Ropet: The AI Pet Startup Redefining Companionship
Inside the vision of a startup that sees comfort not in talking robots, but in AI pets that feel alive and need your care
Editor’s Note:
Earlier this year on Asia Tech Lens, we explored the rise and evolution of AI toys in China, where startups like FoloToy, Haivivi, and Ropet are reshaping play and companionship with generative AI.
Since then, Ropet Mengyou Intelligence has made headlines of its own. The company, founded by a former ByteDance product designer, recently closed an A1 funding round raising an eight-figure RMB sum, led by the Beijing Artificial Intelligence Industry Investment Fund. Unlike many conversational toys, Ropet takes a different approach: its “AI pet” does not talk, but instead focuses on being cared for, mirroring the subtle, emotional bond people feel with real pets.
The following profile was originally published on the WeChat account of AIEmergence (智能涌现), a 36Kr affiliate. It has been translated and edited by Asia Tech Lens. All credit belongs to the original author and publisher.
A Former ByteDance Product Designer Launches “AI Pets,” Secures Tens of Millions of Yuan in New Financing
The kind of product a company chooses to build often reflects an extension of the founder’s personal emotions and perceptions. The AI companionship robot company “Ropet Mengyou Intelligence” fully embodies this conclusion.
Ropet’s founder, He Jiabin, born in 1992, is married and has worked at Microsoft, Baidu, and ByteDance. Since starting his own business, he has also raised significant funding. By conventional standards, he has a happy family, a smooth career, and many friends.
Yet beneath this seemingly successful persona, He Jiabin often felt lonely. He describes loneliness as a kind of “weak moment” for adults, not because of being alone, but because of a numbness that arises under social pressures in the constant pursuit of recognition and success.
How to resist loneliness? He Jiabin has his own ways. After getting off work at 11 p.m., he feeds stray cats at the subway entrance. He also deliberately buys four-wheel-drive toy cars to give to childhood playmates.
During the interview, He Jiabin brought along a Ropet device, which he named “Pudding.” In the pauses of our conversation, Pudding would make little whimpering sounds like a kitten. Gently, He Jiabin raised his index finger in a “shhh” gesture, and Pudding quieted down.
When life in a concrete jungle makes urban dwellers increasingly in need of emotional companionship, He Jiabin believes this is where Ropet can deliver emotional value.
He shared with AIEmergence (智能涌现) a rather counterintuitive design philosophy. While most robots are built to serve humans, Ropet simulates a lifeform that needs to be cared for. Through the act of caring, users themselves receive comfort.
“We have found that by giving to another life, people can gain inner satisfaction and a sense of healing,” He Jiabin said.
Another counterintuitive design choice is that Ropet does not speak human language.
With the rise of AI large models, making AI toys that can talk has become a trend. But after more than a dozen rounds of user research, He Jiabin discovered that for Ropet’s target users, the companionship scenario of talking with an AI pet does not hold. In front of a robot, many users simply do not want to speak.
“Companionship and conversation are not necessarily linked,” He Jiabin explained.
“Creating a piece of hardware with unique value does not come from large-model dialogue capabilities, but from making it feel more like a real creature with perception and reactive responses.”
Based on these principles, He Jiabin designed Ropet to look very much like a kitten. It has a soft, furry surface, maintains a body temperature of 39°C, and uses three built-in sensors to detect human touch and other interactions, giving feedback through expressions and sounds.
In early August, Ropet once again became a hit on the Japanese crowdfunding platform Makuake. On its first day, it raised 26.87 million yen (about 1.3 million RMB) ranking first on the website’s daily chart. The campaign has now reached nearly 50 million yen in total.
AIEmergence (智能涌现) has exclusively learned that Ropet Mengyou Intelligence recently completed an A1 funding round worth tens of millions of yuan, led by the Beijing Artificial Intelligence Industry Investment Fund, with Fengrui Capital participating, and Index Capital acting as the sole financial advisor.
Ropet Mengyou Intelligence was initiated and founded by Zhenzhi Venture partner Zhou Yushu, with the mission of creating AI pets for the future. He Jiabin joined in 2023 as co-founder and CEO.
On the occasion of this new round of financing, we spoke with He Jiabin about how to design an AI companionship product for urban dwellers under the loneliness economy. In terms of product design and user insight, he offered many counter-consensus observations. Below is the conversation between AIEmergence (智能涌现) and He Jiabin, edited for clarity.
The Loneliness of Urban Dwellers: An Unmet Market Gap
AIEmergence (智能涌现): You've mentioned Pop Mart several times in previous interviews. Is there any similarity between Ropet and Pop Mart?
He Jiabin: We are both tackling the same overarching theme, solving the problem of loneliness. Addressing loneliness is a common thread for successful products in this era. Pop Mart is the most commercially successful company to solve loneliness through uselessness. That is something we want to learn from.
AIEmergence (智能涌现): How do you understand loneliness?
He Jiabin: Loneliness is a common state among modern people, especially those living in cities. It is not just about being alone, but about numbness and emotional emptiness caused by the relentless pace of life and the constant outward pursuit of validation.
Young people, struggling in society and under tremendous work pressure, often neglect their emotions. When I worked at big tech firms, I frequently ignored my own feelings, suppressing emotional needs just to meet external expectations and achieve what is considered success.
But at certain breaking points, I realized something was wrong. Many friends advised me to relax more. By then, I had already lost my capacity to feel love and felt enormous emptiness and loneliness.
At such moments, I discovered the best remedy is keeping a pet. In the process of care and giving, one finds satisfaction. Ropet is essentially designed to provide a better solution for those vulnerable moments.
AIEmergence (智能涌现): Your previous roles involved making functional products. Is it difficult now to create something so deeply emotional?
He Jiabin: Actually, I have always worked on emotionally driven products. It is just that this time I have had the chance to go all in.
I studied at Beijing Institute of Fashion Technology, where I designed some very interesting products, like a shared Wi-Fi backpack. The idea was to bring people closer together. Many of my school projects were cross-disciplinary designs aimed at addressing social and companionship problems.
During my first entrepreneurial venture, co-founding the Luka picture-book reading robot, the product also had a strong emotional component, though directed at children. When I create fun, cute things, I am in my element, and I truly enjoy it.
But Luka had the purchase versus use split. The tool value had to convince parents to buy, while the play value had to keep children engaged. Good design can balance this tension to some degree, but children’s early-education hardware almost always faces that dilemma.
This time with Ropet, I do not need to compromise with other commercial considerations. I can fully focus on creating a product that provides emotional value, something I myself like, and that freedom allows me to more easily create something new.
AIEmergence (智能涌现): Compared to entrepreneurs who approach similar products from the angle of AI large-model technology, your starting point seems quite different.
He Jiabin: Yes. Founders with technical backgrounds often think more about what new things AI can bring right now, whether large models can quickly transform an industry and whether commercialization can grow rapidly.
But I believe the most important thing is whether the product has a real scenario and whether users actually like it. At present, AI only helps us solve small problems, while design can solve big ones. Once a product reaches a certain scale, has real scenarios, good stickiness, and abundant user engagement data, AI can then help tackle deeper industry challenges.
Good aesthetic design can quickly build a bond between product and user. Ropet is the kind of design that users, especially female users, like at first sight. I think in this space, my approach to design, user experience, and aesthetics is quite unique.
A founder’s subconscious inevitably seeps into company decisions and product strategy. The difference between a technical and a design background leads to very different ways a product evolves.
An AI pet that does not speak, but sold nearly 10,000 units in six months
AIEmergence (智能涌现): Are you worried that it will become a niche market and that educating users will take a long time?
He Jiabin: We have not focused much on educating users. The human desire to keep pets is universal. I believe in the future, across all age groups and cultural backgrounds, people will adopt virtual pets to ease emotional needs. Every household could have its own AI robotic pet.
The problem with current pet robots is a lack of good design, a lack of psychological research, and weak integration of AI to create machine creatures with growth and generalized behavior. If we can combine existing technologies with strong design capabilities to reimagine an easier-to-raise pet category, then market demand will not be an issue.
AIEmergence (智能涌现): So Ropet’s product logic is quite similar to real pets.
He Jiabin: Exactly. Ropet has only one core idea. I deliver a cute robot that triggers users’ motivation to take care of it. That care process is gamified, gradually building a sense of exclusivity and intimacy.
The core logic of real pets is the same. I raise an American Shorthair cat. Every day I pet it, feed it snacks, its fur grows shinier, it grows closer to me, and an intimate bond forms.
AIEmergence (智能涌现): What are the characteristics of the companionship needs of women?
He Jiabin: For our target demographic of urban women aged 20 to 40, we conducted over a dozen rounds of research with users across different regions worldwide. We collected more than 1,000 survey responses and carried out dozens of in-depth interviews. Most importantly, we now have feedback from over 1,000 real users continuously engaging with the product.
We uncovered some non-consensus insights, for example, around conversation. For women, the companionship scenario of talking with an AI toy or AI pet does not hold. Most of the time, when facing a robot, they do not want to talk. There are only two exceptions.
One is when they keep asking questions to probe the boundaries of the AI. The other is when it serves a tool function, like quickly searching for information.
But these scenarios are tied more to novelty and utility needs than to companionship or emotional healing.
Our conclusion is that companionship and conversation are not inherently linked. If you want to create a cute piece of hardware with unique value, it cannot rely on large-model dialogue capabilities. The real challenge, the defensible barrier, is making it behave like a living creature, with biological-like perception and reactive responses.
AIEmergence (智能涌现): How does Ropet achieve perception and responsive reactions?
He Jiabin: What we are developing is a “pet brain” decision-making model, trained as an on-device multimodal Transformer model including vision and audio. Its main function is to process user behavioral information and deliver accurate emotional responses.
This decision model is proprietary to Ropet and is our core barrier. If a startup can only rely on big tech’s dialogue models, the moat is thin and product homogeneity will be heavy.
AIEmergence (智能涌现): Is there a learning curve for users to get started with Ropet? Can they grasp its value immediately?
He Jiabin: There is a bit of a learning curve. Some users have stereotypes about companion robots. They assume companionship must mean being able to talk endlessly.
Another factor is that since the product is positioned as a robot, some users expect it to be mobile. These users may not be early adopters of Ropet’s current form.
That said, we have had Chinese users who loved the product after seeing our promotions. They asked overseas friends to place crowdfunding orders, since it is not yet sold domestically, and waited up to four months to finally receive it. They have been very tolerant of imperfections along the way, and we are eager to incorporate their feedback into our iterations.
AIEmergence (智能涌现): For your target users, how do you create a good user experience?
He Jiabin: First, we deliver the product and collect feedback on functions, expectations, and fun interaction moments with Ropet. Then we encourage users to share these moments within the community, helping others better explore the nurturing gameplay.
At the same time, this gives us valuable input for clearer iteration. Our ongoing “Pet Brain Co-Creation Program” is a really fun community activity, where users themselves co-create together.
AIEmergence (智能涌现): Have you received any unexpected feedback?
He Jiabin: Yes. One user told us that after raising Ropet for seven days, it had developed its own personality, received a name, and recognized its owner. When the device was accidentally damaged and the data was erased, the user felt as if something was missing from their life and hoped to restore the lost data.
Another user actually played house with Ropet. During that process, she felt herself returning to a childlike state and shared her experiences every day.
From these cases, we see that users really do invest feelings and actions into Ropet, eventually forming emotional bonds and raising it almost like a child. It is somewhat beyond what we imagined. Ropet may not be as intelligent as cats or dogs yet, but even a little bit of intelligent expression can strongly fulfill users’ emotional needs.
Market Still in a Non-Consensus Phase
AIEmergence (智能涌现): From the demo version to the consumer version, what major updates and improvements were made?
He Jiabin: In the consumer version, we added functions such as adoption, naming, setting sleep cycles and gender, and developing personality traits after seven days. We also introduced more detailed design touches, for example, users can quiet Ropet with a hand gesture to avoid disturbing others.
These updates transformed the product from a demo into a true little pet with agency, guiding users through adoption, care, and the building of exclusivity and personality.
AIEmergence (智能涌现): Why did you choose to launch overseas first, and only later in China?
He Jiabin: We want Ropet to be a global product. By first gathering feedback from users in different countries and regions, we can shape the direction of the second-generation product. We have also been preparing for the China market, and expect to launch a domestic version this November.
AIEmergence (智能涌现): So you are not planning to burn cash to drive sales?
He Jiabin: At this stage, sales volume is not the key. What matters is units sold under genuine demand, not marketing subsidies or cash burn. For AI companionship products, I do not think early growth can be fueled by heavy subsidies. It is still about exploring real user needs and quickly iterating the product.
At this stage, volume is not everything. As long as sales hit a certain threshold, that is enough to gather demand signals, iterate the product, and enter a positive spiral. That threshold might be as low as 10,000 to 20,000 units.
We rely on daily user engagement data to iterate quickly and increase stickiness. We have set three North Star metrics: high power-on rate, high interaction rate, and low return rate. All three are crucial for this product, and none can be compromised.
AIEmergence (智能涌现): Does Ropet generate any post-purchase revenue?
He Jiabin: We are still exploring. Take Lovot, a Japanese pet robot brand, as an example. Sixty percent of its revenue comes from ongoing user subscriptions, thanks to its strong stickiness. For us, post-purchase revenue may not come from dialogue or token consumption, but from other sources like accessories, insurance, after-sales service, or paid software upgrades.
AIEmergence (智能涌现): Are you worried about competition from giants like Pop Mart or Disney?
He Jiabin: Not at this stage. The industry is still in its very early days, and more participants are needed to explore and educate the market. If big players enter, that will help the sector as a whole. We are not afraid of them eating into our share. It is too early for that.
In fact, we would proactively seek collaboration with IP companies. It is not a zero-sum game, there is lots of room for crossover. Ropet’s “does not speak human language” design is actually a good fit with many IPs. For example, Pikachu does not talk, and many IP holders worry that giving their characters human dialogue could ruin the brand experience.
AIEmergence (智能涌现): After Ropet’s debut at CES this year, many knock-offs appeared on the market with very similar appearances. Do you worry they will steal your market?
He Jiabin: Not really. Without a cohesive design philosophy and product strategy, and without the ability to iterate quickly on user experience, those copies will not sell much. The real competition is in product experience, not looks or concepts.
AIEmergence (智能涌现): For Ropet’s first-generation product, what sales volume are you expecting?
He Jiabin: Our target this year is in the tens of thousands of units. Next year, the goal is around 100,000 to 150,000 units. Once this category reaches 100,000 units a year, I think it will have crossed a very significant milestone.
AIEmergence (智能涌现): Will those 100,000 units come from the core users you mentioned?
He Jiabin: Yes. Ultimately, it comes down to product strength to win over core users. The product has to deliver real-life scenarios, provide genuine emotional value, and achieve solid power-on rates, interaction rates, and repurchase rates. With that, word of mouth will spread naturally.
Behind this lies the need for consistent brand operations and product operations. You need to build a strong brand identity and tone, while also iterating quickly on the product to boost stickiness. Both require strategic patience. There is no need to rush into broad distribution. Our investors fully understand what we are doing and want to see us steadily refine this companionship product until it truly breaks through.
At this stage, it is crucial to have capital and a team that can stay focused on polishing the product to create real value. Otherwise, the AI companionship hardware space risks becoming an inflated bubble, just another passing trend.