Enterprises Struggle with AI Agent Accuracy

A recent VB Pulse survey found that 57% of enterprises have traced confident but wrong AI agent answers to missing or inconsistent business context. The survey of 101 qualified enterprises with more than 100 employees revealed that retrieval over documents is the default way agents get business context for 38% of enterprises. However, this approach often leads to accuracy problems that only become apparent after the system is live. The reason for this issue is largely due to the way enterprises choose retrieval systems, prioritizing ease of ingestion and operational simplicity over retrieval accuracy. To address this problem, vendors are racing to roll out context platforms, with 25% of respondents already running a governed context layer and 34% building one. The remaining 41% have not started, but companies that have experienced confident-wrong failures are more likely to be building a fix.