The future of finance: From supporting decisions to shaping them

By: Katie Krasnikova and Graham Webster

Most conversations about AI in finance start with the wrong question. They focus on which processes to automate. The real question is: “Which decisions finance is still too slow to influence?”

Finance has historically had two major mandates: to look backward with rigor, providing insight into what happened and why, and to help the business look forward through budgeting, forecasting and resource planning. Both sides benefit from process improvement and automation, and these are foundational investments worth making regardless of broader strategy. But it is the forward-looking mandate where the real transformation lies: moving finance from a facilitative role, where it supports planning processes, to an influencing role, where it actively shapes the business decisions that flow from those plans.

Three forces are driving this:

• Real-time data integration makes it possible to detect performance shifts as they happen, not after the close.
• AI-driven modeling enables finance to test dozens of scenarios in minutes, replacing static assumptions with continuously refreshed, probabilistic forecasts.
• Agentic automation absorbs routine transactional work, freeing finance capacity for judgment-intensive activities like trade-off analysis and strategic planning.

The result is that finance can now operate at the speed of the business decisions it seeks to influence. But realizing this requires more than technology. It demands trusted data, redesigned decision processes and a finance workforce that combines technical skills with deep business understanding, equipped to exercise judgment earlier, more often and under greater uncertainty.

Figure 1: Finance’s shift from facilitation to influence

What’s actually breaking down — and why

Finance has long been trusted to reflect the facts of the business accurately and to facilitate the planning processes that guide it forward. But the environment has changed materially, and the traditional operating rhythm is strained on both fronts, particularly on the forward-looking side where finance’s opportunity to influence decisions is greatest.

Forecasts lose relevance before they are finalized

A traditional forecast requires analysts to gather actuals, reconcile data across systems, build driver-based models and route assumptions through multiple rounds of review. In most large enterprises this cycle can take two to three weeks. That timeline was adequate when the business environment shifted on a quarterly basis, but it falls short when input costs, demand signals and competitive dynamics can move materially within that same window. The problem is that the forecast’s shelf life is now shorter than its production cycle. By the time it reaches decision-makers, the assumptions behind it may already need revision.

Investment trade-offs are made in silos and revisited too late

In most organizations, capital and operational expenditure allocation decisions are made within business units or functions, typically through an envelope model where each group receives a fixed budget and manages spending within those boundaries. These allocations are reviewed annually and adjusted only when performance clearly misses plan. Enterprise-wide trade-offs, where one initiative’s underperformance should trigger reallocation to a higher-returning opportunity elsewhere, are rare not because leaders lack the intent but because the budgeting structure doesn’t accommodate them. That enterprise-wide perspective often doesn’t materialize until a quarterly review, if it surfaces at all.

The shift is portfolio-level simulation. AI models can ingest financial performance data, market signals and operational key performance indicators across initiatives simultaneously, then run thousands of allocation scenarios to identify optimal trade-offs.

Understanding what happened takes too long

In complex, multi-segment enterprises, diagnosing the root cause of a forecast miss can take weeks of investigation across systems, geographies and business lines. Forecast retrospectives are often the most time-intensive example: explaining not just that a miss occurred, but why and what the organization should learn from it. AI-powered anomaly detection and automated variance analysis compress this cycle dramatically, surfacing which drivers contributed, how they interacted and where intervention is most likely to matter.

Transactional work absorbs capacity needed for strategic analysis

In a typical large finance organization, skilled analysts spend a disproportionate share of their time on process-heavy tasks: reconciliations and missed-forecast retrospectives, journal entries and transaction monitoring. These activities are essential but repetitive, and they crowd out the higher-value work the enterprise increasingly expects finance to deliver: scenario analysis, trade-off evaluation and strategic decision support. The problem isn’t that the work is done poorly, but that it consumes the people best equipped to do something more impactful.

Agentic AI workflows are changing this equation. Unlike simple robotic process automation, agentic systems can execute multistep processes, make rules-based decisions and escalate true exceptions for human review. The result isn’t just efficiency, but the reallocation of skilled analysts to work that requires human judgment.

The takeaway: Finance is structurally advantaged to lead this shift. It is cross-functional by design, balances near-term performance with long-term investment and is grounded in evidence and analytical rigor. The question is whether it builds the foundations to act on that advantage. Finance organizations that don’t make this shift risk being bypassed, as business units build their own analytical capabilities and the function’s influence narrows rather than expands.

The foundations that determine whether AI creates value or stalls

The most common failure pattern in finance digital transformation isn’t choosing the wrong technology. It is deploying capable technology on top of weak foundations. Three interconnected pillars determine whether insights translate into better decisions or stall before impact: data readiness, process design and people capability.

Figure 2: Three foundations that determine whether AI creates value

Data is often called a strategic asset, but in practice it functions as a minimum requirement. Without trusted, timely and well-governed data, AI models produce outputs that finance professionals can’t interpret, defend or act on.

What “good data” means for finance is specific:

• Trusted and consistent: Clear ownership, documented definitions, reconciled lineage and a single source of truth.
• Timely: Available at the pace decisions are made, not locked to month-end close cycles.
• Integrated: Combining financial, operational and external signals across domains.
• Accessible and structured for use and reuse: Organized as managed products that are easy to find, extract and connect into models, rather than rebuilt from scratch for each report.

When data is fragmented, delayed or inconsistently defined, AI outputs become noise rather than signal. The first investment in any finance transformation should be an honest assessment of data readiness. This isn’t to achieve perfection, but to identify and close the gaps that matter most for the decisions you are trying to improve.

Process: Connecting insight to action

Strong data and capable models are necessary but not sufficient. If the process through which decisions are made was designed for a world of static, periodic analysis, then faster insights will simply queue up behind the same bottlenecks.

Most finance processes follow a sequential pattern: gather data, perform analysis, produce a report, review it and eventually make a decision. Each handoff introduces delay, and when AI surfaces a material shift in near real time, the existing process has no mechanism to respond. The gap isn’t analytical but organizational: who sees the signal, who owns the response and how quickly the organization can act.

For finance, this means planning cadences must shift from annual lock-ins to rolling, decision-triggered reviews that treat the annual plan as a living baseline rather than a fixed contract. Decision ownership must be explicit: when a model flags a reallocation opportunity, it must be clear who evaluates, who escalates and who acts. Analytics, too, need to be part of how decisions actually get made, not material that informs them from a distance.

Process redesign is less visible than a technology deployment, but it is often the binding constraint. AI creates value only when processes are designed to absorb and act on its outputs.

People: Judgment under new conditions

Of the three foundations, people are the most critical and the most difficult to transform.

AI changes what is asked of finance professionals in two ways. First, judgment is required earlier and more often. When AI generates a probabilistic forecast with confidence intervals, someone must decide which scenarios to present, which assumptions to challenge and which signals to prioritize. Second, the experiential path to judgment is compressed. If AI handles the data gathering, reconciliation and initial modeling that once occupied early-career analysts, those analysts lose the repetitive exposure that traditionally built deep intuition about the business.

But there is a more fundamental capability gap. To influence business decisions, finance professionals must understand the business itself, not just its financial outputs. This means familiarity with products, customers, competitive dynamics and the operational realities that shape performance. A finance team that can model scenarios but can’t contextualize them commercially will remain a facilitator, not an influencer.

The generational dynamic adds complexity. Newer professionals are more fluent with data and technology; senior leaders bring institutional knowledge and the credibility to drive change. The most effective transformations create space for both. But as AI absorbs foundational work, organizations must ask where the next generation of finance leaders will develop their instincts. The answer requires deliberate investment: structured rotations, scenario simulations, mentorship on live decisions and cross-functional exposure that builds commercial intuition alongside analytical skill. Organizations that solve this earliest will have a compounding advantage in finance leadership quality over time.

5 use cases where finance can expand its impact

These foundations come to life in five use cases where finance has the greatest opportunity to expand its influence: AI-powered forecasting, continuous planning and resource allocation, next-best investment decisions, intelligent automation in finance operations and the CFO control tower. The interactive companion to this paper explores each in detail.

Most organizations should not pursue all five simultaneously. AI-powered forecasting and continuous planning are natural entry points because they sit close to core financial planning and analysis processes and surface data and process gaps early. Portfolio optimization and the CFO control tower typically come later, once leaders are comfortable acting on dynamic, model-driven insight. Start where the decision need is clearest and the data is most ready.

Figure 3: Use cases mapped to the value chain

What to do now

These use cases illustrate what is possible. But they share an implication for how finance operates: the function must maintain its rigor on accuracy, compliance and control while building new capabilities on the forward-looking side — commercial fluency, comfort with probabilistic reasoning and the organizational standing to influence decisions beyond the profit and loss (P&L).

Transformation doesn’t require a “big bang.” Start with the choices that build the foundation for what comes next.

1. Redefine how the business sees finance. The enterprise has long treated finance as the team that reports the numbers and tracks the budget, and finance has often organized around that expectation. Changing it requires both sides to move. Business leaders must bring finance into decisions earlier, and finance must build the commercial fluency to contribute meaningfully when they get there.
2. Start where finance’s untapped influence is greatest, not where automation is easiest. Without clarity on which business decisions finance should be shaping, technology investments will default to making existing processes faster. Identify the decisions where finance has strong analytical capability but isn’t yet influencing the outcome, and pilot there.
3. Assess data readiness for decision support, not just reporting. The relevant question isn’t whether data is clean enough to close the books, but whether it is timely and connected enough to inform trade-offs as they arise. Focus the assessment on the gaps that would block the use cases most relevant to your organization.
4. Match the depth of investment to the decision complexity it serves. Data-driven capabilities like forecasting and planning tend to deliver value broadly. For transactional automation, the return depends on the volume, variability and cost structure of the underlying process.
5. Invest in talent and team structures that make influence possible. Data literacy, probabilistic reasoning and the ability to challenge AI outputs can’t be bolted on after deployment. Neither can the business acumen finance needs to advise on decisions beyond the P&L. Build these capabilities deliberately, and structure teams so finance professionals are embedded alongside business partners where they develop commercial intuition through shared ownership of outcomes, not just reporting relationships.

What separates organizations that benefit from AI from those that simply adopt it is straightforward: it isn’t about making finance faster at the same work. It is about enabling finance to do fundamentally different work— influencing decisions as they are being made, managing uncertainty with rigor rather than false precision and shaping enterprise outcomes rather than facilitating the processes that report on them. Organizations that build these foundations position finance where the enterprise needs it most: not as a back-office function that supports decisions after the fact, but as a strategic voice that helps make them.

Read more insights from ZS.

This article reflects our personal views. They do not necessarily represent any official position of ZS.