A robotics startup has reached $1 billion valuation by making industrial robots that learn new tasks by watching human workers demonstrate them once or twice, eliminating complex programming. The technology uses the same neural networks behind ChatGPT, adapted for physical tasks instead of language. While aimed at factories, the underlying principle is transforming how small manufacturers and makers can adopt automation. Instead of hiring specialized programmers or sending staff for training, a shop floor worker can simply show the robot what to do. The system adapts when parts shift or conditions change, unlike traditional rigid automation. For small-scale producers considering automation but deterred by setup complexity and costs, this demonstrate-and-learn approach makes robotics accessible without engineering expertise or major workflow disruptions.