This research work is oriented to determine the feasibility of determining work stress using facial emotion recognition software designed for this purpose. First, the software was developed for facial recognition of emotions, working with free tools such as PHP, Apache, MySQL, and Laravel. We integrated routines from Amazon web services designed exclusively for facial emotion recognition and applied an Extreme Programming (XP) methodology for the development of this software. Then an experimental group of 20 people was formed, these initially evaluated their stress levels in a software that automated the Maslach questionnaire, this software was also created by the researchers of this work. Then, facial recognition was taken from all of them for 3 weeks, achieving about 400 samples. Finally, a simple linear regression model was made, and the values obtained were stratified for classification. The values between both tests did not coincide, which leads to the conclusion that artificial intelligence-based facial recognition of motions requires further study and further integration of intelligent algorithms to make the results more reliable.