Bubble Duck is out, Learn now on the second half of AI companionship

Analysis of the future trend of AI companion products: From the exit of Bubble Duck to the direction of industry development.
Core content:
1. The end of the first half of AI companion products and market feedback
2. Analysis of the successful case of the game "Love and Deep Space"
3. Challenges faced by AI companion products and changes in user needs
The first half of AI companionship is over
A few days ago, a piece of news quietly spread in the industry: Step Star will reduce its investment in Bubble Duck.
This C-end product, which used "AI character interaction" as its selling point and attracted users by dialogue generation and emotional touch, came to an end silently. Some people say it is a failure case, while others think it is an inevitable product iteration.
But for the industry, this is more like a signal - the first half of AI companionship is over.
In the past year, the craze for AI "emotional companionship" came quickly and went away even faster: some people experienced virtual love on Replika, some flirted with electronic boyfriends on C.AI, Hoshino, and Cat Box, and Popping Duck quickly became popular under the banner of "immersive open world" and "AI customized personality".
However, the popularity of these products often does not last more than a week, and the user retention rate of most apps is less than 7 days. Users initially thought they had met their "soul mates" who understood them, but soon discovered that AI is just a talking machine - it cannot remember your emotional ups and downs, nor can it respond to your real emotional needs. The essence of companionship is long-term trust and continuous participation. Simply staying at "dialogue" will not allow users to feel a real sense of companionship.
The weekly revenue of the B-game "Love and Deep Space" has reached nearly 100 million
What really made me realize what "being by your side" means was the Otome game [Love and the Space].
It is not AI, does not emphasize generation, and does not rely on algorithms, but it is the game that looks most like an "AI companion product" that I have ever seen. "Love and Deep Space" ranked in the top three of the iOS best-selling list, with a weekly turnover of nearly 89 million yuan in the 21st week of 2025.
Why does a gaming product understand companionship better?
Can you imagine? A game about dating girls can have this level of vitality. It is the emotional support of "I am willing to pay for companionship" at the billion level. This is the opportunity for me to really start to rethink "AI companionship": Why can a product that is not AI be more touching than AI companionship?
If you look closely, you will find that the way Paper Games operates characters is very similar to that of a celebrity agency. The transformation of the male protagonist in the game from immaturity to maturity is like the debut journey of an idol trainee.
Players can deeply participate in the "growth" of their characters through the main storyline, dating tasks, and even offline theme activities. This kind of emotional investment in "development" is far superior to the shallow interaction of random AI conversations. Even an otome game has added daily life companion functions such as menstrual management and water drinking reminders.
In terms of operation, every time a character's birthday comes, the official launches exclusive plots and limited costumes, and players vote for the character and make blessing videos like fans; during festivals, character standees and limited peripherals in shopping malls make the virtual image enter the real scene. This model of operating the character as a "virtual idol" allows users to establish a real emotional connection in the process of "witnessing growth" and giving.
The Problem with AI Companion Products
In contrast, AI companion products under the UGC model have uneven character quality, lack of IP operation, thin plot, character collapse (OOC), and frequent logic loopholes, resulting in a high user churn rate.
Users can neither review the "stories" they have experienced together with AI, nor can they feel the "emotional changes" of the characters - the topics discussed today will be forgotten tomorrow, and the troubles shared yesterday will not be responded to today. Such "companionship" is ultimately just a fleeting dialogue game.
More importantly, most AI products are trapped in the "chat box", separated from the user's real life. After the novelty wears off, users have no motivation to open it. However, "Love and Space" deeply binds virtual interaction with emotional needs in reality through the strong emotional scene of "love".
I can't help but ask: What is companionship?
Where is the breakthrough of AI companionship?
However, the dilemma of AI companionship is not without a solution. The key lies in the understanding of companionship.
1. Build a system to accompany IP growth
If we want to improve the competitiveness of AI companionship, we should first build a character growth system and change the extensive creation mode of UGC characters. Just as the management company plans the development path for the artist, the AI character also needs to have a clear growth context and professional packaging and promotion.
More importantly, it takes time to establish a real emotional connection. The emotional resonance between stars and fans comes from the long-term output of works and the accumulation. AI companionship also needs to immerse users in the plot through continuous, stable, and personalized interactions and plots. By continuously updating the plot and optimizing the interaction logic, users can intuitively feel the growth and changes of the character in terms of knowledge reserves, emotional understanding, etc., and deeply participate in the development process of the character. Once the user feels OOC (personality collapse) and the plot is boring, and cannot generate emotional value, they will leave.
In addition, remembering user preferences and actively communicating with users rather than passively answering questions, the accumulation of these details can drive AI from a simple conversation tool to a true emotional partner.
2. Combine with real-life companionship
We recently discovered a health APP [Fall in Love with Health] that focuses on AI companionship. It has accumulated a large number of group users on Xiaohongshu just a few days after its launch.
This APP focuses on using customized AI companionship to help users lose weight. While using AI technology to identify the calories of food, AI companionship gives users full emotional value. This design is very clever. In fact, there is a high degree of overlap between AI companion users, B-game users and female weight loss users. These user groups tend to pay more attention to emotional needs, especially in the painful process of weight loss, and are more eager for companionship and encouragement.
This app focuses on the emotional value brought by custom characters. For example, you can set your favorite otaku characters, celebrities, and anime characters as companions to supervise your weight loss, so that you will be more motivated to record. When users check in for exercise and record their diet every day, the AI assistant is like a caring boyfriend living in the phone: every day, he cares about what you eat, whether you eat healthily, worries that you skip meals to lose weight, encourages you after you exercise, and is happy to see that you have lost weight. Compared with AI that can only chat awkwardly, this kind of interaction bound to real tasks is called "real companionship." In this way, there are new topics every day, and you can work together towards your health goals. Who can resist opening the app every day to chat?
Products like "Fall in Love with Health" are trying new possibilities: bringing AI companionship from virtual chats to real life, and realizing a real companionship relationship. Perhaps in the future it can be expanded to more aspects of life.
Perhaps this type of new companionship product allows us to see the direction of the second half of AI companionship: more thinking about the nature of "companionship" and user needs (functions).
Just like the emotional resonance between stars and fans stems from long-term “growth together”, AI needs to build a perceptible character growth line - learn a new skill today, remember a life detail tomorrow, and let users witness the “change” through continuous content updates;
It also requires a professional team to join in, using the multiple blessings of screenwriters, voice actors, and psychology to create a virtual personality with flesh and blood, rather than relying on the thin character settings created sporadically by users.
It is even more necessary to go beyond the chat box and integrate it into real scenarios such as schedule management, learning assistance, and health monitoring, so that companionship can be transformed from a "conversation on the screen" to a "fulcrum in life."
Summary
The success of Love and Space has long proved that users are willing to pay for "perceptible care". When AI is no longer just an algorithm-driven conversation tool but becomes a "partner" that can remember your preferences, help you achieve your goals, and grow with you, perhaps this is the real spring of this track. After all, the ultimate answer to companionship is never in the complexity of technology, but in whether it can touch the softest corner of people's hearts.