The Rise of Vibe Coding

Recently, the term “vibe coding” has become extremely popular, with many people showcasing their ability to quickly create products using AI. In any AI product community or social circle, it’s common to see posts about someone launching a product overnight or developing a SaaS system over the weekend.

Indeed, AI’s coding capabilities have improved significantly, enabling users to create functional products by clearly describing their requirements. However, this has led to a widespread misconception: many believe that knowing how to use AI to code means they can successfully create products or start their own businesses.

While most can use AI to develop a feature-rich product, there remains a considerable gap to achieving commercial viability. This gap often stems from a lack of understanding of product design logic, which requires professional skills beyond just coding.

A Real Example

I recently encountered a friend’s project. Although he is not a product manager, he managed to create a feature-rich product in a short time using AI. To be honest, the amount of functionality he included would typically take a product manager several months to organize, design, and schedule.

From the perspective of “creating functionality,” AI has indeed increased efficiency dramatically. However, when discussing the product with him, several issues quickly became apparent:

• When I asked who the primary users of the product were, his answer was vague: “It should be useful for many people.”
• I inquired about how users previously solved the problem and why they would switch to his product, but he couldn’t provide a clear answer.
• I asked about the user journey from opening the product to completing core tasks, but he hadn’t thought through the specific usage process.

In essence, while the product has many features, it lacks clarity on “who it is for,” “what urgent needs it addresses,” and “how it is used.”

This situation is akin to a chef using the latest kitchen tools to prepare a visually appealing dish, only to find that the guests actually wanted noodles. No matter how advanced the tools or how quickly the dish is prepared, if the direction is wrong, the meal will still be a failure.

In product development, functionality should naturally result from clarifying these initial questions. Whether I am developing a product myself or leading a team, it is crucial to understand what problem the feature solves for which user, why this feature is necessary, and its role in the business process before moving on to prototyping, reviewing, and delivering functionality to developers.

AI Can Code, But Not Make Product Judgments

It is important to break down the process of “making a product.”

Many people perceive “making a product” as merely the final step in the process: turning ideas into functional code. At this stage, AI can indeed significantly speed up the process and even replace a portion of entry-level development work.

However, before the coding step, there are several critical and often overlooked tasks that together form the true “skeleton” and “soul” of a product:

1. Business Model: Have you clarified the business model? How will this product generate revenue? Is it subscription-based, pay-per-use, or free with conversion strategies? Why would customers pay for this product instead of using free alternatives? If the business model is flawed, even the most impressive features will amount to nothing more than a self-indulgent creation.

2. Target Audience: Have you clearly defined your target audience? Saying “many people can use it” is equivalent to saying “there is no target user.” Is it for operations personnel in small to medium enterprises, or for a specific role in a large corporation? Different audiences require different feature priorities, product language, and pricing strategies. If you can’t articulate “who it is for,” all subsequent designs are mere guesses.

3. Customer Needs and Pain Points: What are the real needs and pain points of customers? This is often confused; believing that a feature is useful does not equate to it being necessary for customers. Many individuals creating products are essentially addressing problems they imagine rather than real issues faced by customers. Features developed without user interviews or validation of pain points, no matter how numerous, are merely “appearing useful.”

4. Prioritization of Needs: Which needs are core necessities, which are enhancements, and which can be deferred to later phases? This judgment must be based on a comprehensive assessment of business goals, resource allocation, and competitive landscape. AI will not make these trade-offs for you; it simply executes whatever you instruct it to do, leading to an ever-expanding list of features while the product becomes increasingly disjointed.

5. Business Logic and Processes: Business logic and processes are often underestimated but are crucial to whether a product can function effectively. How does data flow within the product? What are the relationships between different roles? What steps and validations must a task undergo from initiation to completion? If this chain is not clarified during the design phase, each functional module generated by AI might work in isolation but fail to integrate cohesively, resulting in mismatched fields, incomplete status transitions, and overlooked critical scenarios.

These five areas fundamentally represent the “think clearly before doing” aspect.

They do not depend on coding ability but rather on your depth of understanding of the industry, customers, and business. This is precisely the capability that professional product managers develop over time, and it is an area where most AI tools currently fall short.

The Distance Between “Creating” and “Succeeding”

Returning to my friend’s project, from the perspective of “creating,” he has achieved a speed that many professional teams cannot match. However, from the perspective of “succeeding”—whether the product can be continuously used by customers, generate positive business cycles, and evolve closer to real needs—there is still a significant gap.

This gap will not automatically close as AI capabilities improve.

AI addresses the “how” (how to implement), while product managers must focus on the core questions of “what” and “why” (what to do, why to do it). If the “what” and “why” are incorrect, no matter how quickly or well the “how” is executed, the result will still be an unusable product.

Conclusion

Vibe coding is not a bad thing; on the contrary, it is a gift of this era for anyone looking to create products, lowering the barrier to implementation to an unprecedented low.

However, because “implementation” has become too easy, the ability to “think clearly” has become the truly rare and valuable skill.

If you are also using AI to create products, take a moment to ask yourself a few questions:

• Who is my product for?
• How are they currently solving this problem?
• Why would customers be willing to pay for my product?
• Among the features I’ve developed, which ones are core necessities, and which are merely what I think should exist?
• Can I personally navigate the path from opening the product to completing a task?

If you can answer these questions clearly and thoroughly, AI will be your best accelerator. But if you cannot articulate them, no matter how powerful AI is, it will only amplify a product that is misdirected, potentially leading you further away from your goal.