The quantum computer concept was first introduced by R. Feynman in 1982. During the 13 following years, its implementation was impossible as quantum systems are subjected to relaxation and decoherence leading to the destruction of information. In 1995, P. Shor demonstrated it was possible to protect information using a quantum error correction (QEC) code. With those codes, the fundamental bricks of the computer, the physical quantum bits(qubit), are combined into a logical qubit. The later is protected from the relaxation and dephasing or equivalently from the bit-flip and phase-flip errors. In the superconducting circuit field, this approach is adopted by worldwide compagnies such as Google and IBM which use the QEC surface code. Because physical qubits suffer from two errors (whereas classical bits suffer from only one) and QEC codes must deal with the measurement backaction, the surface code needs 1,000 physical qubits to obtain only 1 logical qubit containing only 1 bit of information. To reduce this overhead, resource-efficient approaches called bosonic codes were developed. They consist in encoding the information inside states of bosonic modes such as superconducting electromagnetic resonators. Bosonic codes such as Gottesman-Kitaev-Preskill and pair cat codes can fully protect information, while Kerr cat and dissipative cat codes protect information against only one error but enable the use of a QEC code with lower overhead (about 30 instead of 1,000). Along the past few years, bosonic codes were developed using superconducting circuits in private and academic groups such as Yale, AWS, ENS Paris and Alice&Bob.   I will discuss why bosonic code is an interesting approach to reduce the overhead of QEC codes and to implement gates in quantum computing. I will also explain how they can be implemented with superconducting circuit and present the latest results in this field.