Jack Paul 
Genpact’s new research, released on November 12, 2026, shows a split picture in enterprise AI. Most large companies have moved agents into production. Far fewer have changed the way they run the business around those agents. The survey covered more than 2,000 executives across 17 industries and 14 business functions, with companies ranging from $1 billion to over $50 billion in annual revenue.
Deployment outpaces operations
About 79% of surveyed firms say they have AI agents live in production. But only 20% use those agents broadly and consistently across functions. That gap matters. Putting an agent into a workflow is one step. Running it alongside people, rules, and financial targets is another.
Only 15% of enterprises have fully connected AI spending to P&L outcomes. On average, 49% of AI initiatives funded in the previous 24 months delivered expected value. Just 9% said they had completed substantial changes to their operating models. Genpact calls the missing payoff the “autonomy dividend,” meaning gains that show up in business performance, not just in productivity checks. Maybe the issue is less about model quality and more about management.
Sanjeev Vohra, Genpact’s chief technology and innovation officer, said enterprises have become effective at deploying AI agents, but running them is the hard part. That view fits the data. The research suggests many firms are still treating agents as tools added to old processes, rather than as part of a redesigned operating model.
Agents take on more decisions
AI now leads or fully handles 40% of decisions, more than double the share from a year earlier. People remain involved at high rates. At each decision stage, humans make or audit the call in at least 87% of cases. Among respondents with agents in production, 92% reported unchanged or higher headcount.
That does not point to simple replacement. It points to a division of labor that many companies are still working out: where agents act, where people decide or check the work, and which outcomes the combined workforce owns.
What leaders do differently
Genpact identifies a small group of AI Leaders, about 17% of respondents, at the top of its maturity index. They tend to have CEO-level support and connect operations, governance, and financial ownership. Most firms have not yet linked all five building blocks the study assessed.
The practical takeaway is not that agents are failing. It is that deployment can run ahead of the operating changes needed to turn AI into profit. For now, the 79% figure is real, but so is the 15% one.