AI toys are undergoing a harsh process of “eliminating the false and preserving the true”

Recently, the question that I was asked most frequently was like a persistent curse: “Should we enter this AI toy market now?”. Initially, when faced with this question full of longing and anxiety, I even had a moment of hesitation. After all, from any macro data perspective, this seems like a gold rush that is about to erupt: the capabilities of large models are already mature, and they can be accessed at any time like water, electricity, and gas; emotional consumption has been steadily increasing in the post-pandemic era, and people are willing to pay for “healing”; while the traditional toy industry is desperately seeking new growth curves. These three currents converge, easily giving rise to a strong intuition: The opportunity has arrived, we must take off immediately.

However, during this period, as if a diver, I intensively explored the deep waters of this industry, looking at one project after another. A stronger intuition than the initial one emerged instead: This matter is not as simple as it seems. The surface excitement often masks the underlying crisis.

If you are in this circle, the intense atmosphere almost burns your skin. New products are launched on a weekly basis, the machines in the factories roar day and night, the containers of cross-border logistics come and go continuously, and the data of the first batch of orders seems quite impressive. The entire industry seems to have crossed the initial stage of “can it be done” and entered the practical cycle of “who can sell the goods”. This seems like an era of abundant gold, but if you are willing to squat down and wipe off the foam on the data, you will see an extremely bone-chilling reality: Many products can successfully ship the first batch of goods, but it is difficult to have the second and third batches. In today’s technical conditions, shipping is not difficult; the hard part is “continuity”. When the goods in the warehouse are removed, it does not mean the victory of the transaction, but the beginning of the test of “relationships”. If the product is left in the corner after reaching the user, and is abandoned after only three days, then the prosperity of the first shipment is merely accelerating the transfer of inventory from the factory to the landfill. The excitement is for the industry, while the inventory and bad debts are for oneself.

Why does this “peak at entry” situation occur? Because the vast majority of current entrants are “making products based on the excitement” rather than “making products based on the demand”. Their logical chain seems flawless: AI is very popular, so I must incorporate it; plush toys sell well, so I make a shell; add a little conversational ability, and the product is all done. This process is logical and conforms to all the imagination about “opportunities”. But the fatal mistake lies in mistaking “industry trends” for “user demands”. It’s like selling umbrellas in a desert because the weather forecast says the rainy season is coming, but ignoring that there is no rain in the desert. The result is that the product “seems” completely correct – it has AI, plush toys, and conversational capabilities, but after the user takes it home, they find that it neither solves loneliness nor provides tool value, and ultimately becomes an expensive “electronic garbage”. Such a product is for investors, not for users to use.

During my conversations with many teams, I discovered an extremely interesting and dangerous phenomenon. When everyone talks about the product, they often focus on showing technical parameters: We used which large model, the voice synthesis is so natural, the memory function can be precise to the second. However, almost no one can clearly and forcefully answer a more essential question: Why do users repeatedly use it? This is the dividing line between “product” and “sample”. A product is established not just because it “can be used”, but because it “will be continuously used”. If you use the simplest test standard to measure: Will users feel worried if they don’t use it for three days? If you put it away, will users feel indifferent? If the answer is affirmative, then this product, from a business perspective, is “unviable”. It may gain temporary attention, but it will never have a lasting life.

The current biggest misconception lies in directly equating “AI capabilities” with “product value”. Many AI toys merely package and encapsulate AI capabilities. These are undoubtedly manifestations of technical capabilities, but the problem is that capabilities do not equal value, nor do they equal users’ motivation for use. Users will not stay because your toy “has AI”, just as users will not be moved by a phone being able to make calls. Users will stay because the product is “interesting”, has established a “certain relationship” with the product, and has developed a “dependence” on the product. AI is merely a tool for establishing these emotional connections, not the connections themselves. When we mistake the tool for the goal, the product loses its soul.

Returning to the original question: “AI toys, this wave of opportunity, should we go for it now? ” My answer is: You can go for it, but don’t rush to act. More accurately, the current problem is not an external judgment of “is the timing right”, but an internal questioning of “do you really understand what you are doing”.

If you are only motivated by seeing others shipping, hearing that someone has received an order, thinking that this is a no-misunderstanding opportunity, and getting excited, then I would actually suggest you to stop immediately. Because decisions driven by FOMO (Fear of Missing Out) emotions often make you the one who takes over. But if you have ruthlessly stripped away all the noise, thought clearly about these things: your user profile is not just a vague “young person”, but a specific person, such as that anxious urban single woman in the night; your product identity is clear, it is either a tool, a pet, or a friend, and an ambiguous identity is a fatal flaw; will there be a closed-loop relationship like “feeding” or “talking out” between the user and it. Then, entering at this time is not an adventure, but a certainty. Because in this turbulent race, the real opportunities do not belong to those who run the fastest, but to those who see the most accurately. When the tide recedes, we will eventually discover that only those products built on real needs can avoid the fate of floating without a net.