
Is it blind following or a choice made after careful consideration?
Cover image generated by AI
Author|LainyaEditor|Yin Guanjiao
In June of this year, the highly popular AI toy “Fuzozo” officially launched on JD.com. This plush accessory has attracted many consumers, and currently, Fuzozo ranks second on JD’s AI toy sales chart, with data from the business observation media “Baobian” indicating that Fuzozo’s monthly sales can exceed 20,000 units.
In September, an AI companion robot in the shape of an accessory, Eiliko, completed its crowdfunding campaign on Kickstarter, garnering support from approximately 4,700 backers and raising about $600,000.
In October, LIVING.AI, a hardware brand focused on the AI companion field, suddenly updated its product promotional images on its official website, introducing a new product that also resembles an accessory.

From left to right: Fuzozo, Eiliko, and LIVING.AI
New product preview image|Image source JD.com, Kickstarter, LIVING.AI
After Fuzozo gained popularity, several AI toy brands launched new products in quick succession, all choosing the accessory form. It is not uncommon for a particular market segment to follow trends, but with such impressive results, we can’t help but wonder why the “accessory” form? Is this a case of market players converging in their choices, or is it a decision made after careful consideration?
Similar path choices? From desktop to personal use
According to Qichacha, the three AI companion accessory products mentioned above come from Shanghai Robopoet, Shenzhen Energize Lab, and Shenzhen LIVING.AI. Coincidentally, all three companies initially chose to launch desktop robots as their first products.
. Robopoet-Fuzozo:
Robopoet was established in 2024, and its first product, Sunday, is officially positioned as an “AI trendy toy,” claiming to support three-in-one scenarios for desktop, ground, and wearable use. Sunday stands about 15cm tall and supports features such as voice chat, facial recognition, long-term memory, and personality development. From the official introduction, compared to its later star product Fuzozo, Sunday lacks the social feature of “tap to add friends” and has significant differences in appearance, but most other settings have been carried over to Fuzozo.

Robopoet’s first product prototype, codenamed “Sunday” (left),
not yet officially sold, now displayed on the official website as a life assistant
Robohelper; later launched Fuzozo (right, initially debutedunder the name “Fuzzoo”)|Image source Robopoet official account
. Energize Lab-Eiliko:
Energize Lab was established in 2019, and before Eiliko launched on Kickstarter, it had another desktop companion robot, Eilik (the new product name simply added an “o” at the end), which raised about $770,000 in 2021. Eilik is equipped with sensors that can recognize user touch actions and provide emotional feedback through facial expressions on a display, supporting conversation, timing, and interactive games, with multiple devices able to interact.
This year, Eiliko launched, and at first glance, it appears to be a smaller version of Eilik, with a height reduction of 5cm. It adds a modular design for outfit changes based on the accessory form. In terms of functionality, it incorporates AI technology support into the conversation feature and enhances social attributes, where different Eiliko units increase intimacy with more frequent encounters.

From left to right: comparison of Eiliko and Eilik,
Eilik, Eiliko. Eiliko also adopts a “once in a lifetime”
pairing mechanism, which cannot be unbound once successfully paired. Its head can be detached from
the body for different combinations.|Image source Kickstarter
. LIVING.AI-HICOCO:
LIVING.AI was established in 2020, and its first product EMO is also a desktop robot that supports conversation, photo-taking, and schedule management functions; equipped with multiple sensors, it can express emotions through a facial display and multi-axis joints in response to user voice commands and touch. As usage time accumulates, EMO’s understanding of user commands also improves.
Later, AIBI was launched, positioned as a pocket robot (which can also be worn with a strap), supporting facial recognition, photography, voice Q&A, and schedule management functions. At this stage, LIVING.AI has clearly been reducing the size of its robots, enhancing their portability. The AI accessory, known to many as fuzzoo, seems to have been explored earlier by LIVING.AI’s pocket robot.
Regarding the recent sudden update of the homepage product image, LIVING.AI indicated in a promotional activity that the new product HICOCO may be a “surprising addition to EMO and AIBI.” Following the trend of previous products shrinking from desktop to pocket size, and based on the preview images released, HICOCO is likely to also take on an accessory form.

On the left, EMO and AIBI, below is the updated product image on the official website.|Image source LIVING.AI
As we observe the unified shift of AI toy manufacturers towards the “accessory” form, the trendy toy attributes are also evident.
Fuzozo and HICOCO have chosen appealing, attractive plush materials, while Robopet’s name change for “Fuzai” coincidentally occurred within two months after Labubu’s explosive popularity; Eiliko supports outfit changes and adopts the typical “blind box” sales model found in trendy toys.
It must be said that manufacturers may be “grasping” the psychology of young players, but the explosive popularity of Pop Mart has indeed shown entrepreneurs the value of emotional engagement.
In terms of audience, Robopoet’s founder Sun Zhaozhi has explicitly stated that Fuzozo’s target audience is young women; the promotional images for Eiliko feature numerous clothing/bag matching elements, and the “once in a lifetime” pairing mechanism clearly positions young users and couples as core audiences. The age range for LIVING.AI’s EMO and AIBI is broader, while the audience for HICOCO remains to be observed after its official release.
Smaller AI companions, a social shift after being “seen”
From the iterative product development of the aforementioned manufacturers, we can see that the basic functional practicality has remained almost unchanged from the early “desktop pets” to the recent “AI accessories.” There is hardly any difference compared to Eilik, which was already crowdfunded in 2021, with the improvements largely benefiting from advancements in AI technology (such as enhanced model capabilities and improved long-term memory).
In other words, although AI toys have been thrust into the spotlight in 2025, they are not fundamentally different from four years ago. The wave of returns faced by many AI toy sales earlier this year also indicates that the experience may not yet meet user expectations.
However, the success of Fuzozo after its launch and the collective following of multiple manufacturers align with the trends we have observed in the AI companion software sector, as everyone is trying to break out of a “misconception” about using AI to accompany humans.
The core change in this wave, the “accessory” form, not only provides young people with a reason to like it due to its trendy attributes but also allows this liking to be shared and seen, thereby increasing the probability of two AI accessory toys encountering each other, transforming the toy’s value from merely being an AI companion to being an AI that accompanies and facilitates social interactions between people.
On social media platforms, many consumers who purchased Fuzozo are looking for “parents” to make friends or share their experiences of meeting offline through Fuzozo. A friend of mine who bought Fuzozo mentioned that, for her, the companionship attribute of Fuzozo currently outweighs its social attribute. The likelihood of using social features also relates to the user’s personality; a more introverted friend expressed that she would not proactively seek out “like-minded” individuals in real life, but if someone asks, she would be happy to try the tap-to-friend feature.

Zhu Xiaohu posted about Fuzozo as a package on social media
and published a friendship post (left); some users purchased both
Fuzozo and AIBI (in the blue box)|Image source Xiaohongshu
On Kickstarter, overseas users also express high expectations for social features. During the crowdfunding phase of Eilik, many users left comments hoping to add social interaction features; when Eiliko was launched, users had high expectations for its intimate social attributes.
At the same time, from Eilik to Eiliko, Energize Lab’s settings have undergone subtle changes. Originally, two Eilik units could interact with each other, with the user being the same person; whereas in Eiliko, both the social features based on “intimacy” and the “once in a lifetime” pairing mechanism typically involve different users.

Comments section for Eilik and Eiliko|Image source Kickstarter
Energize Lab shared that after the release of Eilik, the team found that the usage of Eilik as a desktop robot extended far beyond the desktop, with many users taking Eilik to cafes, beaches, and even hanging it on their travel bags while traveling the world.

The background of Eiliko’s birth|Image source Kickstarter
Moreover, the collective shift of AI toys towards “accessories” seems to facilitate industry advancement and market validation.
Robopoet’s Sunday is not yet officially sold, while Fuzozo is priced at 399 yuan (currently around 300 yuan after discounts for “Double Eleven”), which is considered a highly acceptable price among AI toys. Energize Lab’s Eilik was initially priced at $100 (about 710 yuan) during its early bird phase in 2021, while the early bird price for Eiliko this year was directly halved to $50 (about 355 yuan). Although LIVING.AI’s HICOCO has not yet been officially released, the prices for EMO starting at $279 (about 2000 yuan) and AIBI at $249 (about 1800 yuan) clearly indicate a downward trend in product pricing.
In conclusion
According to the business observation media “Baobian,” Vivian, an AI investment expert with a background in NLP (Natural Language Processing) who has worked at investment firms such as Jingwei and Guangyuan Capital, emphasized that AI companionship is not a single product but should be designed around the lifestyle of target users. Many teams are currently fixated on the word “companionship” without understanding “who needs companionship,” “what kind of companionship,” and “in what scenarios companionship is needed.” This has become a core reason for the difficulty in monetizing AI toy manufacturers.
The toy manufacturers choosing to make “AI accessories” have all experienced a shift from desktop robots to portable devices. This consistent shift actually reflects that AI companion hardware manufacturers are adjusting their thinking, focusing on young users as the core audience, and using the “accessory” form to achieve more frequent and intimate companionship scenarios, but the ultimate goal may still be “using AI to attract people to accompany people.” We will continue to monitor the subsequent developments of AI accessories as they explore new paths.
References:
1. Zhu Xiaohu’s bet on the 300 yuan AI toy must first overcome the “return wave” – Baobian
2. After raising an “AI pet” for a month, I found this business full of bubbles – Silicon-based Research Institute
3. Not making “AI companionship” robots, but raising tens of millions – AI No Quadrant
Data from SimilarWeb, Diandian Data, Semrush, and other third-party platforms may have certain discrepancies with real data and are for reference only.
