Bain & Company: Why AI Budgets Grow But Returns Don't, and How Amazon Finance Tech Saved 92% of Time

Every year, boards approve larger automation budgets, and CEOs sign off on the next waves of implementation: RPA, machine learning, generative AI, now AI agents. And every year, the expected savings fall short of projections. Not catastrophically, not enough to kill the programs, but consistently, quietly, and by a margin that should be making executives uncomfortable. However, there are those who succeed, like the Amazon Finance Technology team, which reduced tax update processing time by 92% using an AI agent.

Companies pour billions into AI, but often fail to see the expected returns. Budgets grow, ambitions escalate, yet real savings and business value remain elusive. The reason isn't poor technology, but fundamental flaws in approach: automating broken processes, funding new projects with non-existent savings from past ones, and underestimating data and organizational change challenges. These are not just lost dollars; they are missed opportunities and reduced competitiveness. But there's a way out, and it doesn't require magical technology.

Why AI Budgets Grow But Returns Don't: Bain & Company Research

Bain & Company's survey of 951 global companies revealed a troubling trend: while 37% of companies aimed for 11-20% cost reductions through AI, nearly 40% of those who measured outcomes achieved only 0-10% savings. The technology works, but the value doesn't arrive. And instead of pausing to understand why, 90% of these companies are again increasing their budgets — this time for AI agents that will operate with even greater autonomy and complexity.

However, a meaningful group of companies is breaking this pattern. They are realizing their targeted savings, deploying agents with genuine confidence, and funding the next wave from returns that actually materialized. They achieved this not by finding better technology or bigger budgets, but by treating data access, governance, and process redesign as CEO-level problems rather than IT problems. The gap between these companies and everyone else is widening.

Three Uncomfortable Truths Hindering AI Success

1. Agents aren't as autonomous as they seem. Only 7% of companies are running fully autonomous agents in production today. The dominant model (38%) is "human approval required," and another 32% operate with guardrails and exceptions, meaning a human steps in whenever the agent encounters something it can't handle confidently. If your business case was built on full automation, and the reality is a system routing a significant share of decisions to a human, then the CFO approved one set of numbers, and the organization is living with another.
2. The next wave is being funded by returns that haven't arrived. 44% of companies plan to fund generative AI and AI agent investments with savings from prior automation programs. This sounds like discipline, but in reality, it's a circular bet with a structural leak. 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.
3. Data is still the wall. Data access and integration is the single biggest barrier to AI progress, cited by 41% of respondents. Despite a decade of investments in data modernization, the number one reason AI programs underperform is that companies cannot reliably get access to their own data.

How Amazon Finance Technology Achieved Success: A Concrete Example

Instead of waiting for the perfect data solution, leading companies are using AI to solve the data problem itself. Amazon's Finance Technology team did exactly this by implementing an AI agent that tracks value-added tax (VAT) regulatory updates across global markets. What previously took tax teams 26 minutes per regulatory update now takes 2 minutes — a 92% reduction. Moreover, 80% of the AI-generated summaries are accepted without modification by human experts.

This isn't a moonshot; it's a bounded, specific workflow where the data was already there, and AI replaced the manual assembly and processing. This approach allows for quick value generation and uses these early wins to fund more extensive data modernization efforts.

How to Replicate Success: Five Key Decisions

Companies that succeed with AI make a small number of specific organizational decisions:

• Pay down your workflow debt before deploying AI. The single most costly mistake in AI deployment is automating a broken process. AI doesn't fix workflow debt; it locks it in, speeds it up, and makes it vastly more expensive to unwind. Before any AI program is approved, ask: "If we were designing this process from scratch today, what would it look like?" Only then should the technology conversation begin.
• Validate the investment case and name a governance owner before programs launch. Before approving the next wave of AI spending, CFOs should audit actual returns from prior automation programs, not projected returns. If the previous program delivered 60% of its targeted savings, size the current investment accordingly.
• Use AI to solve the data problem. Don't wait for the data problem to be solved first. The fastest path to value is often automating one repeatable, high-value workflow where humans are currently pulling data manually, consolidating spreadsheets, and producing reports, and replacing that entire sequence with AI.
• Redesign the operating model, not just the process. Deploying AI agents without changing how people work around them pretty much guarantees an organization will underdeliver on the business case. In an agent-led operating model, employees are no longer moving work along a process; they are orchestrating, supervising, and making the high judgment calls that agents cannot.
• Measure outcomes at the enterprise level, not the program level. Programs will always optimize for what they were designed to measure — typically cost and hours saved. What matters for the enterprise is whether AI investment is producing better decisions, faster responses, and stronger customer outcomes. If those metrics aren't on the CEO's dashboard, programs will keep delivering the wrong things efficiently, and the value gap will persist regardless of how much the budget grows.

The turning point for most companies is not finding the best AI technology. It’s the moment when leaders decide — before the next budget cycle, before the next vendor pitch, before the next program launch — that they have a personal responsibility to create the organizational conditions for AI success. The window to make that decision ahead of the competition is still open, but it’s narrowing faster than many executive teams realize.

If this case sounds like what's happening in your company, our manager can help: he'll analyze your business and niche for free and point out where an AI agent would bring a real result in your case. Message the manager