A common use of systems thinking (ST) is for guiding our practices of systems making (SM). One style of ST for SM centers on making designs with deterministic rules, as in the hard sciences, for guiding engineered applications. Another style mimics natural development, with a process by stepwise learning and improvisation to produce evolving designs; examples including architectural design, scientific research, and the practice of action research (AR). All these use exploratory pathfinding to search for better ways to work with reality, and this is the main subject of the paper. Both deterministic and adaptive ST for SM are widely found in differing roles, each having capabilities the other lacks. I start with simple models, such as step-wise improvisation for adapting recipes when making dinner. Another example is Robert Rosen’s model for how scientific and other cultures learn to work with nature, by turning attention back and forth between nature and theory for creating their cultural language. A review of the modern history of the systems sciences, as practices of ST for SM, then further broadens the view and context. That leads to introducing a new paradigm of natural systems thinking (NST), using commitments to critical awareness, emancipation, and methodological pluralism for working with natural systems.