Two-thirds of finance leaders use AI tools. What are they using them for?

Finance leaders are using AI for more than drafting emails. CFO Connect’s Top CFO Tools Report 2026 found that 67% of respondents use AI-powered tools in their work, with commonly reported valuable applications including analysis, reporting, modelling and reconciliation.

For CFOs, the next step is making those workflows dependable and measuring their impact. That means choosing useful applications, providing the right data and keeping appropriate checks around the results.

This article explores how finance leaders are using AI, how to build reliable workflows and how Spendesk combines automation with AI-assisted access to company spend data.

Key takeaways

• 67% of surveyed finance leaders use AI-powered tools in their work.

• Applications include analysis, reporting, modelling and reconciliation, with outputs checked against source data.

• Start with one measurable workflow, using approved data and human review to assess quality and time savings.

What are finance leaders using AI for?

The survey measures whether respondents use AI, rather than how extensively they have automated their work. Commonly reported valuable uses include financial analysis, reporting, modelling or forecasting, and reconciliation. Respondents also described data queries and workflow automation.

The examples below illustrate how these applications can work. They are practical suggestions rather than additional survey findings.

Financial analysis

AI can help explore a prepared dataset, suggest questions and produce a first draft of variance commentary. For example, a team might ask it to identify departments where expenditure changed materially and propose areas to investigate.

Finance should verify the calculations and investigate the causes. A plausible explanation for a movement is not evidence that the explanation is correct.

Reporting

An AI tool can turn approved figures and notes into a draft management narrative. It can also help adapt an explanation for different audiences or identify questions readers may still have.

Finance teams should check the final version against the underlying numbers, reporting period and business context.

Modelling and forecasting

AI can help explain formulas, challenge assumptions and suggest scenarios. It may also help develop spreadsheet formulas or code that a qualified team member can test.

The model needs independent validation. Fluent output does not demonstrate that a formula is correct or an assumption is reasonable.

Reconciliation and workflow support

AI-assisted workflows can help investigate potential matches, organise exceptions and prepare items for review. Finance teams should set the level of automation according to the quality of the data and the consequences of an error.

A pilot that proposes matches for approval is a different control decision from a workflow that posts entries automatically. Make that distinction explicit before implementation.

The same applies to AI agents: decide which actions, if any, an agent may take without approval and who reviews its exceptions.

Anomaly and duplicate detection

Beyond the uses respondents described, AI can help finance identify transactions that need a closer look. It can flag unusual spending or potential duplicate invoices for review.

A flag may indicate an error or potential fraud, but it is a prompt to investigate rather than a confirmed finding.

Why do general-purpose AI tools appeal to finance teams?

A general-purpose model can assist with several tasks in one environment: explaining a formula, summarising a policy, exploring data or challenging a forecast assumption. That breadth can make initial experimentation easier than introducing a separate application for every use case.

However, a general-purpose model does not automatically know the company’s chart of accounts, approval policy or current financial position. Its usefulness depends on the context and authorised access provided to it.

Connecting an assistant to approved company data can make its responses more relevant, but finance still needs to verify that the information is complete and the conclusions are supported.

How to turn AI use into a reliable finance workflow

Choose one recurring task with a clear review point. Drafting commentary from an approved variance table is easier to assess than a broad instruction to automate financial reporting.

Before testing, record how long the task takes and what a good result looks like. Define the permitted data, approved environment and person responsible for reviewing the output.

Run the pilot alongside the existing process. Compare outputs, record mistakes and measure the effort needed to correct them. Include review time when assessing whether AI saves work.

Follow company policies on data access, retention and permitted use. Give system connections appropriate permissions, and keep a clear record of actions. Check that the team can trace outputs to their source data and record how they were produced.

Expand once the workflow produces reliable results under those controls. Continue monitoring as models, inputs and business requirements change.

Explore how other finance leaders are adopting AI

The 67% adoption figure provides a useful starting point, but the more meaningful benchmark for your team is whether a particular application improves speed or quality while preserving the checks finance needs.

Use peer examples to identify workflows worth testing, then assess them against your own data, processes and review requirements.

CFO Connect’s Top CFO Tools Report 2026 reveals the AI tools respondents use, a breakdown of their most valuable applications and examples shared by finance leaders.

Download the report to discover how your peers are using AI and identify ideas for your next pilot.

How can Spendesk AI help finance teams automate work and analyse spending?

Spendesk combines AI-powered automation within its platform with a connection that lets finance teams explore spend data through their existing AI assistants.

Reduce manual preparation and focus on exceptions

Spendesk’s AI and automation features extract information from invoices and receipts, suggest expense accounts and VAT rates based on transaction history, and flag potential duplicate invoices or unusual spending for review.

Invoice matching connects invoices with purchase orders and, where relevant, delivery notes, helping teams identify discrepancies before approval. These capabilities reduce repetitive preparation work while keeping finance involved in reviewing suggestions and exceptions.

Ask questions about live spend data

Spendesk AI Connect uses the Model Context Protocol (MCP) to connect Spendesk data to AI assistants such as Claude and Dust. Finance teams can ask plain-language questions about supplier spending, overdue payables or open purchase orders, then turn the answers into structured reports.

For example, a finance leader could ask which suppliers account for the most spending over a selected period or request a summary of outstanding invoices. Recurring updates can also be configured through the connected assistant.

Access follows users’ existing Spendesk permissions, and the connection is read-only: it cannot create payments, approve invoices or change records. This gives teams a practical way to bring company spend data into AI-assisted analysis while retaining control over financial actions.

Frequently asked questions

How many finance leaders use AI tools?

In CFO Connect’s 2026 survey of 215 finance leaders, 67% reported currently using AI-powered tools in their work. This is a rounded survey result rather than an estimate for all finance professionals.

What can AI help with in finance?

AI can assist with analysis, report drafting, model development, data queries and parts of reconciliation or other workflows. Suitability depends on data quality, the task, access permissions and human verification.

Does using AI mean a finance team has automated its processes?

Not necessarily. A team might use AI for occasional drafting or research without connecting it to operational systems. Process automation requires defined inputs, actions, controls, exception handling and ownership.

About the data

CFO Connect’s Top CFO Tools Report 2026 draws on a survey conducted in June and July 2026, exploring the tools finance leaders use across core finance functions.

About CFO Connect

CFO Connect is a global community of finance leaders founded by Spendesk, offering research, expert insights and opportunities to learn from peers.