The growing popularity of digital assistants (from Microsoft’s Clippy to Amazon’s Alexa) is changing how consumers acquire information and make decisions. Often embodied in anthropomorphized forms, digital assistants (DAs) are designed to serve consumers by suggesting relevant products to simplify purchasing decisions. In this work, we aim to understand how consumers evaluate social relationships with different types of DAs and their subsequent effects on purchasing. Our findings show that consumers judge DAs as being more socially close both when DAs are anthropomorphized and when they provide higher-quality recommendations. Evidence from fMRI indicated that both recommendation quality and anthropomorphization fostered greater feelings of social closeness by recruiting similar brain mechanisms involved in mental simulation (i.e., inferior frontal gyrus and cortical midline structures). Although anthropomorphized DAs were evaluated as more socially close, they did not facilitate increased purchase interest, suggesting that stimulation of neural reward networks is also necessary for driving greater purchasing.