AI Is Technology, Not a Product: Why the 'Killer AI Device' Is a Myth

The current discourse surrounding Artificial Intelligence is often dominated by a search for the "killer app" or the "killer device." From AI-powered pins to wearable agents, the industry is obsessed with finding a standalone product that defines the AI era in the same way the iPhone defined the mobile era. However, this perspective may be fundamentally flawed.

As argued by John Gruber in a recent critique of Steven Levy's analysis of Apple, the mistake lies in confusing technology with a product. AI is not a product to be shipped; it is a pervasive technology that will eventually inhabit every product we use.

The Fallacy of the "Killer AI Product"

There is a prevailing narrative that AI will obviate the current smartphone ecosystem. The vision often presented is one of "always-on AI agents" that anticipate our needs—ordering a ride-share before we even step outside a restaurant or managing our schedules without a single tap.

While this makes for a compelling futuristic narrative, it often ignores the physical realities of user experience. Any "always-on agent" still requires hardware: a microphone to hear the command, a speaker to provide feedback, and a screen to verify details. For the foreseeable future, the most efficient piece of hardware for these tasks remains the smartphone.

As Gruber notes, the iPhone didn't succeed because it was a "mobile technology product"; it succeeded because it was a meticulously designed experience that solved real problems. The iPod wasn't about the technology of MP3s or 1.8-inch hard drives; it was about the experience of having your entire music library in your pocket.

AI as Pervasive Infrastructure

To understand the role of AI, it is more helpful to compare it to wireless networking. Apple does not sell a "killer wireless networking product." Instead, Wi-Fi, Bluetooth, and cellular connectivity pervade every single device they make. Wireless networking is a foundational technology that enables the features of the product, but the user doesn't buy the device because it has a wireless chip—they buy it because it allows them to browse the web, stream music, or send messages.

AI is likely to follow the same trajectory. Rather than a single "AI device," we will see AI integrated into existing form factors to enhance specific utilities. This is echoed in community discussions, where users point out that the most valuable AI features are often the most invisible:

"The killer feature of AI for Apple is to finally make Siri work... The user does not like AI for its own sake... The user likes messaging their friends and playing music."

The Commodity Trap and the Ecosystem Play

There is a significant tension between AI labs (like OpenAI and Anthropic) and hardware manufacturers. AI labs are currently attempting to build ecosystems around their models to avoid becoming disposable. However, many observers argue that AI inference will eventually become a commodity—similar to cloud storage or electricity.

If AI becomes a commodity, the value shifts away from the model itself and back toward the integration and the user interface. This is where the "work backwards from the customer experience" philosophy becomes critical. When a company markets a feature as "Powered by AI," it often signals a lack of a clear value proposition. True productization happens when the AI disappears and only the solved problem remains.

Counterpoints: Could the Form Factor Change?

Despite the argument for the persistence of the phone, some argue that the smartphone is not the final form factor. The friction of pulling a device out of a pocket is a genuine pain point that could be solved by smart glasses or advanced wearables.

Furthermore, some suggest that humanoid robots could be the one area where AI is the product. In robotics, the hardware, sensors, and AI model are so tightly coupled that they cannot be separated; the AI is not just a feature of the robot, but the essence of the product itself.

Conclusion

AI is an immense inflection point, but it is not a category of product. The companies that will win the AI era are likely not those that ship the most "AI-centric" devices, but those that use AI to make their existing products more intuitive, invisible, and useful. The goal is not to build a product that makes the user think about AI, but to build a product so good that the user forgets the technology is even there.