The Agibot R2 wheeled humanoid robot working on a Longcheer tablet production line.

Agibot

Agibot’s wheeled G2 humanoid robots are now working on live production lines for tablet computers alongside humans, says the company. These are actual jobs working alongside human co-workers on the main production line, so it’s not a test or demo environment. This is, the company says, “the world’s first large-scale industrial implementation of embodied AI systems within core production workflows in consumer electronics manufacturing.”

Agibot’s G2 is not the first humanoid robot to get a job, of course.

Digit by Agility Robotics famously got the world’s first humanoid robot paying job back in November of 2024, plus a bigger job with Toyota Canada early this year. And Figure’s second-generation model, since supplanted by Figure 03, was right behind it. Plus, there are plenty of robots doing test gigs in logistics or manufacturing, or working in reception-type roles.

But this is a first for high-speed electronics manufacturing, Agibot says.

“2026 marks the beginning of large-scale deployment for embodied intelligence,” Yao Maoqing, SVP of Agibot’s embodied business unit, said in a statement. “This project demonstrates that embodied AI is no longer experimental. It is a practical, production-ready capability that can operate reliably under real industrial conditions and deliver measurable economic value.”

Agibot recently celebrated its first AI Week, where it announced out a full-stack vision for embodied intelligence that stitches together data, simulation, models, developer tools, and deployment into a single massive but coherent system. The clear goal: being able to control the entire robotic intelligence stack from data to deployment.

Among the announcements:

• A large-scale, real-world robotics dataset
• LLMs to generate massive simulated environments, enabling scalable training and evaluation
• Foundation models for robots combining vision, language and action
• World model research to give robots predictive understanding of physical environments
• No-code development and end-to-end deployment workflows for companies employing robots

That investment in AI, robotic development, and a holistic approach to managing and implementing robotic workforces is starting to pay off.

Agibot’s customer on this production line job is Longcheer, an electronics manufacturing company based in Shanghai with 5,500 workers and seven research and development centers, ranking 328th on the Fortune China 500 for 2025.

“This deployment represents more than a technical milestone, it marks the first real-world validation of embodied AI in consumer electronics precision manufacturing,” Li Long, Longcheer’s robotics division general manager, said in a statement. “In just four months, the AGIBOT G2 was integrated into Longcheer’s mass production line, delivering stable, continuous operation and meeting all key targets. This milestone highlights a scalable path forward for embodied AI in manufacturing.”

The actual job, like many in electronics manufacturing, is not especially glamorous, which makes it perfect for robots. Agibot’s G2 robots work on “high-speed assembly lines,” picking up tablets, moving them to a testing area, placing them with "millimeter-level accuracy," then sorting finished or defective units as required.

As you can see in the video above, the G2 models that Longcheer are using seem to have some custom-made grippers that exactly match the use case: picking up and placing tablets for testing. The millimeter-level precision is impressive, and the speed is also impressive: almost certainly better than any legged humanoid robots today could accomplish.

You could argue that a bodiless robotic arm could do the job too, and perhaps even better, but Agibot says a key part of the functionality is being able to adapt to changes in the production line for short-run products, which might be a bit more challenging with an anchored robotic arm system.

G2’s production stats aren’t bad, according to Agibot:

• Throughput is up to 310 units per hour
• Cycle time is approximately 19–20 seconds per operation
• Success rate is over 99% in continuous operation
• Production output is approximately 3,000 units per shift
• Operational stability is good: the R2 works autonomously with minimal human intervention and “over 140 hours of cumulative continuous operation, with downtime loss below 4%”

Overall, this is impressive. And, given the improvement rates of electronics, technology and AI these days, it’s only going to get better.