From little leagues to world-class think tanks

Jake Spurlock helps run some of the biggest websites on the internet. As a forward deployed engineer at Automattic's enterprise division, WordPress VIP, he works with publishers like Condé Nast, Pew Research Center, and government agencies. He's been in the WordPress ecosystem for over 20 years.

With three kids of team-sports age, he's also a Little League player agent, a baseball field dragger, and the guy who built a rain forecast app for a single field a quarter mile from his house—in about an hour!

The “I can solve this” itch

Jake helps run a Little League in the Bay Area. A quarter inch of rain shuts the fields down. He got tired of checking AccuWeather, so he decided to build a hyper-local forecast that would aggregate all of his exact field locations in one easy-to-check place.

He already knew how to build a website. He knew how to work with an API. Apple had kept the tech from Dark Sky (the beloved hyper-local weather app they acquired and shut down) alive as an API called WeatherKit. So he wired it up, passed in GPS coordinates, got a clean response, and in about an hour, he had a working site showing forecasts for all 13 fields in his league.

"It was kind of a big unlocker for me," Jake said. "It showed me the acceleration that was possible. Here's this problem, this little itch, and here's a way to really speedily solve it."

That moment rewired how he thinks about problems. After that, every itch became a question: Can I solve this quickly? And if the answer was yes, he followed it. The rain app led to more apps, more side projects (and eventually to a workflow that changed how one of the most cited research institutions in the country builds software).

The 80% rule in action at Pew Research

Pew is one of the most cited research institutions in the country. They publish five to ten dense, well-crafted articles a week. They have a small development team. Their question was simple: how do we maximize the return on every hour of engineering effort?

Jake and the Pew team settled on a target: get AI to produce 80% of any given pull request, so a developer only handles the remaining 20%.

Here's how it works: 

• They created a GitHub issue template with a clear structure: a description of the problem, and detailed acceptance criteria for what "done" looks like. 
• When someone files an issue that meets the template's criteria, a bot reads it, confirms it qualifies, and hands it to an AI agent. 
• The agent opens a pull request. It writes tests. It follows coding standards already established in the codebase.

Well written acceptance criteria, and really training the org on how to write good acceptance criteria, is what makes this whole thing work. 

"If I can be incredibly clear in what I want, and the standard I want it built to, and the skill I want it to be using, I can really get over the fence at 95, 99, even 100% a lot of the time," Jake said.

The target was 80%. They're hitting 99%. The most interesting thing here is who gets to participate.

Before this workflow existed, only developers could turn ideas into working software on Pew Research websites. Product managers could request things. Web producers could file bug reports. Researchers could describe what they wanted in a meeting. But the actual act of creating was bottlenecked through a small engineering team.

Now, product managers write issues. Web producers write issues. And researchers, the people who actually know what the next report needs to do, write issues too. If they can write clear acceptance criteria, an agent builds it while everyone sleeps.

"AI is democratizing a lot of processes that were previously held only by developers," Jake said. "If this allows more people to be part of this workflow, it helps out everybody."

There’s a vision of the future where anyone writing intent is how much of it gets created—instead of strictly engineers writing code. 

Your kid deserves a walk-up song

Not all of Jake's projects are enterprise workflows. Some of them are just fun.

Every Major League player gets a walk-up song. A booming announcer. The crowd on their feet. Jake thought his kids deserved the same treatment at Little League games. So he built Walk Up DJ.

The app lets you build a roster, assign each kid a song from Apple Music, and play it when they step up to bat. He added AI-generated stadium announcers. One voice is called "Los Angeles." Another is called "Milwaukee." If you know, you know.

Walk Up DJ is on the App Store. Real families are using it. It's his only paid app, because the AI voice generation costs real money. Somewhere right now, a ten-year-old is walking up to the plate, feeling like they’re in the big leagues already.

Today: making RSS useful again

One of Jake’s “itches” was redefining how we interact with the news on a personal basis. Also, bringing back RSS. So, Jake built an app called Today, an RSS reader that uses Apple's on-device AI to read your feeds and tell you what actually matters.

The inspiration came from his work at Automattic, where the internal communication system generates a massive RSS feed that can be overwhelming to read. He wanted to press a button and have AI surface the important stuff. So he built it.

The app pulls in feeds from anywhere (Reddit has full RSS support for every subreddit, for example), runs them through Apple Intelligence on-device, and generates a daily briefing written with personality. Jake modeled the tone on Dave Pell's Next Draft newsletter: enthusiastic, opinionated, human.

Today is available as three separate builds: Mac, iPad, and iPhone.

Building agents or being agentic?

In modern parlance, what Jake is doing would be called “agentic.” He isn’t waiting for permission, for a vendor to solve his problem for AI to be perfect—he’s just identifying friction, writing clear intent, and shipping things that real people use.

At Pew, that looks like researchers writing GitHub issues that become production code overnight. At home, it looks like a rain forecast for a baseball field and a walk-up song app that makes ten-year-olds feel like big leaguers.

The common thread is that AI is most valuable when it's in the hands of the people closest to the problem. Just clear thinking, clear writing, and a platform that turns intent into action becomes a major unlock for every organization willing to put the tools in the right hands.

Jake Spurlock is a forward deployed engineer at WordPress VIP (Automattic). You can find him at jakespurlock.com or @whyisjake on basically every platform. Walk Up DJ is available on the App Store. Today is available for Mac, iPad, and iPhone.