This study focuses on the impact of team cognitive load on compliance with an Artificial Social Intelligence agent’s (ASI) advice. It also expands on some of the factors that influence team performance, including cognitive load, compliance, and ASI interaction dynamics. The study design comprised three types of ASI agents that advised all-human teams, each generating their advice based on variations in message length and frequency: long messages at low frequency, moderate lengths and frequency, and short messages at high frequency. Three team members collaborated to locate and save victims in a simulated Urban Search and Rescue (USAR) task environment, while the ASI provided intervention messages (i.e., advice) through text chat. The ASI monitored the team members in the USAR task environment via its interaction-based analytic components. Then, ASI predicted human team members’ behaviors based on their past and current interactions to provide teamwork interventions to maintain team effectiveness. The findings indicate that (1) team cognitive load was not associated with team compliance with ASI advice, (2) both team cognitive load and compliance with ASI messages were positively related to team performance score, (3) Teams assigned an ASI that had moderate advice length and frequency performed better than the teams that were paired with the other two types of ASIs which demonstrated either short message length and high frequency or long message length and low frequency. Overall, these findings show that the ASI advice interventions are helpful as long as they have moderate-level message length and frequency and are complied with by the team members in the USAR task. Future designs of ASI agents should target these types of intervention message characteristics and prioritize compliance to improve team performance.