We examine the issues that arise in extending an estimatedregression (ER) planner to reason about autonomous processes that run and have continuous and discrete effects without the planning agent’s intervention (although the planner may take steps to get processes running). An ER planner is a classical planner that searches situation space, using as a heuristic numbers derived from a backward search through a simplified space, summarized in the regressionmatch graph. Extending the planner to work with processes requires it to handle objective functions that go beyond the traditional step count or cumulative step cost. Although regressing through process descriptions is no more difficult than regressing through standard action descriptions, figuring out how good an action recommended by the regressionmatch graph really is requires "plausibly projecting" the subtree suggested by the action, which often requires forcing actions to be feasible. The resulting algorithm works well, but still suffers from the fact that regression-match graphs can be expensive to compute.