Six months ago, 36Kr sat down with Guo Renjie, who had just stepped down as executive president of Dreame’s China operations to start a new company: Joyin. At the time, Joyin had yet to produce its first product. But Guo had already secured backing from a lineup of staunch investors, including IDG Capital, Matrix Partners China, ZhenFund, and HongShan.
“My only real advantage is that I’m relentless,” Guo said.
In a follow-up conversation in late May, Joyin had just closed another funding round, this time an angel extension round worth a nine-figure RMB sum, led by Jinqiu Capital. Existing backers Matrix Partners China, Vitalbridge, and Monolith returned, with Lighthouse Capital also participating and serving as exclusive financial advisor.
This marked Joyin’s second funding round in just three months, bringing its total angel financing to nearly RMB 300 million (USD 42 million).
Just as notably, the company publicly revealed the focus of its first product: a compact, multipurpose robot designed for home use.
After six intense months, Guo emerged with a central question:
Why are there many promising technologies, yet few actually achieve product-market fit, with artificial intelligence-powered toys being one of the few categories to break through?
“Good tech isn’t making it into AI hardware,” he said. In his view, the slow march has less to do with technical capability than with accessibility. “Motors, algorithms, models—they are already very capable. They can perceive and understand the environment. But most of that capability stays locked inside massive humanoid robots.”
Instead, today’s large-format robots mostly live in labs and showrooms, priced anywhere from tens to hundreds of thousands of RMB, far from mass market viability.
Guo’s goal is to reverse that trajectory and bring the tech into everyday homes. He’s seen it before: the explosive growth of consumer robot vacuum cleaners only came after LiDAR (light detection and ranging) sensors, originally built for autonomous vehicles, were adapted for home use.
Rethinking household robotics
Meanwhile, the embodied intelligence space is booming. Companies like Unitree Robotics, Leju Robot, and Noetix Robotics have all teased packed order books for the year.
But despite the buzz, the road to productization and scalable commercialization remains unclear. This uncertainty made GSR Ventures partner Zhu Xiaohu’s recent revelation that the firm is exiting humanoid robot investments “in bulk” all the more unexpectedly provocative.
In many ways, it simply reflects reality: the sector is still in its infancy.
Having helped usher a hardware category like robot vacuums from zero to mass market, Guo brings a pragmatic perspective.
Over the course of the interview, he used the phrase “consumer product” more than 20 times. No matter how ambitious the goal, or whether the robot has two legs or eight, Guo is focused on one thing: building robotics products for everyday users.
The compact, general-purpose robot for home use? That’s just step one.
Building the hardware equivalent of Manus
Joyin’s product planning began by zeroing in on the home, which stands as one of the few domains where consumer hardware has scaled both profitably and sustainably.
From there, Guo and his team identified three essential traits for a home robot: human-like design, mobility, and integration.
Joyin’s first two products reflect this framework. One is Z-Bot, a humanoid-style indoor robot around 50 centimeters tall that can freely navigate household environments. The other is W-Bot, a smart companion built for outdoor and mixed-use settings.
Getting there wasn’t straightforward. At one point, the team tried to retrofit hardware into imagined use cases without clearly defining the problems they were solving. This usually started with an initial idea like building a humanoid robot that roams the house. But that only led to more questions. Should it have mechanical arms? Bipedal locomotion? Tracks?
Eventually, Guo hit pause and redirected the strategy. Instead of hypothesizing use cases, Joyin looked to actual household needs.
The process was as simple as it was personal: the team made lists of daily life challenges, interviewed users, and reflected on personal routines. “I’d think about how I usually communicate with my family. Or about safety issues for my grandfather, who lives alone. How could a mobile robot help with that?” Guo said. One team member even brought a pet to the office to test interactions.
Those insights crystallized into a core design goal: make the robot fully mobile in the home. Then, address the long tail of needs that arise in family life.
A prime use case is eldercare or childcare. Fixed home security cameras are limited in both field of view and emotional acceptability. Many seniors dislike being surveilled. A robot that patrols specific areas at set times could offer both broader coverage and greater peace of mind.
Other long-tail applications include security patrol, environment sensing, and more. Guo likens it to the “mobile hardware version of Manus,” referencing the AI agent positioned as an all-in-one virtual workspace.
Once a robot can move and accompany users, it transforms into a richer interface. “Say you’re watching television. You might want a voice-controlled agent to adjust the lighting, check security footage, or talk to you. A mobile terminal changes the whole user experience,” Guo said.
Individually, these use cases may seem niche. But combined in one device, they could justify purchase and drive ongoing use.
Emotionally aware robots?
The release of DeepSeek-R1 was a turning point for Joyin’s product direction.
“This was the best thing to happen to me in 2025,” Guo said.
Models like DeepSeek-R1 offer major gains in inference capability. When integrated into a robot’s processor, they enable smarter decision-making and task execution, which are crucial for a field still early in its technical arc.
As part of its roadmap, Joyin is also developing an open robot platform for developers, students, and researchers. The idea is to let users shape the product from the outset. That means letting users swap in sensors, adjust software, and tailor the robot’s behavior to suit different home environments.
Guo believes reinforcement learning will be a key differentiator. In older systems, even basic actions like waving a toy for a cat had to be pre-programmed by engineers. With reinforcement learning, users can teach robots new behaviors more intuitively. The robot can then plan and complete those actions independently.
This also gives the user a more intimate, human-like experience. By spending time in the home and absorbing contextual cues, the robot can start to respond in more personalized ways.
Guo stressed that when Joyin talks about “human-like design,” it’s not chasing the humanoid ideal of robots folding laundry or making dinner. Instead, it’s about emotional resonance: building a presence that feels comforting. This sets Joyin apart from many startups focused on perfecting humanoid tasks. Rather than competing on technical mastery, Guo is betting on emotional connection.
As an example, Joyin’s first robot might not even have voice functions. “Even if it only uses body language and screen visuals, it can still convey emotional presence,” Guo said. The priority is trust and companionship. Other features can come later.
To support that goal, Joyin is co-developing emotion- and consciousness-mimicking models with its large model partners.
The first step is building a bond. Once that happens, users naturally offer more context. From there, the robot learns autonomously, and the open platform allows for continuous expansion.
Still, Guo is pragmatic:
“Right now, we have to make money while feeding the dream.”
In just six months since its founding, Joyin has completed a working product demo. Both its team and strategy are evolving rapidly.
It’s been only six months since the company was founded, and already the product demo is done. Both the company and its strategy are evolving at warp speed.
“When we think it’s the right move, we go for it,” Guo said. “But we also admit we could be wrong. The real question is: can we iterate and adapt fast enough to get closer to the vision?”
That vision? Making robots a real part of family life.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This article was written by Deng Yongyi for 36Kr.