Bain & Company’s survey of 951 global companies finds that while 37% targeted cost reductions of 11% to 20%, nearly 40% of those who measured outcomes landed in the 0% to 10% bucket instead. The technology worked, but the value did not arrive as expected. 90% of those same companies are now increasing their budgets again, this time for Artificial Intelligence agents that will operate with even greater autonomy, complexity, and consequence.
Only 7% of companies are running fully autonomous agents in production today. The dominant model, cited by 38% of respondents, is human approval required. Another 32% operate with guardrails and exceptions, meaning a human steps in whenever the agent encounters something it cannot handle confidently. Only 38% of companies that fell short have agents at guardrails-level autonomy or above, compared with 50% of those that delivered. The gap between the business case and the operating reality has become a central financial issue.
When asked how they plan to fund generative Artificial Intelligence and agentic Artificial Intelligence investments, 44% of companies, the largest group, cited savings from prior automation programs. The prior wave underdelivered, the savings pool is smaller than assumed, and the investment case for the current wave was sized against projections rather than actuals. Companies that do not validate reinvestment math against actual automation returns are compounding risk rather than managing it.
Data access and integration is the single biggest barrier to Artificial Intelligence progress, cited by 41% of respondents, above compliance concerns, budget, skills gaps, and executive buy-in. The companies that delivered on their targets cite data as a bigger barrier than those that missed, 44% compared with 40%. Stronger performers have made data access, governance, workflow redesign, and accountability CEO-level issues, while underperformers more often point to budget limits, lack of a Center of Excellence, and competing priorities.
The recommended path is organizational discipline before another budget cycle: pay down workflow debt, audit returns from prior automation programs, assign clear governance ownership, use Artificial Intelligence where data is already bounded and accessible, and redesign human roles around agent-led operating models. Amazon’s Finance Technology team used a generative Artificial Intelligence solution for valued-added tax regulatory updates; what previously took tax teams 26 minutes per regulatory update now takes 2 minutes, a 92% reduction, with 80% of the Artificial Intelligence generated summaries accepted without modification by human experts.
