AI and the finance function: from data entry to strategic steering

A financial controller who consolidates spreadsheets, prepares a monthly report, and then arbitrates a budget priority is performing three activities that have almost nothing in common when it comes to automation. The first is mechanical work, largely automatable today. The third involves a judgment call about company strategy that no model replaces anytime soon. The finance function may illustrate better than any other why you need to stop reasoning at the job level and start reasoning at the task level. Here’s where the line falls, and what it changes for how you organize your teams.

The finance function is no exception: it isn’t a job, it’s a stack of tasks

From accounts payable to the CFO’s office, every role in the finance function mixes repetitive execution work with judgment work. The difference between roles isn’t the nature of the tasks they contain, but their relative proportion. A junior accountant spends most of their time on execution; a CFO spends most of theirs on arbitration. Treating the finance function as a homogeneous block underestimates the automation potential of execution-heavy roles and overestimates that of steering roles.

What shifts first: entry, reconciliation, consolidation

A substantial share of finance function time concentrates on activities with a stable, verifiable pattern:

• Data entry and bank reconciliation — a systematic comparison between two data sources, with no room for interpretation once the rules are defined.
• Multi-entity spreadsheet consolidation — mechanical aggregation work, a recurring source of human error and close-cycle delays.
• Recurring regulatory reporting — standardized formats, produced on a fixed calendar, from already-structured data.
• First-level consistency checks — flagging discrepancies or anomalies against known rules, before handoff to an analyst.

What stays at the core of the role: judgment, arbitration, relationships

By contrast, some finance activities resist automation because they involve judgment that goes beyond the available data:

• Budget arbitration between competing priorities — a decision that factors in internal political and strategic stakes, not just numbers.
• Negotiation with financial partners — banks, investors and auditors expect a trust-based relationship built over time.
• Analysis of atypical situations — an acquisition, a restructuring or a litigation case follows no repeatable standard model.
• Financial communication to governance bodies — translating numbers into a strategic narrative for a board remains an exercise in judgment, not calculation.

Automating the close doesn’t make the finance function less strategic. It finally makes it able to be — provided the freed-up time gets redirected to the right place.

The close as a laboratory: what a diagnostic reveals about your cycle

The monthly close cycle is a particularly revealing testing ground for a role-by-role diagnostic: it concentrates, over a short and recurring period, an unusually high share of the finance function’s most automatable tasks. Decomposing the roles involved in the close, checking each task against a reference library, then quantifying the freeable time in full-time equivalents gives a concrete measure — often several days of close reduced to a few hours of review — instead of a vague promise of “a faster close thanks to AI.”

From financial controller to decision steward: a role redesign, not a headcount cut

The most poorly anticipated risk isn’t headcount reduction in the finance function — it’s the silent, poorly managed transformation of the roles that remain. A controller relieved of manual consolidation doesn’t automatically become more strategic: they become more strategic if the role is explicitly redesigned to include more analysis and more support for operational divisions, with the corresponding skills developed to match. Without that explicit redesign, the freed-up time simply dissolves into lower-value work.

Three horizons for the finance function

Transforming the finance function is managed in phases, not a single switch:

1. 0–6 months — automate entry and reconciliation. High-potential, low-risk tasks are handled first, with reinforced quality control during the transition.
2. 6–18 months — redesign FP&A and controller roles. Job descriptions evolve to explicitly include more predictive analysis and more support for business divisions.
3. Beyond 18 months — reorganize the function around strategic steering. Hiring and training plans shift toward analysis and modeling rather than accounting execution.

The finance function doesn’t become less necessary as AI advances. It simply becomes less defined by data entry, and more by what it does with what’s left.