Dendral was an early artificial intelligence system developed in the 1960s at Stanford University to help chemists identify unknown organic molecules. It used mass spectrometry data and knowledge of chemistry to generate possible molecular structures and test them against the data. Dendral consisted of two subprograms: Heuristic Dendral, which produced potential structures, and Meta Dendral, which learned to explain the correlation between structures and spectra. The system pioneered the use of heuristics, knowledge engineering, and the plan-generate-test problem-solving paradigm in expert systems.