We propose a subtle self-tracking feedback by mirroring the self's state into an artificial agent. By eliciting empathy towards the artificial agent and fostering helping behaviors, users would help themselves as well. We reflect on the implications of this design framework, and the methodology to design and implement it.