Abstract : This research aims to eventually create problem solvers that automatically adapt their problem solving techniques, representations, and heuristics to fit the current problem and computing environment. In this report, we discuss our prototype, RA*, which automatically adapts its heuristics to the current problem and computing environment. RA* does this by computing the impact of its candidate heuristics upon its ability to solve the problem in this environment and using the heuristic with the most impact. Our experiments show that RA* can solve more of the 2011 IPC Deterministic Optimization Track's problems than current state-of-the-art heuristic selection systems and RA* can solve them faster.