Abstract-People-driven service engagements involve communication over channels such as chat and email. Such engagements should be understood at the level of the commitments that the participants create and manipulate. Doing so provides a grounding for the communications and yields a business-level accounting of the progress of a service engagement. Existing work on commitment-based service engagements is limited to designtime model creation and verification. In contrast, we present a novel approach for capturing commitment-based engagements that are created dynamically in conversations. We monitor commitments identifying their creation, delegation, completion, or cancellation in the conversations. We have developed a prototype and evaluated it on real-world chat and email datasets. Our prototype captures commitments with a high F-measure of 90% in emails (Enron email corpus) and 80% in chats (HP IT support chat dataset) and provides promising results for capturing additional commitment operations.