Wie kann man im Zeitalter der KI ein gutes Produkt entwickeln?

Quick Overview of Key Points

• Tony Fadell, the so - called "Father of the iPod" and co - founder of the iPhone, recently shared his views on how to develop good products in the era of artificial intelligence in a podcast.
• Fadell shared his product philosophy of the "Three - Generation Rule": In the first generation, the product is developed; in the second, it is refined; and in the third, the business model is refined – no product is perfect on the first try.
• He believes that AI is an amplifier. However, if founders and product managers completely delegate the top - level thinking and system architecture to AI, cognitive problems will arise – the real difference is always made by taste and decisions that AI cannot replace.
• He assesses that the AI's demand for computing power triggers a new wave of infrastructure reconstruction. Pure software has no protective walls; only products that integrate hardware and software have real barriers.

In the era of artificial intelligence, the barriers to entry for product design and development have reached a historical low. Prototypes of many products can be created with a click, and code can be output in large quantities. Anyone can develop a "seemingly good - looking" product in a few days. What does such a change mean in the eyes of experienced product experts? Tony Fadell, a former Apple manager and the so - called "Father of the iPod", recently expressed his views on this in a podcast.

It Takes Three Generations to Get Things Right

Fadell shared a product philosophy he developed over decades – the Three - Generation Rule:

In the first generation, the product is developed; in the second, it is refined; and in the third, the business model is refined. I've never seen anyone get everything right on the first try.

Fadell says that this is an objective rule of innovation. He takes the iPod as an example: The first iPod was released in 2001 and only supported Mac at that time. The target audience was only a few Apple enthusiasts. The second generation was the same. At that time, Apple was on the verge of bankruptcy. It was only in the third generation that the iPod was compatible with Windows, and the iTunes music store was launched simultaneously. Only then was the entire ecosystem truly complete – a hardware product completely changed the landscape of the music industry through a system restructuring at the software level. The development of the iPhone is almost identical: The first generation only supported AT&T and had a 2.5G network. It was only in the third generation that the profit structure and stability were truly formed.

The Three - Generation Rule solves the problem of "how to proceed". But at an earlier stage, there is an even more fundamental question without an answer: Where should the first step be taken?

Fadell recounted the long internal debate during the development of the first iPhone in the show – virtual keyboard or physical keyboard. The team conducted tests for several months and compared typing speed and error rate. The data never led to a conclusive result. At that time, BlackBerry only had a 1% - 2% market share in the mobile phone market, which means that 98% of users had never been in contact with a physical keyboard. Surveys can tell you what users are used to now, but not what they will accept in the future.

Finally, Steve Jobs decided that the iPhone should have a virtual keyboard. And in Fadell's view, this is the necessary form of a "Opinion - Based Decision":

When you develop a 1.0 product, a completely new product category, you have almost no historical data to refer to. You need one or two people with taste and judgment to make decisions – they must have the imagination to create a product from a blank sheet of paper.

Fadell says that data is well - suited to optimize known things but is unable to define unknown categories. Revolutionary innovations occur outside the boundaries of data and require the "Informed Instinct" that is accumulated through a long time in the user environment. This instinct is not a talent but is honed – by constantly validating with a large number of prototypes, by telling the product story to people who don't understand technology at all, and constantly calibrating the story based on their confusion and resonance, one can act decisively at a certain critical moment.

To find the direction worth pursuing, Fadell has a simple underlying formula: Long - standing problems plus newly matured technologies, but one must also note that "technology serves the user, not the other way around". He also added that both "iPod + iTunes" and "iPhone + App Store" follow the same line of thinking. One cannot attack from a single point but must build a complete ecosystem around the core product.

He believes that this way of thinking has new significance in the era of AI. Since the tools have become powerful enough, the biggest problem is no longer the lack of technology but that many people voluntarily give up control over technology.

Where Will AI Ultimately Lead Products?

Fadell says that he supported the integration of AI and hardware earlier than most people, so he is not against AI. What he fears is that founders, product managers, and architects completely delegate the top - level thinking, strategy definition, and system architecture to AI. He believes that this poses cognitive risks.

The speed at which AI gives answers is much higher than the speed at which humans form judgments. In the past, the framework was built first and then filled with tools. Now it's the opposite – AI gives the answer first, and then humans decide whether to accept it or not. This reversal of the order is the problem.

Fadell assesses that AI is well - suited to create prototypes, handle local tasks, and accelerate validation; but architecture design, conclusive opinion - based decisions, judgment, and ethical boundaries must be controlled by humans. He used a direct metaphor: The code written by AI is like "Fast - Fashion": It looks like you can wear it, but it falls apart after one wash. A truly good product is carefully crafted and stands the test of time. AI is an amplifier that enhances the user's judgment – if the judgment has already been outsourced, there is only emptiness left.

In the era of AI, when every team can quickly create prototypes and output code in large quantities with large models, the general AI tool itself cannot form a differentiated barrier. The real difference is made by products that are deeply refined and clearly positioned at the top. If anyone can develop a product, those who can develop good products are truly valuable.

Fadell says that telling stories is one of the ways this judgment is ultimately implemented. The most common mistake technology teams make is only telling "what the product does" but forgetting to explain "why it is important to you". Steve Jobs polished the new story behind "1000 Songs in Your Pocket" every day for two and a half years before the iPhone's release and practiced it repeatedly among friends. By the release day, he had told it thousands of times. "Why" is the core of the story and the connection between the product and people's hearts.

Among all these things, Fadell has also drawn an ethical line that he considers inviolable: Product designers must be responsible for the things they create. They must not deliberately make users addicted to increase traffic and retention, and they must not do things that dissolve real human connections. When Steve Jobs rejected the inclusion of adult content in iTunes, he asked the question: "Do you want your children to grow up in such a world?" This question is still valid in the context of AI products today.

Where will AI ultimately lead products? Fadell made some assessments in the show.

Regarding the form of future AI smartphones, his conclusion is surprisingly "conservative". He believes that future smartphones will still need a screen. So far, the failure of some screen - less devices is not due to positioning problems but to the lack of natural information storage capacity – "If you want to look at a map, do you only want to hear on a headset that you should turn left in 200 meters? You'll say: Shut up, I just want to take a look."

He believes that the real change in the future will be the Reversal of Interaction Priorities. Currently, touch input is the first priority, the keyboard is the second, and voice control is the third. In the future, voice control will be the first priority, the keyboard the second, and touch input the third. But this change will take a long time – users' trust in AI voice control cannot be built in one or two product generations.

Regarding the cycle of "AI +", Fadell says that trends have alternated between software and hardware in the past few decades. Today, the AI's demand for computing power triggers a new wave of hardware infrastructure reconstruction – from end - devices to edge computing, from data centers to new forms of interaction. He believes that currently, pure software has almost no protective walls. Anyone can use Vibe Code, while products that integrate hardware and software are difficult for competitors to copy in a short time.

This article is from the WeChat account "Hong Shan Hui" (ID: Sequoiacap), Author: Hong Shan. 36Kr published it with permission.