While nearly eight in 10 enterprises report having AI agents in production, relatively few have connected their AI investments to measurable financial outcomes, according to new research from business process and technology services provider Genpact.
The company’s Autonomy in Practice report, based on a survey of more than 2,000 executives across 17 industries, found that 79 percent of organizations have deployed AI agents. Only 20 percent, however, use them consistently across business functions, and only 15 percent have fully connected AI investments to profit-and-loss outcomes.
The research also found that only 49 percent of AI initiatives funded during the previous two years delivered their expected value, on average. Nine percent of respondents reported completing significant operating model transformations.
“Enterprises have become effective at deploying AI agents. Running them is the hard part,” said Sanjeev Vohra, chief technology and innovation officer at Genpact. “Many organizations have deployed agents into operating models designed for human-only work, and much of the value they expected has yet to arrive. Rethinking the enterprise operating model for people and agents together is what we call agentic operations. Agentic Operations connects AI with context-rich process intelligence, human judgment, and governance to capture the autonomy dividend.”
Despite growing adoption, human oversight remains prevalent. Respondents reported that AI leads or fully executes 40 percent of decisions, while humans continue to make or audit decisions in at least 87 percent of cases at every decision stage.

