CFOs prioritise trust over speed in AI adoption journey

The finance function demands high levels of compliance, and integrity of data and people. But, at the same time, it relies on repetitive processes that take up considerable time and take personnel away from higher value tasks such as resolving complex issues and contributing to business strategy. This makes finance a fertile ground to reap the harvest of AI.

While AI promises great efficiencies and labour savings, CFOs and other finance leaders must recognise the risks that come alongside opportunity.

Rosie Cairnes, the Regional Vice President for ANZ at BlackLine, said there are three main risks finance leaders need to focus on when embarking on an AI journey.

"The first is data readiness. You can't automate intelligently if your underlying data is not clean, harmonised and accessible. Poor data quality is a massive amplifier of risk. Number two is visibility and chain of thought. You can't allow probabilistic AI. You need a deterministic control layer without an element of randomness. Finally, CFOs need visibility into what AI is acting autonomously on. They need to know what data and what controls apply and ensure that human accountability remains."

Cairnes said the goal for finance isn't to slow AI adoption, but to ensure there are controls and guardrails in place. Citing research from Gartner, she said that by 2028 an average enterprise will have over 150,000 agents in use but only 13% believe they have the right AI governance in place.

One of the core principles of compliance is the segregation of duties. This is no different for AI agents. Cairnes said digital agents need to be managed similarly to human workforce. She argued that agents need to be deployed within a broader end-to-end context rather than a standalone feature.

For most organisations, financial data is seen as the absolute source of truth. It is reported to boards at least monthly and appears in official data such as annual reports that are subject to strict governance. But, as the data often comes from multiple sources in a variety of formats, it can be very challenging to ensure the high levels of quality that are required.

"Data is often a mess," said Cairnes. "Finance teams are spending way too much of their time manually gathering data. There's research suggesting 70% of finance time is spent just gathering data."

Although it may seem obvious that AI can offer finance teams a huge productivity boost, Cairnes said building trust is essential. BlackLine's platform is responsible for supporting finance teams with over $56T in market capitalisation.

That means trust is imperative. For finance, trust comes through transparency and understanding how data is being treated and managed, and ensuring the necessary controls are in place.

"It is not enough for AI just to produce answers faster. AI needs to understand how it's getting to those answers and it needs to be able to defend the outcome. Without that there's no trust," said Cairnes.

The automation benefits that come with AI still require human oversight. Like any team member, Cairnes said AI agents must have appropriate authorities, operating boundaries and oversight and review.

"Sometimes we talk about black box AI. The idea that you know these outputs are made without the transparency. We focus on a glass box approach where finance has visibility of AI decisions, actions and controls," explained Cairnes.

Typically, the business case for adopting AI is based on cost savings and efficiency. But in Cairnes' interactions with CFOs, these are seen as side benefits. The biggest benefit they see is having confidence in financial statements and eliminating risk. They want to avoid enforceable undertakings and audit issues.

"For most senior leaders in finance at the CFO level, this is more of a trust premium than just speed and efficiency," said Cairnes. 

Organisations evaluating AI must undertake a realistic assessment of their capabilities. Jumping to conclusions about AI's potential risks overlooks fundamental requirements such as accurate, integrated data, alongside understanding context including chart of accounts, materiality thresholds, and accounting rules requirements.

Successful AI implementation relies on leveraging the technology to automate scalable tasks and freeing up human resources for higher-value analysis of financial data. It is crucial that organisations prioritise data integrity, contextual awareness, and agency to mitigate risks and maximise the value of AI solutions.