Survey finds only a third of enterprise agentic AI projects reach production due to knowledge gaps

A survey of 300 data, AI, and technology executives by MIT Technology Review Insights identifies a persistent barrier to scaling enterprise AI agents: a lack of organizational knowledge integration. On average, only 34% of agentic AI projects reach production, with legacy data systems, security and privacy constraints, and insufficient contextual understanding cited as the primary causes of failure. Even high-tech firms—often early adopters of AI—struggle to move beyond pilot stages.

The report distinguishes between organizations that successfully deploy agentic systems and those that do not. A subset of "production leaders"—where an average of 61% of agentic projects advance beyond pilot—demonstrate markedly stronger knowledge capabilities, particularly in semantic understanding. This correlation suggests that robust knowledge integration is a key differentiator for scaling agentic AI.

Fragmented data across systems is the most frequently reported challenge to expanding agent access to knowledge, cited by 55% of respondents. In contrast, production leaders are more likely to flag security and privacy concerns as a major obstacle, with 72% of this group identifying it as a top challenge. The contrast underscores differing priorities between lagging organizations and those already operationalizing agents at scale.

To address these gaps, executives report plans to invest in technologies that strengthen the structural foundation between enterprise data and AI agents. The most impactful steps, according to respondents, involve enhancing retrieval technologies—such as ingestion pipelines, AI-ready APIs, and retrieval-augmented generation (RAG)—as well as deploying AI evaluation agents and knowledge graphs. These investments aim to provide agents with the contextual grounding required for reliable reasoning and decision-making.