A research team from the University of Toronto and other institutions has published a paper proposing the concept of a Self-Building Laboratory (SBL). Unlike Self-Driving Laboratories (SDL), which operate automatically only within preset capability boundaries, SBL incorporates the definition and expansion of laboratory capability boundaries into the automated closed loop, enabling autonomous identification of capability gaps and their fulfillment based on research goals. The system is built on digital twins and operates through the collaboration of scientific agents, twin agents, and safety agents. It has been validated through cadmium sulfide quantum dot synthesis experiments, demonstrating the feasibility of executing the entire process from scientific goals to real experiments. Future work will adapt to more experimental task equipment and cover larger-scale experimental spaces. SBL aims to enable laboratories to proactively expand their execution capabilities according to research goals, freeing researchers from building and debugging automated systems to focus on the scientific questions themselves. The field still needs to address issues such as experimental safety and the gap between simulation and reality.