The rapid evolution of artificial intelligence from predictive tools to autonomous agents fundamentally transforms how humans and machines collaborate in organizational settings. While hybrid intelligence theory has established frameworks for understanding human-AI problem-solving, critical gaps remain regarding the cognitive biases that shape these collaborations. This systematic literature review synthesizes research across three converging streams: hybrid intelligence theory, AI agents’ evolution, and cognitive biases in human-AI collaboration. Drawing on the behavioral theory of the firm and extending Raisch and Fomina's (2025) hybrid problem-solving framework, we identify seven theoretical gaps at the intersection of these literatures. Our analysis reveals that automation bias, algorithm aversion, confirmation bias amplification, and expertise paradoxes operate distinctively in agentic AI contexts, yet remain undertheorized in management scholarship. We propose an integrative framework mapping how cognitive biases manifest across autonomous, sequential, and interactive hybrid problem-solving processes. This review contributes to management theory by establishing cognitive biases as a critical lens for understanding hybrid intelligence outcomes, offers practical implications for organizations deploying AI agents, and outlines a research agenda addressing the recursive dynamics between human cognition and increasingly autonomous AI systems.