Recent breakthroughs in the field of neural prostheses have shed light on innovative approaches for restoring lost functions, such asspeech in completely paralyzed patients or the ability to walk in para/tetraplegic patients. Beyond the impressive results achieved and the evident benefit for the patients, brain implants used in such contexts mainly rely on energy intensive hardware-based solutions and are unidirectional (ie. either only stimulating the target brain area, or only recording neuronal activity). In this view, closed-loop neurostimulation systems, which would adapt the stimulation parameters to any physiological/pathological change in the target nervous tissue, are much needed. To address these issues, we have chosen a material approach based on an original artificial neuron with optimized energy efficiency previously developed in the team (Sourikopoulos et al. 2017, Front. Neurosci. https://doi.org/10.3389/fnins.2017.00123), and conducted an interdisciplinary study involving neurosciences, electronics and biological microelectromechanical systems (BioMEMS). We developed a novel type of neurobiohybrid interfaces based on a biomimetic spiking artificial neuron relying on a standard 65nm CMOS TSMC technology. The biomimetic properties of this artificial neuron in terms of frequency, time constant and amplitude of its generated action potentials enable a direct analog communication with living neurons thus providing real-time adaptability to the system. Different types of living neurons including human iPSC-derived cells were maintained in vitro in such neurobiohybrid interfaces and either stimulated by emulated biomimetic signals or recorded. Concomitant calcium imaging acquisitions enabled to ensure the biological reliability of recorded electric signals on an in-house electronic bench. Our results show the ability of such neurobiohybrid systems: a) to stimulate directly and specifically electrically responsive cells maintained in vitro, the biological specificity of the electrical stimulation triggered being corroborated by concomitant calcium imaging assessments, and b) to record specific electric signatures from different neurons in culture. Altogether, the present work sets the bases for further studies towards a real-time bidirectional communication between artificial and living neurons. In conclusion, such next-generation devices enable the development of an efficient closed-loop communication between neural prostheses and living neurons.