Most financial firms can describe their AI program in detail, but only a few can put a dollar figure on it. A review of earnings calls from large U.S.-listed financial firms found that analysts asked for a specific dollar figure on AI savings many times and rarely got a good answer on AI impact. Many AI programs report savings that never reach the budget. Closing the gap starts with knowing which line item each use case is supposed to move.

AI Adoption is widespread, but financial impact is uneven

The 2026 Global AI in Financial Services Report, from the Cambridge Centre for Alternative Finance at Cambridge Judge Business School, surveyed financial firms, vendors, and regulators. Of the financial firms surveyed, 81% said they were adopting AI at some level, but only 14% described AI as transformational to their strategy and competitive position. Among the firms surveyed, 40% reported higher profitability from AI and 43% reported no change. The same study found that 55% of industry respondents find AI value hard to measure. Among large institutions, that figure rises to 76%.

Where AI Is Delivering Financial Impact, and Where It Is Still Catching Up

Retail and commercial banks, investment banks, wealth and asset managers, insurers, and payments firms are concentrating their AI work in a familiar set of areas: back-office operations, lending, fraud, client service, adviser and banker support, and software engineering. Agentic AI in financial services has already moved into production in several of them. In the Cambridge study, four of the five most common use cases were back-office functions. Respondents saw the largest productivity gains in technology and data functions (79%), followed by back-office operations (75%) and front office roles (69%). Those gains have been slower to reach earnings. Across industries, McKinsey’s 2026 research found that only 37% of organizations report a positive enterprise level earnings impact from AI. Public disclosures from banks and insurers help explain the gap, because use cases differ in how far a gain has to travel before it appears as a dollar figure.

Gains that reach a number already reported

Fraud prevention and software engineering use cases show the clearest financial results, because the gain lands in a figure the business already tracks. At one Australian retail bank, customer fraud losses fell more than 20% in the first half of its 2026 fiscal year compared with the prior year. Its agentic system proposes new detection rules, and fraud analysts approve each rule before it goes live. At a global bank, 14,000 engineers now use an agentic coding tool, and technology vendor spend fell 30% year over year as more work moved in house. Neither result needed a new metric, since fraud losses and vendor spend were already on the management report. Our Enterprise CTO’s Guide to AI-DLC and Spec Driven Development covers how engineering teams can capture these gains with the governance a regulated firm requires.

Gains that show up as speed

Lending, operations, and client service use cases produce the largest cycle time improvements. A large Canadian retail bank cut mortgage and home equity pre-adjudication from an average of 15 hours to under three minutes. The system drafts summary memos, and underwriters still make the credit decision. A midsized U.S. commercial bank now completes loan reviews 35 to 40 percent faster on a legacy portfolio and produces an initial third-party risk assessment in about five minutes, with a person approving every piece of AI generated output. At a global bank, AI tools used by service agents in 82 countries have improved client response times by about 20%, and about 100 of its most manual processes have been or are being reengineered.

These improvements are meaningful, but faster work becomes financial impact only when the firm handles more volume with the same team, redeploys capacity to higher value work, or wins business it would otherwise have lost.

Gains still measured as activity

Front office tools have the widest rollout and the thinnest evidence. A large U.S. bank has more than 300 approved use cases, 114 of them live generative AI tools. These include tools that help wealth management advisers and bankers prepare for client meetings and automate research and presentation materials, yet the bank has shared no time, cost, or revenue figures for them. In insurance, a large U.S. property and casualty carrier launched an agentic voice service for auto damage claims in February 2026 and is retraining its call center staff for more complex work, also without published results. The value of a better prepared adviser or a smoother claims call shows up months later in conversion, retention, or claims cost.

Zilbix recommends building measurement in from the start: set a baseline at launch, compare teams using the tool with those that are not, and have finance convert the difference into revenue and productivity impact after the use case goes live.

Why Financial Impact Stays Hard to See

Two issues hold back the numbers across all three groups. The first is that speed is easy to count, while rework, returned files, repeat calls, complaints, and false positives are harder to count and often left out. A faster loan review that sends more files back for correction can look like a saving when the work has only moved somewhere else.

The second is that value and cost both spread across budgets. Lower fraud losses, faster underwriting, and better adviser preparation land in different cost centers, and no single line of the P&L shows the total unless someone owns it. Costs spread the same way, since the full cost of an AI use case includes integration, monitoring, and the time people spend reviewing output vs. just the model cost. Our article on the AI pricing shift looks at how that changes enterprise technology spend.

At Zilbix, we address both issues before a use case goes live. We work with client leadership to agree on the baseline, the outcome measures, and the line item each use case is expected to move, so the impact can be traced to the P&L once the work is in production.

What This Means for Financial Services Leaders

The firms with the clearest results chose use cases tied to P&L impact. In our work with leadership teams, Zilbix sees the strongest returns when a firm concentrates on a small number of priority use cases and gives each one a clear owner and an agreed measure of success. Three questions help a leadership team test where its program stands.

  • Transforming existing processes: Which processes did we redesign end to end, and where did we add a tool to an unchanged workflow? Tools added to unchanged processes tend to produce speed without savings. Our article on building the AI native operating model covers what end to end redesign involves.
  • Human approval: Who approves AI output for each live use case, how often do they change it, and who reviews new use cases before launch?
  • Impact ownership: Which line item is each use case meant to move, and who owns the outcome? A program reported only in percentages is not yet ready to guide budget decisions.

In the Cambridge study, 40% of industry respondents named data quality and availability as the leading obstacle to AI adoption. Budgeting for data work in the business case gives a use case a realistic path to measurable results. For a practical view of turning AI priorities into delivered results, see our guide to AI Strategy vs. AI Implementation.

How Zilbix Helps

Zilbix helps financial services leaders connect AI programs to measurable business outcomes. Our Expert Advisors work alongside client leadership to identify where process redesign is most likely to pay back, establish baselines before launch, and define the oversight each use case needs. The same team that shapes priorities stays to deliver. Our AI Transformation and Business Transformation practices support banks, wealth managers, insurers, and private equity investors from strategy through execution. For the investor view, see The AI Value Creation Playbook for Private Equity Operating Partners, and explore our work across financial services and other industries we serve.

To explore how Zilbix can help, schedule a consultation or reach out at contact@zilbix.com.

About Zilbix

Zilbix is a premier management consulting firm specializing in Business and AI Transformation. Zilbix partners with senior leaders across Fortune 500 corporations, Private Equity backed companies, Emerging Enterprises, and Public Sector organizations to drive complex initiatives from strategy through execution. By combining the agility of a boutique firm with the rigor of global consulting methodologies, Zilbix enables enterprises to accelerate growth, optimize costs, and harness the power of advanced AI to build future ready businesses.

For more insights on AI strategy, enterprise transformation, and the future of agentic AI, follow Zilbix on LinkedIn.