Published on August 4, 2026 · The mitza.es team
The term "vibe coding" was coined by Andrej Karpathy, former director of AI at Tesla and one of OpenAI's founders, in February 2025. It describes a radically different way of programming: instead of writing code line by line, you describe in plain language what you want to build, an AI generates the code, and you iterate until it works. A year later, it's already a widespread practice among developers and non-technical people alike.
How it's already being used in Spain
Spain hasn't sat out this shift. Startups and scaleups like Glovo, Cabify and Wallapop already use tools like GitHub Copilot to speed up development of new features. Development agencies in Madrid, Barcelona and Valencia are using tools like Replit or v0 to prototype solutions for clients in a matter of hours instead of weeks. Even banks and large corporations like BBVA and Santander are exploring AI use within their development teams.
What vibe coding is genuinely good at
- Validating a product idea in hours, not weeks.
- Building functional prototypes to show a client or investor.
- Automating simple internal tasks without relying on a technical team.
- Lowering the barrier to entry for people who can't code but have a clear idea.
Where it falls short
A prototype generated with AI is rarely built to scale, to handle real traffic, to meet security requirements, or to integrate robustly with other systems. The generated code itself usually lacks any long-term architecture: it works for the demo, but becomes fragile as soon as the project grows or requirements change.
- Security: AI doesn't always apply good practices around data protection or authentication.
- Performance: quickly generated code usually isn't optimized for real user volume.
- Maintainability: without a clear architecture, every new feature gets harder to add.
- SEO and structure: a "vibe-coded" website is rarely built to rank in search engines.
When it makes sense to move to professional development
The tipping point usually arrives when the project stops being a proof of concept and the real business starts depending on it: when you need to handle customer data, process payments, guarantee load times, or rank in search engines to consistently win customers. At that point, the right move isn't to throw away the prototype, but to use it as a requirements baseline for building something solid on top.
Prototyping fast and building well aren't incompatible
The most efficient way to work today combines both: use AI to validate the idea fast, and professional development for the version that will actually carry the business. If you have an AI-built prototype and want to know what it needs to reach production, check out our custom web development service, or tell us about your case in web development.