Research shows only 34% of enterprise AI agent projects reach production due to knowledge gaps

A survey of 300 data, AI, and technology executives found that data and knowledge weaknesses consistently stall AI agent progress, with only 34% of agentic AI projects making it into production. Data fragmentation (cited by 55%) is the top challenge to knowledge access. Organizations with stronger knowledge capabilities, especially in semantics, achieve higher success rates, with 61% of their agentic projects advancing beyond pilot. Executives prioritize strengthening the structural link between organizational data and AI agents, with knowledge graphs being a key focus.