The main advantage of autonomous biomolecular computing devices over electronic computers is their ability to interact directly with biological systems. No interface is required since all components of molecular computers, including hardware, software, input, and output are molecules that interact in solution along a cascade of programmable chemical events. Here, we demonstrate for the first time that the output of a computation preduced by a molecular finite automaton can be a visible bacterial phenotype. Our 2-symbol-2-state finite automaton utilized linear double-stranded DNA inputs that were prepared by inserting a string of six base pair symbols into the lacZ gene on the pUC18 plasmid. The computation resulted in a circular plasmid that differed from the original pUC18 by either a 9 base pair (accepting state) or 11 base pair insert (unaccepting state) within the lacZ alpha region gene. Upon transformation and expression of the resultant plasmids in E. coli, the accepting state was represented by production of functional beta-galactosidase and formation of blue colonies on X-gal medium. In contrast, the unaccepting state was represented by white colonies due to a shift in the open reading frame of lacZ.