Raising the quality of insights in agentic analytics

Judgment

CIOs and analytics leaders should not scale agentic workflows by buying bigger models; they must fund semantic layers and governance first, because more than half of agent insights still fail to improve the business. The tradeoff is slower deployment and higher verification cost against faster automation.

Basis

MIT Technology Review Insights, in a report sponsored by Snowflake, reports that 48% of insights generated by analytics agents at surveyed companies lead to actions that improve the business, up from 37% a year earlier—meaning more than half still fail. When quality is substandard, 55% of executives cite lack of business context and 45% cite fragmented or inaccessible data. The report also finds 52% of organizations generate too many insights to evaluate and act on, and only 24% can evaluate all agent insights; review is mostly case-by-case and human-intensive. Insight leaders—a small cohort—see more than 70% of agent insights lead to business-improving actions. Respondents prioritize AI/data governance (60%) and adding a semantic layer (45%). Mechanism: context and semantics improve prescriptive insight quality, but human verification remains the bottleneck, so scaling automation without governance raises the cost of checking flawed actions.