AI’s Untapped Potential in Finance: A Five-Step Guide to Implementation
While AI is revolutionizing various sectors, its adoption in finance departments, particularly among mid-sized companies, is lagging. A recent report, “AI in the Finance Function,” by Kleer, reveals that nearly half of the surveyed companies have yet to integrate AI tools into their finance operations. However, the desire and belief in AI’s potential are strong, with eight out of ten companies anticipating AI becoming standard practice within a few years. The hope is that AI will free up time for finance professionals to focus on more strategic initiatives.
So, why the slow uptake? Pontus Björnsson, CEO of Kleer, suggests it’s not due to resistance but rather a lack of time and expertise, coupled with the expectation that existing financial system providers will solve the AI integration puzzle. “The step to incorporating AI into finance processes feels significant,” he explains.
To bridge this gap and ensure AI implementation is successful, Björnsson outlines five key steps:
1. Map Out Time Consumption
AI agents can work tirelessly, making them ideal for repetitive tasks. The first step is to identify where the finance department spends its time. “If you want to use AI to facilitate work, you have to start where it makes a difference,” Björnsson advises. “It’s important to find clear ‘quick wins’ to both create time and foster a positive experience.”
2. Solidify the Foundation
Standardized processes, centralized data, and real-time data flows are crucial prerequisites for AI implementation. “I usually say that an AI agent is like a very ambitious intern who knows nothing about the business,” Björnsson notes. “You have to be very clear about how the process looks and what needs to be done.”
3. Choose a Limited Test Case
Starting with a specific use case, such as supplier invoices or variance analysis during financial closing, increases the likelihood of success. “You don’t want to test big and be disappointed,” Björnsson cautions. “An AI agent needs to be trained. If you do something small that saves two hours a week, you are much more likely to continue.”
4. Run a Pilot with Human Oversight
AI agents require monitoring, training, and verification. A skilled human is needed to guide the AI and identify areas for improvement. “AI is neither human nor code,” Björnsson emphasizes. “Code always does the exact same thing, AI does not. Therefore, you must have a human in the loop in the beginning and verify that it is correct.”
5. Build for Organizational Scalability
Once the pilot yields results, consider how to scale the solution across the organization. While leadership support is essential, Björnsson believes ownership should reside with those working within the processes. “If you have domain expertise, you can instruct AI yourself, without going through IT,” he says. “People feel ownership and become curious, and then it becomes a natural part of the company culture.”
The Importance of the Right System Partner
Selecting a system that allows AI to become an integral part of the team is fundamental. Kleer’s all-in-one financial system, designed for SaaS and consulting companies, offers open APIs and is already highly automated. “In an age of AI, a consistent data structure becomes very important,” Björnsson states. “Then it is easier to collect everything related to finance in one system and let an agent talk to and find information via open APIs.”
Björnsson concludes, “The finance function has been built around monthly and annual cycles. With AI, we can finally leave that behind and get a real-time economy, where recurring tasks are automated and people can focus more on analysis and strategic work.”
Enjoyed this post by Thibault Helle? Subscribe for more insights and updates straight from the source.
