Abstarct: With the increasing use of smart wireless devices and the advent of the IoT idea, wireless communication systems are faced higher data rates as well as better quality services. To achieve such a high data, the wireless networks should be work at high frequencies, such as millimeter and THz bands. On the other hand, with increasing frequency, the waves are not able to pass through obstacles such as walls and are lost in the environment that cause less penetration depth. Multi-path fading also limits the use of such high frequencies of electromagnetic waves due to the interaction of signals from several different paths in the air from the transmitter to the receiver. To control this, the researchers came up with the idea of covering objects in the environment with Software-Defined mextasurface to turn publishing environments into a programmable environments. In this method, all the walls are covered with pieces of the same size from the mextasurface, called tiles. The main challenge is that the environment can change during the communication, for example a sudden obstacle blocks the signal path, this management must be adaptive. This paper, the use of Reinforcement Learning (RL) algorithm to create such an environment is presented. Since an agent in the RL algorithm gradually interacts with the environment, it can be better decision, especially when the environment changes. The results of the proposed method are compared with the previous work, which used a genetic algorithm to create a programmable environment. In the previous work, the average of the received power for 12 receivers is 25.38 dBmW and the reinforcement learning method in this paper, increases the average of the received power by 12%.