Vibe Coding Is Dead How To Actually Make Ai Tools That Unite Ai

Vibe coding - Part 2: Is the design process dead? Plus, why we should trust the effects of our work more than artefacts, rituals or tools. In this note, you will read about changes that are happening in the product building process, especially in the area of design influenced by AI. I will answer why we should stay open to new ideas, understand others’ perspectives, but at the same time keep critical thinking and evaluate if what they say makes sense.

Times are changing, that is certain, but not necessarily in the dramatic way the hottest headlines want us to believe. “Design process is dead”: Is that true? More and more articles around the internet tell us that the classic double diamond design process is dead. They mean the classic workflow where you start with research, a diverge phase, synthesis, and UI design production that gets turned into prototypes and tested. The main argument behind it is that the process is long, time-consuming, and costly.

The next one is that many redundant rituals and artefacts are produced (like personas, journey maps, research activities, and all kinds of design thinking workshops). These evangelists claim that designers or product builders should primarily use AI code generation for rapid execution by creating prototypes without needing to explore UI in classic design canvas tools. Is that true? Only partially. The most valid point is about new possibilities: designers are now able to show how a product should work, not just draw how it would look.

Second valid point is that with AI we may increase the pace of product building process a lot. However, I also detect many biases and weak points in the “design process is dead” way of thinking. They are especially dangerous if they spread without further reflection. Companies that adopt that concept without deeper reflection will not only burn money, but also drive talented creatives away, release mediocre products, and lose customers in the long run.

Here is what you should pay attention to when you hear bold slogans about AI killing design: 1. Conflict of interest First, you need to realize that many of the designers evangelizing this new process are actually somehow connected to the companies building these AI tools. It is clear that they are at least a bit biased, and certainly enthusiastic about the tools they help create.

What’s more, these companies want you to use their products more, so convincing you that you should start your exploration with dozens of AI generated ideas is in their interest. You buy tokens, they make a profit. This series of random generation approach reminds me of a casino, where you pull the handle and get random numbers. From time to time, you get a quick dopamine hit because some of the results give you a small reward, but your wallet will empty sooner than you get a meaningful result.

I believe there is a more thoughtful way of creating with AI, which you will read about in the next paragraphs. 2. Clickbait People just want to grab your attention. Shocking ideas, dramatic shifts, and disruptions always get many views. This is why bold titles appear more and more. I wrote in the “Getting out of information bias” note that negatively framed news headlines have grown increasingly popular over the past decade.

Some posts or short videos just show “the proof”: a nice looking website that, according to their authors, was generated with a few prompts without opening a classic design tool. And yes, AI can do impressive things right now, but that is not the point. Those examples are usually demos, not production apps or sites. Even when published, we don’t know how the solution is performing. I would like to see the numbers behind products built entirely in this “new” way. 3. Velocity as the holy grail Build faster.

Ship faster. Faster is better. Faster is cheaper. This is the illusion we are being fed lately. With that assumption, many product builders often forget the key point. They create the app 10 times faster, but the solution they built is not solving any meaningful problem. I already explained this in the “Climbing the right mountain” note: the key foundation of productivity is not about speed, but direction. It is not a coincidence that I mentioned it here, because the idea came from Sam Altman, OpenAI CEO.

What is more, very often the most meaningful discoveries or ideas need time to grow. This is described incredibly well in “Slow Productivity” by Cal Newport. 4. Lack of innovation Those who say the classic design process is dead often only mean creating products by utilizing existing patterns and components. There is nothing wrong with using proven and stable design systems. They are efficient for business and increase trust in the brand.

But design systems, by definition, are limited, and sometimes you need to expand them with new elements or patterns. A process where AI builds the blocks will not deliver new creative ways to solve user problems and improve business performance, at least not yet. Very often, innovation happens when you talk with each other, sketch something quickly on paper, or explore various directions on a digital canvas. 5. Limited creativity Design is a mix of analytical and creative processes.

Creativity in general consists of two phases: convergent thinking and divergent thinking (this is the reason design process visualization has double diamond shape). The moment where we explore and then synthesize ideas and concepts. Even when you are working within a specific design system and can’t introduce new patterns, combinations of the elements you design may work and feel completely different. Vibe coding, in its current state, is a very linear process.

With existing AI coding tools, it is harder to explore variants since the flow itself does not support exploration. Now that the enthusiasm around AI driven product building has cooled down, let’s look at how the design process is actually changing. The “universal” design process in practice I am not here to say the design process will stay as it is. It has been changing for decades, and we are approaching a time of accelerated change. Design process is not dead, but it is adapting.

As I mentioned, new possibilities are hard to ignore; we just need to use them in a way that makes sense. However, the key thing about design process will remain the same: Design is a thoughtful way of creation, aiming to solve the right problem in the right way. Design is not art in terms of just expressing yourself. Design is to serve people, both users and the business, by building usable, delightful things.

One of the lessons I learned while helping dozens of companies with various team setups is that in practice there is no such thing as a universal design process. Every company is different, and even within an organization, teams’ needs and preferences differ. Sometimes research can be completed through in-depth interviews with stakeholders who have deep domain knowledge and real data about their users. In other cases, you need to understand users through a series of foundational research activities.

Some teams have a major design system you can start working within; others want you to create the look and feel that will form the foundation of their library. I have worked with teams that prepare extensive research reports shared and analyzed across the organization. For others, a quick board with sticky notes is enough to capture user insights. It was never about mindlessly running “how might we’s” and “crazy 8s,” then producing extensive personas or multilevel customer journeys. Those activities serve a purpose: to clarify ideas and identify problems.

They should be used when needed. The resulting artifacts, which AI design evangelists say “users would never see,” were needed to document findings and keep teams focused on what matters when a stakeholder brings in a completely new idea. The number of these assets and their level of complexity depends heavily on the team, not the tools designers prefer to use. I am a strong advocate of the lean approach to product building, but sometimes the basics need to be done thoroughly so the team can move much faster later.

In the end, it is about the outcome, not the tool or activity you used. Leadership will not ask whether you created a persona for a feature, but they will be interested in performance and user reactions. The proper use of artifacts should help you streamline decisions and build confidence in the strategy. They should not be just a mindless ritual completed before going live.

This is why every time a designer approaches a problem, they need to consider the context of the team, the maturity, the knowledge, and the culture of the organization. The number of artifacts and exercises needed is not about a “universal design process,” but about the result you are able to deliver. With artificial intelligence, we may become leaner and smarter in these areas. Let’s now look at how those activities are changing with AI.

Updates to design process: My observations Our work is, and will continue to be, AI supported - that needs to be said. Companies increasingly expect teams to leverage automation and large language models to “achieve more” in the same amount of time. Personally, I would encourage us to achieve “better things” in the same time, but we need to stay anchored in the financial reality of many organizations.

The key thing I see in the updated design process is that the essence, strategic decisions, and intentional design choices will remain driven by human designers. Research is already supported by AI, which can summarize meetings or even conduct sessions right now. The data collected and clustered by the model should then be analyzed, and decisions should be made by the designers and the team. How? Depending on what works best for you and your teammates.

When it comes to creating a solution, you may ask AI to generate solutions to the problems you identified, but the results will likely be only “good enough”, mediocre outcomes. That is your task to raise the bar here. Once you have those standard solutions, think about what alternative approaches you could explore and how the AI output could be improved. How will this creation happen? We are certainly moving toward artifacts that are prototypes or even real coded parts of a solution.

Bringing things to life is faster and cheaper than ever. The barrier to entry for product creation is now very low. The biggest advantage of this is that the handoff process will be much smoother, as engineers will receive a working coded artefact, whether an app prototype, a feature flow, or a component, that they can review easier or use as the foundation for the final product. However, before that, we still need a tool that encourages exploration.

Pen and paper served that purpose for centuries, then digital canvas tools like Photoshop, Sketch, and finally Figma appeared. I believe the future of design is not in the terminal. It is still very early, and there will be another way. We are already seeing first signs, but we are still waiting for an AI supported tool that figures it out properly. Finally, evaluation of the solution you prototyped.

AI enthusiasts will tell you: “just ship it.” What they mean is that you collect data based on how users actually use the product. This is quantitative research done in production. Many companies already mixed approaches like this before the LLMs hype (see how Meta rolled out different variants of features to specific segments of their user base). With that, You will know what works, which is great, but how do you know why something works? This is where qualitative research comes in.

As I mentioned, sessions can already be summarized or even led by AI, but it will still be your job to catch nuances and confirm insights. Does this AI supported process differ in its key points from the classic approach? I don’t think so. Thanks to faster analysis and production, you may raise the quality bar or iterate more quickly. However, claiming that AI makes the design process dead is like saying electric engines will kill the automotive industry.

It is the change, but more of a progression than a total disruption. The key purpose remains the same. Exercise for next week If you haven’t yet tried building a product with AI, pick up one of the most popular tools like Claude Code or OpenAI Codex. Experiment with them and try creating your portfolio by skipping the traditional canvas. How did it feel? Are you satisfied with the result? What do you feel the process is lacking?

It is important to try and experience these new possibilities, some of which are undoubtedly impressive. On the other hand, I am sure you will also feel that in the current state, something is missing. And the essential purpose of design, intentional creation, will not change.

People Also Asked

• Vibe Coding Is Dead: How to Actually Make AI Tools That ... - Unite.AI
• Vibe Coding Is Dead. Here's What Replaced It (And Why You ... - Medium
• Vibe Coding Is Dead. Here's What Replaced It. - peerlist.io
• Vibe Coding Is Dead. Here's What Replaced It. | Codavyn
• Vibe coding - Part 2: Is the design process dead?
• 18 AI Vibe Coding Tools You Need To Know: Build At The Speed ... - Forbes
• Is Vibe Coding Dead? - Global Tech Council