Avian influenza virus (AIV) outbreaks occur frequently worldwide, causing a potential public health risk and great economic losses to poultry industries. Considering the high mutation rate and frequent genetic reassortment between segments in the genome of AIVs, emerging new strains are a real threat that may infect and spread through the human population, causing a pandemic. Therefore, rapid AIV diagnostic tests are essential tools for surveillance and assessing virus spreading. Real-time reverse transcription PCR (rRT-PCR), targeting the matrix gene, is the main official standard test for AIV detection, but the method requires well-equipped laboratories. Reverse transcription Loop-Mediated Isothermal Amplification (RT-LAMP) has been reported as a rapid method and an alternative to PCR in pathogen detection. The high mutation rate in the AIV genome increases the risk of false negative in nucleic acid amplification methods for detection, such as PCR and LAMP, due to possible mismatched priming. In this study, we analyzed 800 matrix gene sequences of newly isolated AIV in the EU and designed a highly efficient LAMP primer set that covers all AIV subtypes. The designed LAMP primer set was optimized in real-time RT-LAMP (rRT-LAMP) assay. The rRT-LAMP assay detected AIV samples belonging to nine various subtypes with the specificity and sensitivity comparable to the official standard rRT-PCR assay. Further, a two-color visual detection RT-LAMP assay protocol was adapted with the aim to develop on-site diagnostic tests. The on-site testing successfully detected spiked AIV in birds oropharyngeal and cloacal swabs samples at a concentration as low as 100.8 EID50 per reaction within 30 minutes including sample preparation. The results revealed a potential of this newly developed rRT-LAMP assay to detect AIV in complex samples using a simple heat treatment step without the need for RNA extraction.