’t have a server to put this on? Rent one! Easy! Pirate bands of innovators tried to change what a unit of journalism was. Journalists learned to code. Coders came into journalism. Cool things were built. Prizes were won.

I have a lot of fondness for this blip in time. I built one of the enduring ones — PolitiFact. I started my career as a night cops reporter. I had no business writing software to create a whole new unit of content and deploying it to someone else’s servers. Yet, here I was.

But as the business model under it all slowly collapsed, all but the biggest places realized that running these apps was expensive. And the more you deployed, the more you had to maintain. With once great newspapers struggling to keep the lights on, and new non-profit entities just trying to figure out if they were going to make it, fancy web apps on scalable infrastructure became a luxury most couldn’t afford.

The news app was dead.

In 2026, thanks to AI, long live the news app.

Today, a whole lot of people are trying to figure out where AI fits into news creation. Some places are using it to create more Stuff to put on the Thing — in spite of the audience not asking for it. Some are using it to help with story discovery. Other places are using AI to help with accessibility, SEO, and translation to other languages. There’s a lot going on.

I’ve got another to suggest: The Throwaway News App.

Instead of expensive and ambitious apps that will require updating and maintaining, in 2026 we will use AI to make quick one-page web apps that let a user search through some data to find something they are interested in. A little HTML, a little CSS, a little JavaScript, Bob’s your uncle and boom, news app. A news app that is going to get some attention from the story it’s tied to for a few days and then fade into the memory hole like so much of what we do.

The New York Times election website this is not. Nor will it ever be.

But here’s the thing: you can make a real simple one of these in about 15 minutes with AI. Give it some materials — a data file, a logo image, a URL to a stylesheet, a few instructions, and poof. It’s pretty well done. A little update here, some additional text there, and all you need is something that will serve up a static HTML page. Good news, just about anywhere can do that. Cheap.

The me alive in 2006 would freak out entirely that something like this can be done in 15 minutes. It’s so easy, I sprung the idea on a group of Nebraska data journalism students one day in class. None had any web development experience. But they all had working websites online in less than an hour, and half of that was downloading tools and setting up accounts. They picked up a new skill, made part of the internet, and left class early that day.

Is this high art? No. Do these students understand the intricacies of JavaScript and how JSON is loaded and how the result set inherits from which CSS classes? Nope. That is not the point. Is the point that you can make a thing you need to go with a story with very minimal time investment and the cheapest of servers?

Absolutely yes.

For giggles, I made this in less than 20 minutes using Google’s Antigravity product, which is a copy of a widely-used code editor with Gemini 3 built right into it. I found the RSS feed for Nieman Lab’s 2025 predictions (which took about two minutes), wrote a couple of prompts to parse the RSS feed (about five minutes), and then told it to create a website around it. Gave it a quick once over, tried a few things, and pushed it to my GitHub account. I provided nothing more than the RSS file and an idea of how I wanted it searched and displayed.

Give it a whirl!

Matt Waite is a professor of practice at the College of Journalism and Mass Communications at the University of Nebraska.