In case-based reasoning, previous solutions are recalled and adapted to fit new problems. However, for complex problems with multiple stakeholders, multiple sources of experience should be considered to increase the diversity and effectiveness of such solutions. Here we present the approach of robust coherence. This approach combines two seemingly contradictory theories from the philosophy of knowledge: coherence and falsification. Using these two theories in concert, robust coherence seeks to justify contributions from several agents in a collective context that also corresponds to reality. We also present rob-coh, and algorithm which applies robust coherence to multi-agent case-based planning. Using rob-coh, multiple agents can suggest actions and goals from experience to address a problem, and a prediction of the best option can be made based on the robust coherence among these experiences.