AI in finance: why adoption is rising faster than value

The executive committee asks how the finance function’s AI shift is going. The answer is reassuring: the licenses are deployed, the teams use them every week. Then the controller asks the awkward question: what has actually improved, in numbers, over the past twelve months?

The silence that follows has a name. KPMG calls it the value gap.

According to KPMG Canada (September 2026), 83% of surveyed Canadian companies have moved beyond the planning phase for broad AI use in finance, yet only 36% report more accurate forecasts, versus 64% globally.

Why is AI in finance producing so little measurable value?

Because most organizations deployed tools without redesigning the processes, roles and controls that turn a prediction into a result. KPMG’s global survey of 1,013 senior finance leaders across 20 countries (March 2026) reaches a blunt conclusion: using an AI tool does not create value on its own.

Value appears when an AI-generated forecast passes through the judgment of a finance professional, supported by trusted data and clear governance. It disappears when ownership is fragmented across finance, IT and data, or when pilots pile up with no path to a business outcome.

Canadian respondents live this daily. 74% use AI in financial reporting and analysis, 68% in financial planning and 64% in risk management (KPMG Canada, 2026). Returns barely follow: 57% say ROI at least meets expectations, and only 19% say it exceeds them.

What do the 2026 numbers reveal about the gap between adoption and value?

They show adoption that is now mainstream among leaders, gains concentrated in a minority of organizations, and frontline teams still largely left out. This year’s published benchmarks fit in a short list.

• 83% of Canadian respondents have moved beyond planning for AI in finance, versus 92% globally (KPMG, 2026).
• 36% report more accurate forecasts in Canada, versus 64% worldwide; for decision-making speed, the ratio is 45% versus 71% (KPMG, 2026).
• 77% of Canadian respondents have moved beyond planning for agentic AI (KPMG Canada, 2026).
• 80% of accounts receivable teams use no AI in their processes, even though more than eight in ten expect further automation (NACM and BlackLine, 2026).
• 12.7% of Quebec businesses used AI for production in the second quarter of 2025 (Institut de la statistique du Québec, November 2025).

A note on method: these figures come from public surveys with different samples and definitions. They give orders of magnitude, not a portrait of your organization.

The cross-reading is uncomfortable. Leaders report record adoption while the teams that process payments still work by hand.

Where does AI create measurable value in your finance processes?

In repetitive, high-volume processes measurable in dollars and days. Cash application is the clearest example. KPMG confirms it: early gains in reconciliations, variance analysis and reporting fund and de-risk the more strategic applications.

The NACM and BlackLine survey (2026) shows how much ground is available: half of accounts receivable professionals name reducing manual, repetitive tasks as their single top priority for the next 18 months. Not an aspiration. Their most urgent need.

The same survey reveals the constraint: not a single respondent said they would be very comfortable letting AI handle customer interactions. Trust is earned on internal tasks first, payment matching before collection emails. Quadient (2026) observes the same caution: agentic automation advances where the rules are defined.

In the accounts receivable mandates PlanAxion leads with B2B distributors, it is the manual sorting of grouped, partial or undocumented payments that eats the team’s hours. The cash application solutions PlanAxion puts in place target that exact point, and the question of who approves what the agent decides is settled at design time.

The value gap does not close with one more tool. It closes one process at a time, with a documented baseline and an owner of the results.

How do you close the value gap in four steps?

By documenting the baseline, choosing a bounded process, framing automated decisions, then measuring quarterly. The sequence fits in four moves.

1. Document the baseline. Hours spent on the process, exception rate, cycle times. Without a documented baseline, no return calculation is credible.
2. Choose a bounded process. One process, one business unit, one result measurable in dollars and days. Cash application checks all three boxes.
3. Frame the decisions. A written scope, human approval thresholds, an audit trail. KPMG finds that 66% of Canadian respondents can efficiently produce AI-related audit evidence, versus 82% globally. That is a direct brake on scaling.
4. Measure and decide. After one quarter, the same indicators settle it: scale or stop.

That is exactly the approach of our Rapid AI Solutions Workshop: 4 weeks, 5 steps (prepare, identify, prioritize, validate the data, decide and deliver), with the same filter applied to every idea: problem to solve, expected value, available data, required effort. The investment varies with scope and is confirmed during a short exploratory call.

Where should you start before the end of the quarter?

Take a single high-friction finance process, often cash application, and document its baseline: hours, exceptions, cycle times. That two-page document will be worth more than the next tool, because it changes the conversation with your executive committee: you will no longer talk about adoption, but about value.

Frequently asked questions about AI in finance

What is the AI value gap in finance?

The value gap is the distance between reported AI adoption and the measurable results it produces. In Canada, 83% of surveyed companies use AI in the finance function, yet only 36% report more accurate forecasts, according to KPMG (2026). The gap closes through process redesign, not through buying more tools.

Why do Canadian companies get less value from AI than their global peers?

According to KPMG Canada (2026), value is constrained by fragmented ownership, pilots with no path to a business outcome and incomplete measurement. Only 64% of Canadian respondents track and act on meaningful AI KPIs, versus 85% globally. Organizational capability matters more than the technology itself.

Which finance processes deliver the fastest AI gains?

Repetitive, high-volume, measurable processes: reconciliations, cash application, variance analysis and financial reporting. KPMG (2026) observes that these early gains fund the more strategic applications. In accounts receivable, half of teams name reducing manual tasks as their top priority for the next 18 months (NACM and BlackLine, 2026).

How do you measure the value of an AI project in finance?

Document the baseline before deployment: hours spent, exception rate, cycle times. Then compare the same indicators after one quarter, and value the recovered hours and faster cash collection. Without a baseline documented before the project, no return calculation is credible.