Original Summary

I’m a professional writer. Been one since 2009, which in internet years makes me roughly 300. I’ve had an idea for an app rattling around my head for the longest time. Tried learning to code a couple of times but between client work and deadlines it never stuck. My brain just refused to care about semicolons and one word lines (seriously, what's up with that??). Then AI showed up. I described the first version of the idea and it spat out pretty much exactly what I had in my head. That was it, I was hooked. Every time I used it I’d think of another pain point it could solve and we’d go build that too. The idea is simple. As a writer I want to spend my time writing, but the business of writing eats more of my week than the actual words do. Proposals, invoices, chasing clients, keeping track of who owes what. So the app goes after all that busywork. It was EXTREMELY tiring and frustrating at times. There were nights I came within an inch of punching thru my monitor. But oh so gratifying when stuff finally clicked and clacked. I went from n8n in mid 2025 to Base44 to Cursor and now Claude Code, which is basically the vibe coding version of a career ladder. Today I have a fully formed product. Not perfect, but presentable. I showed it to my fellow writers and they liked it, and apart from the first three testers (who found plenty, bless them) nobody has reported anything major. Yes I know “my friends liked it” is the weakest validation on earth, I’m working on the rest. The big lesson for me: AI is fundamentally a verification problem. It works great for stuff you can check yourself. For me that was the functions, buttons, forms and business logic. I can click a button and see if it does the thing. Got that down pat. But I’m not a coder, so I had no clue how to check security, latency or the deeper under the hood bugs. I was basically taking the AI’s word for it, and AI will tell you your code is secure with the same confidence it used to tell you there are two r’s in strawberry. So I brought in a developer. He found some problems in the base code and helped me make it a lot more secure. I do think AI can help people like me bring ideas to life. But it has limits which aren't going anywhere. We’re only limited by our imagination, both in what we can build and in all the ways it might break. That second part matters more imo. Turns out it was my paranoia, not my genius, that made the product better. Also, everyone keeps saying AI will kill software development. I don't buy it. Outside of online echo chambers nobody thinks building your own app is easy or even worth trying. Dentists and plumbers didn't start building their own websites just because Wix and Weebly existed. Most people couldn’t care less. It’s still a ton of work, and that's assuming you have an idea worth building in the first place. Curious if other non technical founders here hit the same wall. How did you handle the stuff you couldnt verify yourself?   submitte

  • 情报分类：工作与职业机会
  • 分类依据：内容涉及招聘、求职或职业发展
  • 信息来源：Reddit · SaaS
  • 发布时间：2026/10/15 15:28:23