Money Toolkit: First Release

My first release of my first app. This feels awesome!

Here it is: moneytoolkit.app

✅ Found a solid tech stack

✅ Deployed PoC to a dev environment

✅ Bought a domain

✅ Setup production Clerk config

✅ Setup Vercel production deployment

✅ Setup production Convex config

✅ Monetization - via Buy Me a Coffee

✅ Open for sign-up

It’s still under active development; I’m polishing a few rough spots. Already got my first user, though!

Learning

Honestly, this taught me a lot. A couple months ago, I don’t think I would’ve said I had much to learn from deploying an app I wrote mostly with AI. Weirdly, I used Codex to write the majority of this application, but still felt pretty much in control the entire time.

I’ve never felt more like LLM-driven development is truly changing software, while also being so sure that no one is getting automated away from this work. Consolidated—yes. It may be barebones, but I just shipped app in a week, while working a full-time job.

Choosing a tech stack

This app is written in the (mostly) default “web stack” of the day:

• Next.js
• TypeScript/JavaScript
• Tailwind
• Convex
• Clerk
• Vercel

I feel choosing a highly popular tech stack enables a serious amount of velocity for this AI-driven development. TypeScript is a fine language, and I enjoy working in React/Next.js.

What I still had to "work" for

• Design choices—turning prompts into HTML seems easy; producing a compelling UI/UX remains difficult
• Maintaining quality—through code review and QA checks
• Be ready to edit code—very rare, but there were still a few times I had to manually take over because Codex just wasn’t solving my issue.
• Domain knowledge of tech stack—while I wasn’t “writing” most of the code, there was still a huge value in knowing what to expect from the code.
• Production/Deployment issues—Codex was a serious help in debugging and troubleshooting issues, but it’s clear we’re not ready to trust LLMs to do this work for us yet.