Amino acids are key contributors to nitrogenous emissions during biomass pyrolysis, yet the underlying reaction mechanisms governing their thermal degradation remain only partially understood. In this study, we combine systematic reaction path search algorithms with chemical insight and density functional theory (DFT) simulations to investigate the thermal decomposition of glycine (Gly), the simplest amino acid, with a focus on the formation of ammonia (NH3) — a major precursor of environmentally harmful NOx species. We derive a comprehensive reaction network for the thermal decomposition of Gly. Notably, we show that, at variance with water that can be generated via simple dimerization in the gas phase, NH3 evolution is kinetically unfavorable at moderate temperatures and low-pressure conditions, while it can proceed with much smaller barriers in the condensed phase via many-body mechanisms involving ionic-pair proton-exchange-driven polymerization pathways. Under such conditions, we predict that NH3 evolution competes with H2O formation, reconciling theoretical predictions with experimental observations.