Leadership’s role in driving agility

The ability to operationalize AI and data-driven strategies hinges on having the right leadership in place. CFOs and operating partners must prioritize agility—not just in processes but in decision-making and organizational design. Leaders who understand how to drive agility within finance functions can accelerate the adoption of AI, improve data governance, and ensure that technology investments deliver measurable ROI.

The success of AI initiatives continues to depend on data readiness. Core obstacles—fragmented ERPs, reliance on spreadsheets, inconsistent definitions, and weak governance—prevent most pilots from scaling. And the Gartner study shows the governance gap plainly, finding two-thirds of people already use AI regularly, yet only 46% are willing to trust AI systems, underscoring the need for clear policy, training, and explainability in finance contexts. At the same time, momentum is undeniable. 96% of CFOs say they have prioritized AI integrations in 2025, even as they grapple with trust and control requirements.

Real-world automation application

Data automation, while often a prerequisite for implementing a functional and integrated AI system, can deliver tangible ROI even without AI implementation. For example, Highspring is implementing an API to automate daily transfers between a job management platform and an ERP. The automation is expected to save 80–120 hours of manual data transformation and reconciliation each month and cut the monthly close time from more than 20 days to 10.

Practical use cases for AI/ML

Customer-level P&L: Broader research on pricing strategy demonstrates the power of even modest pricing improvements on profitability. One famous analysis from the Harvard Business Review found that raising prices by just 1% can yield a profit increase of 8–11%, depending on elasticity and cost structures. While this isn't strictly a customer P&L powered by machine learning (ML), it highlights the financial impact of small unit‑economics insights—exactly what customer‑level P&L is designed to uncover.

Dynamic forecasting: A recent industry case shows that ML techniques significantly enhance the accuracy and reliability of budget forecasts, supporting better financial planning. A literature review from FP&A Trends reports that organizations adopting AI for anomaly detection within financial forecasting experience improvements in forecast quality, proactive planning, and faster decision-making.

Anomaly detection: ML models can automatically flag inconsistencies, errors, and unusual transactions faster and more accurately than traditional manual review methods. These systems enhance operational efficiency by allowing finance teams to focus on investigating high-risk anomalies rather than combing through large volumes of routine data.

ROI considerations

High-performing finance functions are already realizing tangible benefits from AI in areas such as intelligent process automation, anomaly detection, forecasting, planning, and AI-augmented close and reconciliation workflows. Emerging research from the World Economic Forum also suggests that 32–39% of finance work is fully automatable, with another 34–37% able to be meaningfully augmented by AI. Forbes projects financial services AI spend to reach $97 billion by 2027, and with executives increasingly viewing technology as a source of growth rather than merely efficiency, the momentum is undeniable.

However, not all AI initiatives deliver on their promise. Gartner cautions against deploying generative AI tools without clearly aligned use cases, warning that organizations that overfocus on generative AI, apply it in the wrong contexts, or lack strong data governance—especially for packaged AI solutions—face a higher risk of failure. Many finance teams have experienced the trough of disillusionment, where pilot projects are launched without clear metrics or scalable plans, causing them to be abandoned.

An effective approach for portfolio CFOs is a crawl-walk-run adoption model. In the crawl stage, focus on foundational steps like centralizing finance-critical data, improving data quality, and implementing basic automation in high-impact workflows. In the walk stage, layer in targeted AI applications such as dynamic forecasting in a single business domain or anomaly detection in accounts payable, integrating these into existing FP&A and controller processes. The run stage is about scaling proven models across multiple domains, embedding them into decision-making, and formalizing AI governance and risk controls.