NFTs (non-fungible tokens) emerged as a frenzy driven by accessibility. Anyone could start a collection and dream of investing millions of dollars. But for many, these were assets with no intrinsic value. Today, a new trend is emerging that claims a space of excessive “hype.” vibe coding.
As generative AI matures, the focus has shifted to programming. Now, advanced tools These allow complex interfaces and functionality to be replicated with simple instructions. A notable example is Abbas Khan, a developer at the Ethereum Foundation, who demonstrated how to create a working clone of a MetaMask wallet using just four. prompt In the Claude 3.5 Sonnet model, a milestone reported by CriptoNoticias.
This means that, in theory, any user could launch their own cryptocurrency wallet without understanding what’s going on inside. This extreme democratization has led to saturation of products lacking real demand, as with many 2021 NFT collections. The similarities are inevitable. From digital works that have lost 99% of their value, like Logan Paul’s famous NFTs, 623,000 USD now worth just 10 USD— Even AI-generated applications are published and quickly forgotten.
What is the origin of the term vibecoding?
This term comes from the vision that: Andrei Karpathyformer head of AI at Tesla. In a February 2025 Twitter post, he describes a process in which a human simply sends the “vibe” or intent of the software to the machine.
In this model, the user acts as the conductor of an artificial intelligence orchestra. AI interprets, Directs in natural language and generates entire technical structure without human intervention.
This trend eliminates the need to learn complex programming languages such as Python and JavaScript. The focus shifts completely to creativity and the ability to give precise instructions and prompts.
By 2021, at the height of the NFT boom, platforms like OpenSea made it easy to create and sell NFTs without moving a single line of code.
The solution no one asked for
The ease with which they can be cloned creates an overabundance of tools and saturates the digital ecosystem. Users launch cryptocurrency wallets and financial apps without understanding the inner workings.
research GitHub It shows that AI-assisted repositories have recently increased by 40%. However, most of these projects do not have frequent updates and end soon after release.
This trend shows that more people are creating solutions. We often see people copying Uber, but is that really what the market needs? Marketing and security aside, it seems like the trend is just to copy the most successful apps in terms of functionality and interface.
80/20 problem
forbes He explains that AI can easily solve the first 80% of applications. This percentage includes only the visual interface and initial basic functionality.
The remaining 20% is the most difficult and expensive task for developers. This final stage covers: Security, scalability, and complex bug fixesessential for the continuation of the project.
Many vibe coding projects attempt to bridge this technical gap and fail. If the AI can’t solve serious structural problems, creators will abandon the app.
The cost of maintaining these tools increases rapidly after the initial release. Without technical knowledge, users cannot manage junk software that generates vibes.
we are still in the early stages
Investors like Arturo Elias Ayub, famous for the show “Shark Tank,” have issued statements saying they will stay away from flash investments that promise to create applications in record time because they may face security issues.
Although the term “vibecoding” was coined a year ago, it still appears to be in its infancy. There hasn’t been a wave of investment in these “solutions” yet. However, it has become so practical that we begin to see advertising campaigns for applications that solve, or rather claim to solve, specific problems, but they are nothing more than pseudo-developers creating solutions to attract investment.
