Companies Are Running AI Agents. Are They Measuring Whether They Work?

Published 14 October 2026

Are companies measuring whether their AI agents actually work? A 2026 survey from Liferay suggests many are not. Its report, based on 500 US professionals involved in AI decisions or day-to-day use, found that 54% of companies are running AI agents in production or piloting them, yet only 25% measure AI's impact with clear KPIs. Only 24% have a company-wide AI usage policy. Liferay published the report, so read it as vendor research.

A separate survey from CrewAI, which sells an agent platform, asked 500 senior leaders at large companies with 5,000 or more employees. It found that 65% already use AI agents. When asked what matters most in evaluating AI agent platforms, security and governance came first at 34%, while only 2% named time-to-value and return on investment as their primary consideration. These figures are self-reported and come from a vendor.

The two surveys ask different groups different questions, so the numbers should not be compared line by line. They do point to the same gap: running agents is easier than measuring them.

This matters because without a measure, you cannot tell whether an agent saves time or makes mistakes. In the Liferay survey, 42% of companies said better accuracy is what AI agents need most to become more useful, and 29% named cost as a leading barrier to getting more value from AI. A clear baseline makes both easier to judge.

At Zamorins Solutions we build AI and automation solutions for business.

What you can do this week

• Pick one AI agent or automation you already use and write down the one result it should improve, such as hours saved or errors avoided.
• Record that number today as a baseline, and check it again in a month.
• Decide who owns the number and who reviews the agent's mistakes.

Takeaway: Running an AI agent is not the same as knowing it works, so set one measurable goal and a baseline before you expand.

Sources

• New Liferay Study Finds 54% of Companies Are Running AI Agents, Yet Only 25% Measure Their Impact, Liferay 2026 Agentic AI Adoption and Governance Report, as reported by Automation Magazine (August 12, 2026)
• Survey: Enterprises move AI agents from pilots to production, Digital Commerce 360, on the CrewAI survey (February 11, 2026)