I used AI to Change How I Think (on purpose)

And, it taught me how to code. Now the only challenge left... what is worth building?

What if we reversed the dynamic of how we interact with AI.

I thought to myself back in February….

"Humans prompting AI… what if, AI prompts human?"

But before I could build that app… I needed to learn to code. So I hired my friend to give me some live 1:1 coaching sessions. And then before my first session, I asked GPT to look at the ‘HOMEWORK’ he gave me, and help me out.

I was fully expecting to need to do 10+ lessons with him… but after about 15 minutes of sending that message to GPT, I had my coding environment up and running. And within 24 hours of sending that message I had my first ever TestFlight app live on my iOS device. In short, AI totally removed the need for me to learn from a human, rather, I 100x’d the speed because I leaned heavily on AI to teach me. And, I learned by doing. Trial and error. Copy and paste. Screenshot after screenshot being moved into GPT, asking ‘what the heck does this mean???’

I was extremely frustrated at first, but that quickly shifted once I had the v1 of my reflection app on the Xcode Simulator. Then I was actually using my own product to self reflect and calm me down.. fighting off bugs became a trial of the product. It was quite cool, I used it as a dev tool, while I learned dev.
I was literally using this mirror to make important bug-squishing decisions during product development.. so I thought, hey that’s a good value prop. And built it into the on-boarding of the product. 😂

After 7 weeks, I eventually became quite decent. I first a basic AI on understanding universal polarities (niceΞmean, freedomΞstagnation, etc…), and structured the AI convo-algo to draw attention to each polarity in the form of questions. This way each conversation leads me from complaining about my perspective, to understanding multiple perspectives. Essentially I wanted to train my brain to be even more optimistic, seeing opportunities to act from multiple perspectives. And I would train my brain by using AI to prompt me in ways I otherwise might not have thought to.

The final version of the mirror is going to be available for free to anyone who joins my event on Wednesday! Co-featuring Peter Rojas from Mozilla (formerly Meta, Betaworks Ventures, Engadget, etc)!

That first custom AI pushed me further down a rabbit hole of teaching myself to fine-tine models, so that I could get my custom GPT onto iOS Store. Most recently I was able to fine tune a forked version of GPT 4.1 which was trained on custom training/validation database that I built using my entire history of GPT convo + Terminal + and a ton of JSONL files.

This was a grind… I ended up submitting over 20 versions to the App Store as I worked through trial and error.

It took me about 7 weeks in total to build mirror, and then in the last 1 week I threw together four other ideas 😂😂😂😂😂😂😂.

Idea 1: LockedIn: A platform where fortune 500’s issue challenges to recent college grads. This way they prove it with action, not a resume or pedigree.

Idea 2: An app that issues questions to a global audience, best answers get voted up and followed. Helping thoughtful people, and intelligent answers get attention

.Idea 3: A stupid simple RSVP system for podcasters. Right now it’s such a hassle to get RSVPs to add event to calendar for a live podcast. Eventbrite ‘add to calendar’ never works, etc.. So I built a Gcal native one-tap rsvp. I think others should use it, bc I love it, and have done 100s of these over a decade.

So.. I’ve gotten to the point of being able to build MVP’s of ideas in a matter of hours or days. Of course the design elements aren’t there, the brands aren’t there, and the ideas are half-baked, but this ‘exploring in the woods’ mindset seems to be a necessary ingredient to finding product market fit. Especially when you’re technical enough to build MVPs for free and wealthy enough to spend $300-500 on testing each with ads.

I’m very thankful for the independence here! And really keeping fingers crossed I can figure this out!!!

Thank you for reading! I hope to see you Weds when we have Peter Rojas joining us! I first met Peter when he was founding partner of Betaworks Ventures, a $50M NYC fund, and then we both spent time working at Meta. So it’s a decade long relationship, tune in to hear us chat AI, catch up, and talk ideas.

p.s rsvp link here for your ease: