A survey of attendees at the WSJ's CIO Network Summit: 61% are experimenting with AI agents, but 21% are not using them at all, citing a lack of reliability
Tech vendors like OpenAI and Microsoft are banking on business readiness to use the autonomous AI bots, but companies aren't so sure
Context & Ripple Effects
This survey captures an early enterprise-agent split: experimentation is widespread among CIO Summit attendees, but reliability remains a barrier to use. That makes the gap between vendor expectations and operational confidence the central issue for Microsoft and OpenAI.
The hesitation sits alongside conflicting definitions of what counts as an AI agent, which adds evaluation friction for buyers. It also echoes earlier executive concern about being able to explain how AI uses data when deploying systems in governed business settings.
First-order effects
- Most attendees are testing agents, creating a near-term market for pilots and evaluation rather than broad autonomous deployment.
- The 21% not using agents because of reliability gives Microsoft, OpenAI, and other vendors a clear adoption constraint: business buyers are not yet uniformly ready to trust autonomous bots.
Second-order effects
- Enterprise AI vendors will be pushed to compete on demonstrable reliability and clearer product boundaries, not simply on agent branding; definitional confusion can further slow purchasing decisions.
- CIO teams are likely to concentrate early use on settings where agents can be tested and supervised, while organizations that cannot establish confidence remain outside the market.
Third-order effects
- If reliability remains the gating factor, enterprise-agent adoption will develop unevenly: experimentation may grow faster than production autonomy, favoring products embedded in workflows where performance can be evaluated and controlled.
- The broader enterprise AI market may shift from selling access to models toward proving the cost and dependability of useful completed tasks, with governance expectations reinforcing that shift.
The trend: Enterprise AI is moving from broad experimentation toward a proof-of-reliability phase in which autonomous-agent adoption depends on operational trust rather than vendor enthusiasm alone.