We analyze how agents' present bias affects optimal contracting in an infinite-horizon employment setting. The principal maximizes profits by offering a menu of contracts to naive agents: a "virtual" contract - which agents plan to choose in the future - and a "real" contract which they end up choosing. This virtual contract motivates the agent and allows the principal to keep the agent below his outside option. Moreover, under limited liability, implemented effort can be inefficiently high. With a finite time horizon, the degree of exploitation of agents decreases over the life-cycle. While the baseline model abstracts from moral hazard, we show that the result persists also when allowing for non-contractible effort.