This study aims to predict carbon monoxide (CO) and carbon dioxide (CO2) levels as indicators of air quality using the Artificial Neural Network (ANN) method with a 21-20-7 architecture. Data was collected over 28 days using the AQMSense device, with the first 21 days used for training and the remaining 7 days for testing. The results show that the ANN model effectively recognizes historical data patterns, indicated by a regression value close to 1. The prediction accuracy reached 89.96% for CO and 95.84% for CO2, demonstrating strong model performance. ANN proves to be an effective tool for daily air quality prediction and holds potential as a decision-support system for air pollution mitigation