Vibe coding won't ship a production app alone, but it changes a developer's job.
A production app built with AI needs a hybrid. The AI writes the repetitive code, and the developer steers it like a product manager. I published this in October 2025, about seven months after the term vibe coding took off. I've been a .NET developer for about 20 years, and I wanted a straight answer on whether you can vibe code an app that's ready for production.
The business case
Vibe coding means letting the AI write the code and going with the flow. That clashes with everything a traditional developer worries about: security, scalability, where the data lives and which policies apply to it.
If you run a business, the pitch sounds like savings. Let the AI do the work, save money and build apps faster. My mission is to build production-grade apps with AI, and the useful question for an entrepreneur is narrower. You need to know the point where you bring in a developer and the point where you can vibe code it yourself.
Chris Kelly made the point in his talk Vibes won't cut it that tractors didn't get rid of farming. That stuck with me, because the news that year was full of tech layoffs blamed on AI, Salesforce among them. I think the debate is more nuanced than a right and a wrong answer.
The term's origin
Andrej Karpathy, a co-founder of OpenAI, coined it in a February 2025 post on X. The idea was to give in to the vibes and let the AI write the code. The post pulled millions of views, and anyone following the news fell into one of two camps. One said this is the future. The other said it won't work for me.
I sat in the second camp for a long time.
Developers hold back
I write real production code in full Visual Studio, over remote desktop every day. In October 2025, Visual Studio 2026 was still the Insiders preview, and I thought it was good. The GitHub Copilot chat in the older version didn't impress me. The 2026 preview's version did.
Developers have always borrowed code. We copied answers from Stack Overflow based on upvotes, and we talked problems through with a rubber duck. AI is a new form of that.
Developers adopt new tech fast, so the reluctance looks odd from outside. It comes from experience. Most developers don't want to break things, because breaking things means staying until 6, 7, 8 or 9 PM to fix the bug. Stakeholders and business people are watching too. Throwing an AI at a random piece of code and letting it rip is a bad idea.
Jump to
- 0:00 Introduction to AI and Vibe Coding
- 0:33 Traditional Development Concerns
- 1:09 Embracing AI in Development
- 4:09 Origins of Vibe Coding
- 8:49 Benefits of Vibe Coding
- 11:44 Challenges and Limitations of AI
- 15:04 Hybrid Approach: AI and Developers
- 17:33 Conclusion and Future Outlook
The full argument is in the video: watch Vibe Coding with AI on YouTube.
AI wins on speed
Speed is the obvious win. AI is great at boilerplate, the repetitive code nobody enjoys writing. If you're an agency building new apps, I'd go all in. Existing apps are the harder case.
Consistency is the second win. You can tell the AI to comment and document every function, and it will. I'm guilty of skipping documentation, and most developers I know are too.
The third is the user interface. AI spins up a UI from your ideas fast and refines it until it looks presentable, and plenty of developers aren't strong at design. UI matters more than developers like to admit. I think it's part of the reason Apple does so well.
Developers still matter
Firing your developers to cut costs because of AI is a mistake. If code has been around for 20 to 40 years and a developer has worked on it for 10 years or more, their intuition on it beats anything AI brings to the project. On scaling and architecture, they usually decide better than an AI, because they take in the company and its history along with the code.
They also know when to bend the rules. It's Friday at 5:30 and something has to go out the door, so every developer bends a rule unless the work is mission critical. If you're testing in a small environment, you don't need the most production-ready system. A vibe-coded app can miss that context and add code noise you'd never have to deal with otherwise.
Developers make trade-offs under pressure too. With a deadline tomorrow morning, a developer cuts features to ship the one that matters. An AI keeps trying to get every single thing done.
Run it as a hybrid
I sit in both seats, business owner one day and developer the next, so I land on a hybrid. The setup I'd use splits the work. The AI handles repetitive code, the first pass at the UI and keeping things clean and organized. The developer owns the business logic and translates requirements, acting more like a product manager.
That shifts which skills matter:
- Standards and documentation, so the AI has clear, repeatable guidelines to follow.
- Testing, so you can verify a large codebase works.
- Clear business requirements. On a big project with complex requirements, a tool like Spec Kit keeps the AI on track, and I wrote up what happened when I ran Spec Kit on a 5,500-line app.
- Code review, because the AI writes far more code, far faster than I can, and my job shifts to critiquing it.
Treat AI like a junior
Don't fight the AI's coding style. Use style guides and write specific, detailed rules to keep it on track, then treat it like a junior developer. Business people can treat their developers like product managers who pass requirements to the junior developer, the AI. The point is to let the AI do the typing while your developers keep the judgment calls.
There's more on this in my post on why vibe coding still needs engineering judgment. If you want help setting up this hybrid for your own team, book a strategy call.
About Dwain
I'm Dwain Browne, Founder of LeadScore AI and SnapSuite. I'm a Toronto-based software architect with 25+ years across software, sales, operations and entrepreneurship. SnapSuite helps contractors and trades teams get paid faster, and LeadScore AI shows you which leads deserve your next call. I also run Vibe Code to Profit, a community helping people build software with AI.