This article describes the views on the architecture of distributed AI systems based on the simulated bio-neurons representing the basis for the bio-neural circuits, which represent distributed AI subsystems and serve as microservices for the AI client-side systems. The article also describes the interface and the demands to the protocol of communication with the distributed subsystems of the AI, the ways of tuning the synaptic contacts in the brand new neural circuits, which represent the distributed AI systems, and finally, the new approach to communication with such the systems based on the new computer language, which will be used in construction and tuning of such the AI systems.