How a Liberal Arts Student Started Writing Code

I see this request 11 times a week from business owners:

Good news and bad news.

Bad news:

That person barely exists.

The tools are too new.

Best practices don’t exist yet.

And forget hiring a “full stack developer.”

Asking them to build AI tools is like asking a custom woodworker to run an IKEA factory line.

Different mindset entirely.

Writing software with AI is about speed and cost—not scale and perfection.
That offends most traditional devs. I get it.

But it’s apples and oranges.

Good news:

You don’t need them.

All you need to do is:

Play.
Play.
Play.

The people writing software with AI today are figuring it out as they go.

They try something, break it, ask why, and try again.

If you told me to find someone good with AI tools right now, I’d run a public contest—like they did in WWII for codebreakers.

“Build something useful with XYZ tool. Winner gets the job.”

Could be a 15-year-old. Could be a high school dropout working in a restaurant. Doesn’t matter.

The companies gaining an edge aren’t the ones with the biggest AI budgets.
They’re the ones setting aside time to mess around.

Worst case: You waste an afternoon.

Best case: You build something that saves thousands of dollars and countless hours.

You don’t need permission. Or credentials. Or to be “technical.”
(Though appreciating how technical stuff fits together helps.)

You just need curiosity.

And a willingness to poke at things.

For reference:
I don’t know how to code.
The last math class I took was in high school.

But here’s what I’ve built with AI tools just by messing around:

Here are a few things I’ve learned coding with AI, playing with a few dozen different tools:

1. Mobile App for Newborn Twins

We just had twins. I didn’t like any of the baby tracking apps. So I built one.

• Took 4 hours

• Taught me I like mobile development

• Helped me see how much is possible now with drag-and-drop frameworks + AI

2. Resource Hub for Sagan’s Digital Power Tools

My colleague Brian built v1. I modified it using AI.

• Took 3 hours

• No login required, so it was fast to deploy

• Shocked me how good the front-end looked

3. Full Tool for Working with Executive Assistants

Wanted a custom tool for our member’s executive assistants at Sagan.

Nothing bloated….just simple and effective.

To build it, I had to figure out:

• GitHub

• Vercel

• Supabase

• Internal API calls

• OAuth

• SMTP setup

• Debugging 30+ authentication errors

BUT IT WORKS.

We’re rolling it out to members soon.

Time: No clue (I blacked out halfway through and woke up with a working app.)

Next up for me:
A single backend that works across both a native mobile app and a web front end—coded with AI.

I’m close.

Here’s what you need to know if you aren’t seeing why you should care about this.

I (and you!) without knowing a line of code, can build fully deployed, custom software for 1/100th of what you used to have to pay.

The implications to this are years away from being fully understood.

Try building something this weekend.

Reply and tell me what you made—or what broke.

Also, I just recorded an episode where I walk my cohost Peter through what AI coding tools I’m using right now, and getting started using them. Dropping the episode this week, but you gotta be subscribed to the pod!

Yallah Habibi,

Jon

Passage of the Week: