With the recent loop-hole free Bell experiments, the concept of device independent QKD and QRNG has gained a lot of attention. However, in practice this approach is much too complicated and useless. Here, we present a much simpler, so called self-testing approach, to generate random numbers. Based on very little and reasonable assumptions, we design a device which can test itself continuously and certify a minimum amount of entropy. This is based on the concept of Unambiguous State Discrimination. In a second part, I will demonstrate a new QKD setup which can distribute keys over more than 400km.