The aim of this report, which relies on a survey of 300 information, AI, and different expertise executives, is threefold. First, it seeks to gauge organizations’ agentic information capabilities (i.e., their means to offer AI brokers a full contextual understanding of the info they ingest) throughout semantic information, episodic reminiscence, and procedural information.
Second, the report probes the challenges organizations face in enhancing entry to information and in the end to getting extra agent use instances into manufacturing. Third, it explores the measures organizations are taking to beat these challenges.
The important thing findings embody the next:
Knowledge and information weaknesses persistently stall AI agent progress. On common, solely round a 3rd (34%) of organizations’ agentic AI tasks make it into manufacturing. Even high-tech companies battle with this. Legacy information programs, safety and privateness issues, and a lack of information and context are the important thing factors of failure.
Robust information capabilities correlate with agent success. A small group of manufacturing leaders (organizations the place on common 61% of agentic tasks advance past pilot) have stronger information capabilities than the remaining, particularly relating to semantics. This benefit tracks carefully with their larger manufacturing price.
Fragmented information massively complicates information entry. Knowledge fragmentation (the insufficient sharing of knowledge throughout programs) was mostly cited as a high problem to increasing brokers’ entry to information (cited by 55%). Manufacturing leaders, against this, usually tend to see safety and privateness issues as a significant concern (cited by 72% of this group).
Most companies purpose to strengthen the hyperlink between information and brokers. Amongst steps that may yield larger high quality agent choices, executives count on the most important affect to return from strengthening the structural basis between the group’s information and its AI brokers. The consultants we interviewed see a information layer as a first-rate strategy to obtain this.
Funding priorities to spice up information vary from pipelines to information graphs. To develop agent entry to information, organizations will prioritize investments in retrieval applied sciences, like ingestion pipelines, AI-ready APIs, and retrieval-augmented technology (RAG); in AI analysis brokers; and in information graphs.
Obtain the report
This content material was produced by Insights, MIT Know-how Evaluation’s customized content material arm, not its editorial workers. It was researched and written by people, with any AI instruments that will have been used restricted to manufacturing processes beneath human oversight.