Organizations have begun to offer customer service using a combination of human and machine speech recognition. Humans provide unprecedented accuracy and enable formerly impossible features, but are more costly. This presentation describes strategies for how humans can efficiently work together with machines and gives examples of new services the union now makes possible. The session reviews case studies of deployed services across multiple companies and vendors; covers issues such as accuracy, automation rates, and response latency; and gives a glimpse into where this approach may head in the future.