AI Governance Market: Why AI Agents Amplify Broken Customer Data Problems

Bryan Cheung, co-founder and CMO of Liferay, argues that deploying AI agents into fragmented customer data environments amplifies existing inconsistencies rather than resolving them. He notes that AI agents deliver whatever data they can access with confidence, leading to wrong answers that erode customer trust. The article identifies governance gaps—such as unclear data access limits, authorization, logging, and accountability—as the real bottleneck to enterprise AI adoption, not AI capability. Cheung outlines six infrastructure requirements for trustworthy AI agents: access control, a reliable source of truth, audit trails, human review checkpoints, an escalation path, and model-agnostic architecture. He advocates treating governance as a prerequisite to scaling AI effectively.