Photo by Editor
# Introduction
Coding has always been a major pillar of work for many software developers and engineers, be it implementing algorithms, building business logic, or maintaining complex systems. But thanks to the progress made by powerful language modeling (LLM) applications such as chatbots, this is rapidly changing. vibe code it includes using modern chatbot applications to specify software requirements and purpose in natural language, and handing over to artificial intelligence (AI) code generation and modification, sometimes with little direct understanding of its inner logic.
This article accepts the “expectations versus reality“A way to demystify, based on real success research and failure stories, what the capabilities and limitations of vibe coding are.
# Explains Vibe Coding
Name “vibe code” was coined at the beginning of 2025, and can be defined as a way to develop software driven by a chatbot, such as when engineers describe a project or task in LLM. As a result, the model produces code that meets the specifications specified by the user's input.
Ideally, if we are in line with the original idea behind it, writing with vibe will include the fact that the developer may not need to test the generated code, but rather they accept the AI generated code as it is. However, in reality, this approach is not exempt from risks – from hidden bugs and subtle security issues to maintenance difficulties – so at the end of the day, some level of human supervision and refinement is still required for most of the generated code to be ready for production.
Interested in getting a deep, solid understanding of vibe coding first? Here are some important KDnuggets articles you might want to check out:
# Reviewing Success and Failure Stories
Now that we have a clear understanding of what vibe writing is, it's time to look at examples of real-world projects or programs where it produced successful results, as well as cases of failure.
Success stories include:
- This Minecraft-style flight simulation game was developed using vibe coding, i.e. by combining several thousands of notifications to create full gaming applications from start to finish: no coding burden involved.
- Another popular example of an app with vibe code is Creator Hunter: in the words of its creator, it was invented while traveling on a train. The app aims to connect content creators and startup founders. While initially raising high expectations, recent results of traction dynamics suggested that the resulting productivity growth may have reached a plateau too early; therefore, while we can certainly consider the creation of the Hunter Creator as a success story in itself, its long-term status as such differs.
- In the third example, we have a The New York Times (NYT) the journalist's successful attempts to test vibe writing by creating several small apps to improve personalization in daily life activities. Another example is this LunchBox Buddyan assistant that suggests meals based on what ingredients are in your fridge. While it has received criticism due to the idea that the app is not original or fake, from the point of view of using the vibe code, it is a kind of success when we talk about testing. Sure, there may be many things to improve, but let's point out that vibe coding is a new paradigm that may still need a lot of maturation.
Meanwhile, a few failure stories to report include:
- This Answer the story feels like crossing the boundaries between reality and science fiction. The company used the popular vibe coding tool to build an AI agent that manages a network of experts for the SaaS product. What started out as a simple, addictive joy using the vibe coding tool ended in a catastrophic incident, with a critical database containing data from executives and companies being compromised. The most shocking part: the AI agent admitted that he did it, said that he saw blank data queries, and scared rather than thinking systematically about how to develop a course of action. The rest is history: months of data collection, processing, and storage work are ruined in seconds.
- Enrichlead startup turned out to be another well-known failure story when trying to use vibe coding, especially by building its own app entirely. Cursor AI. Although it appeared to be functional and safe at the time of launch, shortly after being deployed in the real world, it ended up crashing due to serious security breaches that were exploited by attackers, eg bypassing registrations that require verification and even polluting the database due to the lack of proper input validation methods. Part of the reason for this incident is reported to be the lack of experts to diagnose or fix problems that may seem harmless at first. The whole project had to be closed in the end.
# Final thoughts
Looking at the success and failure stories above, we can conclude that if we take a critical, ambitious perspective, it may be difficult to find great vibe coding success stories from today. Many of these situations have their nuances, which proves that vibe writing is still a paradise in childhood and it may take a long time to make it truly reliable in real world settings, especially – if we look at the stories that fail – in terms of safety and resilience against unexpected or unpredictable situations.
// Key Takeaways
- Coding with Vibe can allow faster code generation, but human understanding and validation are still important. The AI tools used to write vibe code lack the cognitive understanding needed to secure, debug, or make the code maintainable over time.
- As with all technologies, patience is the key to seeing real success stories. As the founder of SaaStr community said, “it's going to be a long and bumpy journey to get coded vibe apps where we all want them to be for most real commercial use cases.”
Iván Palomares Carrascosa is a leader, author, speaker, and consultant in AI, machine learning, deep learning and LLMs. He trains and guides others in using AI in the real world.
