Optical quantum information processing relies on our ability to engineer an effective interaction between single photons. The currently available process - based on photon coalescence with a beamsplitter cube - has already allowed major realizations (quantum teleportation, entanglement swapping, logic gates). Yet this implies drastic conditions on the indistinguishability of the single photons used, which limits the scalability of this technique. To obtain an interaction between distinguishable photons or even delayed photons, other interaction processes between photons have to be to developed. In this framework, two research axes have been recently explored in LPN. The first axis aims at developing a single-photon router, a quantum device able to transmit a first photon and reflect a second one. In this respect, we have recently demonstrated an optical nonlinearity obtained with few-photon pulses. The second axis aims at developing a spin-photon interface, using a single spin as a quantum memory mediating the interaction between successive photons. In this respect, we have recently demonstrated a giant rotation of a laser polarization induced by a single spin. In both cases the major challenge will be the control of a quantum system using only single photons, rather than lasers. As will be shown, this can be achieved with realistic devices.