Genpact released research on November 12, 2026 that shows a gap between AI deployment and financial results. The survey covered more than 2,000 senior executives across 17 industries and 14 business functions. Companies ranged from $1 billion to more than $50 billion in annual revenue.
Agents in production, not everywhere
About 79% of surveyed enterprises have AI agents in production. But only 20% use those agents broadly and consistently across functions. I think that distinction matters. Putting an agent into one workflow is not the same as changing how a company runs. The report calls this a problem of running AI inside operations, not just installing it. Genpact’s Sanjeev Vohra said enterprises are good at deploying agents, while running them is the hard part. That line sums up the tension well.
Spending still misses P&L
Only 15% of enterprises have fully connected AI spending to P&L outcomes. On average, 49% of AI initiatives funded in the prior 24 months delivered expected value. Just 9% reported completing substantial changes to their operating models. Those numbers suggest many projects can show activity without showing up in profit. Or perhaps the measurement is still catching up. Either way, deployment is ahead of the operational redesign needed to turn AI into business performance.
Decisions shift, headcount holds
AI leads or fully handles 40% of decisions, more than twice the share a year earlier. Human involvement remains high. At each decision stage, people either make the call or audit it in at least 87% of cases. Among respondents with agents in production, 92% said human headcount was unchanged or higher. So agents are scaling beside employees, not simply replacing them. The report describes autonomy as a division of responsibility: where agents act, where people decide and validate, and which outcomes the combined workforce delivers.
Leaders tie operations to ownership
Genpact classifies 17% of respondents as AI Leaders at the top of its maturity index. These companies have CEO-level mandates and are further along in blending people and agents into one operating model. The report calls the model Agentic Operations, combining AI, process intelligence, human judgment and governance. Most enterprises have not connected all five building blocks the research assessed. That may be the next test. Not how many agents are live, but whether the company has changed enough to measure the result.
