{"slug": "is-vibecoding-viable-for-business", "title": "Is Vibecoding Viable for Business?", "summary": "A solo freelance developer who quit a software engineering job earned over $200,000 in 10 months building custom software with AI tools, including a budgeting app and a booking platform. However, the developer warns that AI-assisted development remains labor-intensive, with most time spent on client communication and problem-solving, and that hourly rates lag behind a typical tech salary.", "body_md": "# Is vibecoding viable for business?\n\n*\n*\n\nAfter working in software engineering and infrastructure for 7 years, I quit my job last year and became a solo freelance developer. I did this primarily due to a mix of anxiety about AI and what it meant for me (exacerbated by working for a software company that was having an enormous identity crisis), and what I'd call a more normal career curiosity having never done it before.\n\nIt's now been ~10 months and I wanted to share my experience.\n\nFirst, you can make money building software with AI. I see lots of claims nobody does anything of value with it. Not true. Here's what I did:\n\n- A \"secure\" infrastructure project. This involved building Microsoft Azure infrastructure and getting it audited to pass a compliance check. This made 120k. I am dubious how much of what we made is being used as it felt like we were checking a box rather than making something useful.\n- A budgeting tool for an events management company. This made 20k with 7k per year support. The company is using this daily.\n- A booking platform for a tour company. We replaced Salesforce and a few other SaaS tools with a custom built platform. This made 45k, with 7k per year support. The company is using this daily.\n- A few other random small projects. A website for a plumber, an invoicing tool for a decorating company. These collectively have made around 10k I'd say.\n\nI am still working on #3, because like the other two I've learned a valuable lesson: I tend to woefully underestimate how much time these would take. I was completely AI pilled at the start, and that's been tempered a lot - I am deluged daily with minor quality problems, missing buttons, confused staff who don't understand how something works, etc. Even with AI, I still need to take time to parse these findings to figure out if it's actually something that needs to be addressed or not.\n\nIf I had to summarise my experience, I'd say this: I drastically overestimated what building with AI would feel like. As it turns out it's still very difficult, laborious, and time consuming. The vast majority of the time is spent is on people problems. I can build the greatest software I can conceive with GPT 5.6 and Fable, but I need to convince my clients it actually solves their problem, and teach them how to use it. Similarly, I often build too fast, and then when I present it to them it's completely wrong or there's an enormous load bearing assumption I've made that is mistaken and means I have to go back to the drawing board. These usually aren't things that can be pre-specified, they are only found once we engage in the solution with real intent, so at least right now I don't see any way to speed that part up or avoid it.\n\nBuilding with AI has revealed to me that human understanding is, and probably always was (though less visibly) a significant bottleneck to getting any work done when you're coordinating human effort.\n\nAll that said, it does seem pretty crazy and awesome that a single developer can build a SaaS that rivals Salesforce and other SaaS tools for a business in a matter of months. *However I caveat this strongly:* I'm not sure I'd try it again as easily. I now have a far greater appreciation for these tools. Domain specific SaaS tools especially: Those things have been battle tested for like a decade across countless customers encountering similar problems, and as the project drags on I realise I'm just rediscovering all the things they solved from first principles. It's painful.\n\nSo is it viable? I dunno, maybe. If you work out my hourly rate over the past 10 months it would be very bad compared to my cushy tech salary. I'm going to try and improve that over the coming months either by working with higher paying customers, or trying to leverage what I've already built (a mini SaaS almost).\n\nIn some ways I feel like I've traded depth of capability for breadth of capability. I've done so much \"more\" with these tools, but I'm also paying the costs of covering so much ground in such a small amount of time. I genuinely don't know if this nets out as an improvement or not, which has caught me off guard given how much more productive I *thought* I would be.\n\nI'm not sure if it'll work out, but I'm going to keep trying. Financially it needs work, but I am having more fun doing this then I ever did working at a large org. Stay tuned!", "url": "https://wpnews.pro/news/is-vibecoding-viable-for-business", "canonical_source": "https://brids.bearblog.dev/is-vibecoding-viable-for-business/", "published_at": "2026-07-13 12:56:11+00:00", "updated_at": "2026-07-13 13:05:06.374435+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-tools", "ai-startups", "developer-tools"], "entities": ["Microsoft Azure", "Salesforce", "GPT"], "alternates": {"html": "https://wpnews.pro/news/is-vibecoding-viable-for-business", "markdown": "https://wpnews.pro/news/is-vibecoding-viable-for-business.md", "text": "https://wpnews.pro/news/is-vibecoding-viable-for-business.txt", "jsonld": "https://wpnews.pro/news/is-vibecoding-viable-for-business.jsonld"}}