Most Agentic AI Projects in Production Have Stalled Over Data Problems
Help Net Security, Thursday, June 18th, 2026
77% of organizations running agentic AI in production report stalled projects due to data governance, quality, and infrastructure issues.
According to Confluent's Data Streaming Report, 32% of enterprises now deploy AI agents in production for tasks previously requiring human review. However, obstacles impede scaling: a skills gap and limited organizational readiness rank first at 69%, followed by LLM reliability concerns (68%), data infrastructure issues (66%), and governance problems (65%).
Among production deployments, 77% experience project delays while 61% face complete abandonment. The research emphasizes shifting security controls left toward data sources, with 81% of IT leaders viewing inline security enforcement as highly beneficial for autonomous agents operating on unverified data.