AutoDNA connects AI planning to physical laboratory execution

The Nature paper published November 4 describes a system connecting agents with laboratory instruments; Shanghai Jiao Tong University’s November 6 account reports a DNA-storage demonstration involving 11,382 hardware actions and successful recovery of the stored information. The result concerns an integrated experimental workflow, not a general robot scientist proven across arbitrary laboratories. [1][2]

The engineering contribution is to make equipment capabilities available through consistent software interfaces so that planning, execution and analysis can connect. That integration matters because a plausible written protocol does not establish that a machine can execute it safely or that the resulting measurements answer the research question. The demonstration provides a concrete example of that bridge, with its generality still to be tested. [1]