Agibot has launched its AI model, Genie Operator-1 (GO-1), designed for humanoid robots. The system enables robots to quickly understand and execute tasks.

Founded by former Huawei employee Peng Zhihui, Agibot has developed a neural network that acts as the brain for humanoid robots, facilitating learning and adaptation.

“Bots will no longer be confined to laboratories—they will be able to adapt to changing real-world conditions. They will understand natural language instructions and reason beyond pre-programmed procedures,” Agibot stated.

GO-1 is built on vision-based models, which process large volumes of images and video to enhance the robots’ understanding of human actions. Planning algorithms help them break down tasks into steps and execute movements accordingly.

The system is capable of learning from video demonstrations. In a demo, humanoid robots were seen preparing and serving toast, distributing badges, and pouring coffee. According to Agibot, its machines have improved in task execution thanks to this AI.

Earlier, Figure introduced its own AI model, Helix, designed for humanoid robots. The company claims it is capable of “reasoning like a human.”

Agibot previously ended its partnership with OpenAI after achieving a breakthrough in AI development for humanoid robots.