I Took the AI "Red Pill" This Weekend
My first foray into vibe coding and why I can't stop
Over the past two weeks, there’s been something in the air.
Was it those Claude ads? Was it that viral X post,“something big is happening,” that everyone shared and then everyone had an opinion on? Was it my founders incentivizing us to use AI more? Was it my coach hijacking our session to start with this topic?1
It was all of the above.
And then some.
I’ve used ChatGPT religiously for over a year. To partner with me on writing. To challenge my strategy. To ideate. To take my dictated voice notes, my written post-its and white-boarding sessions and put them into clean documents.
It’s been an incredible partner to me.
But I’ve never veered into prototyping. The vibe coding of it all. Which is kind of ironic, because as a product person you’d think I’d be an eager beaver to build independently.
What is vibe coding: Most simply put, vibe coding is a way to prototype and build web apps, sites and tools using written prompts. What previously took technical engineering and design skills can be built with written prompts, like “build me a workout app that tracks my reps and sets.” More on this workout app below…
Anyway, this weekend I got into vibe coding.
I started using Claude Cowork, and I kind of can’t stop.
At Hampton, I’ve been working in a very niche problem space for the past eight months, evolving and developing new materials, working with limited data to make decisions and roll out materials to hundreds of customers. And I've been doing this with a mishmash of different tools, as I always have.
This weekend was the first time I realized: “Holy shit, I can do all of this in one single place… if I build it.”
And so I did.
I spent about six hours building a basic front-end app that serves as a visionary prototype. I stayed up late on Friday. I woke up early on Saturday, simply because I was so excited to keep going (and I kept hitting my limits2).
Here are three things I’m realizing as an amateur vibe coder who’s late to the game3:
1. Having a product background gives me a leg up.
I’ve worked in Product for over 15 years. I know the language. I have basic product, design and engineering principles locked in. Therefore, I know how to prompt quickly because I’ve spent years building websites, apps and more. And because I’ve spent years doing this over and over again, it’s all second nature to me - which gives me a lot more language to use when building a front end. It’s a head start, for now.
2. Having the raw material was necessary for me to move as quickly as I did.
I didn’t use Cowork to build raw materials from scratch, although I’m sure people could. I created all the raw materials over the past eight months. What I did was use the raw materials to make what I was building highly interactive. It basically took a ton of static pages and made them dynamic. Cowork quickly learned what I’m trying to do and started using my raw materials to develop on top of what I gave it.
I know this is terribly vague and while I’d love to share screenshots, its not ready for public consumption.
3. My moat is my ability to exploit these tools.
I am notoriously terrible at being new to technology. I don’t like watching YouTube videos. I don’t like reading manuals. The beginning of learning new technology is painful. But what I’m realizing is that my ability and willingness to sit in the pain - to figure things out uncomfortably - will be my moat as a knowledge worker. And my ability to exploit these tools, to move quicker, to do things that I need to do without waiting on anyone else - that’s my competitive advantage.
The tools are available to everyone.
Dare to hop on the learning curve.
What does it all mean?
Who could pretend to know.
What’s obvious to me: people with clarity on their own requirements, what jobs they actually need done, are going to start abandoning all the different software that’s trying to be everything in favor of building the exact thing they need.
Here’s what I mean.
I’ve been trying to find a workout app that tracks my workouts the way I want. Nothing quite fits. I’ve been using ChatGPT to build workouts from my personal training and physical therapy recommendations, hosting them in my iPhone Notes app, iterating back and forth. What if I just... built my own app that does exactly what I want?
Same thing with shopping. I’ve been saving things I want to buy in Notes, running them through ChatGPT to figure out what goes with what, and where to find it all at the best price. Again, all this raw material, cobbled across tools. What if I just built an app to do the thing I actually want to do?
One of the most painful things about starting my coaching business was hobbling together Squarespace and Google Drive and PaperBell and a dozen other tools. i documented the process here. Not one of them did everything. But now I’m sitting here thinking: “Wait, I can just build exactly what I need, with the exact requirements I want.”
What feels non-negotiable after this weekend is introducing yourself to these tools. Not just because they’re fun (they are so fun, though), but because they are going to be the differentiator between you and another knowledge worker.
Rolling up your sleeves and familiarizing yourself with what’s happening right now is going to be necessary.
You can either wait until you have to, or you can read a little post like this and be inspired to start today.
I’m Jori Bell, VP of Core at Hampton. I’m also a Coach for Product Leaders. My coaching practice is full this quarter but if you need coaching recommendations, reach out.
Thank you, I’m obviously grateful you did this, M.
I have since upgraded, thank you so much.
It’s all relative, right?
yesss