From MIT Technology Review · · 1m 40s
Half of AI agent insights fail to drive business value
A new report finds that fewer than half of insights generated by analytics agents lead to business improvement, with poor context and data fragmentation as key culprits.
In brief
A new report finds that fewer than half of insights generated by analytics agents lead to business improvement, with poor context and data fragmentation as key culprits. Fewer than half of agent-generated insights at surveyed organizations lead to business improvement, with insufficient context and fragmented data…
AI agents promise to transform analytics
Generative AI has made analytics accessible across organizations. Agentic AI promises to automate entire workflows and take actions based on agent insights, but quality remains a serious problem.
Less than half of agent insights drive improvement
On average, forty-eight percent of insights generated by analytics agents at surveyed companies lead to actions that improve the business. That is up from thirty-seven percent one year ago.
Top performers achieve much better results
At a small cohort of organizations termed insight leaders, more than seventy percent of agent insights lead to actions that improve the business. These leaders give their agents' insights higher marks overall.
Lack of context is the main barrier
When analytics agents deliver substandard insights, the main reason is lack of business context, cited by fifty-five percent of surveyed executives. Fragmented or inaccessible data is the other major factor, according to forty-five percent.
Too many insights overwhelm human reviewers
At a majority, fifty-two percent, of surveyed organizations, analytics agents generate too many insights to be properly evaluated and acted upon. Less than one-quarter, twenty-four percent, can evaluate all agent insights.
Companies focus on governance and semantics
Surveyed companies are prioritizing initiatives to strengthen AI and data governance, cited by sixty percent. Forty-five percent are adding a semantic layer to their data architecture to improve agent performance.
The bottom line
Fewer than half of agent-generated insights at surveyed organizations lead to business improvement, with insufficient context and fragmented data identified as main barriers.
