Jul 2026
ANALYSIS OF ARCHITECTURAL COMPONENTS OF MULTI-AGENT SYSTEMS
It is found that on the examined tasks the multi-agent mode underperforms the single-agent baseline in quality and a pattern is discovered: the integration of three verification loops produces a self-diagnosis capability not observed in any of the examined systems individually.
T. Saratova, Aleksandr Apuhtin
· Scientific and analytical jo... · 0 citations