Original Summary

I don't think the problem with ai coding is that people will build too much software. I think the problem is that most people are going to build software that shouldn't exist in the first place. They have an idea, see that someone else already built it, make a slightly worse version, add "ai" somewhere in the landing page and call it a SaaS and now the cost of doing that is basically nothing. The problem isn't even that all these products will have competitors. It's that a huge percentage of them don't solve a real problem. They automate something nobody cares about, save five minutes nobody needs saved, or copy an existing product without understanding why customers use it in the first place. Example: someone sees a SaaS doing $20k MRR, thinks "I can build that", spends a weekend with Cursor and ships a clone. Except the original product spent years understanding the problem, talking to customers, fixing bug (with real developers) and figuring out what actually creates value. The second problem is the software itself. A lot of ai products look impressive until you actually use them: authentication breaks, permissions are wrong, data gets duplicated, the main features dont work, database is a mess and nobody really understands what happens when something goes wrong. works on local host, then doesn't work for the final user. And I think we're going to see a lot more of this because AI has removed one of the biggest filters in software: the difficulty of building it. When building something takes months, you have to think pretty hard about whether it's worth building. When it takes three days, you can be wrong for almost free. So if you're thinking about launching a SaaS, I honestly think the answer isn't "build faster", but do the fucking research first. Talk to people who actually have the problem. Understand what they do today, what they've already tried, what they pay for, what they hate about the current solution and what would actually make them switch. Then figure out whether you're creating meaningful value or just moving buttons around in a prettier interface. And once you know the problem is real, build the smallest version that actually solves it, not the smallest demo, not a landing page with five fake features, but a real product with real users, real data, real authentication and infrastructure that doesn't collapse the first time someone does something unexpected. This is basically the idea behind what we're building with Foundel . If you want to test a SaaS idea, you should be able to build it on top of working infrastructure instead of spending weeks rebuilding authentication, databases, billing, permissions and all the boring shit every SaaS needs. If the idea doesn't work, kill it and try again. If it does, keep building, but at least make it work properly. Ai made building software cheap. I think the next advantage will be knowing what the

  • 情报分类：商业与市场研究
  • 分类依据：内容涉及商业、投资或市场动态
  • 信息来源：Reddit · SaaS
  • 发布时间：2026/9/18 02:36:58