Let AI take the wheel: how augmented intelligence is transforming F&A

Artificial intelligence (AI) is fueling innovation across industries with the rise of self-driving cars attracting much attention. Tech companies and car manufacturers have been making great strides in advancing autonomous vehicles, with the promise of fewer accidents due to fewer distracted drivers, and smarter, quicker driving routes thanks to real-time traffic information.

In the fast-paced world of finance and accounting (F&A), AI can also help CFOs and their finance teams steer their organizations with improved agility, sharper insights, better-informed, data-backed decisions, and reduced risk and costs.

For those interested but still hesitant about adoption, uncertainty tends to be the problem. How can AI address the top F&A problems and questions?

Where AI presents the greatest benefit is in work with high-volume records, data, and analytics, elevating CFOs and finance teams to deliver greater value to their business. For example, it would normally take people considerable time to process hundreds, if not thousands of documents, to close the books or run a financial forecast. Today, AI can review large data sets to connect the dots, identify patterns, and easily produce results and new intelligence.

With AI performing more time-consuming transactional work, F&A teams can use the analysis and insight to get better outcomes. This is augmented intelligence – where the combination of human with machine intelligence delivers real business results, such as growth, profitability, competitive advantage, and customer satisfaction.

Instead of reviewing line after line of financial documents, people can refocus and dedicate more of their time and resources to looking at the outputs from AI to guide their business in the right direction. Likewise, CFOs can become strategic partners, aligning finance functions—including the technology setup, reporting, KPIs, goals and ongoing day-to-day executions—with the overall business strategy.

For example, a large retailer in Mexico manages around $18 million of foreign exchange in its reserve to cover imports. The amount of foreign exchange and capital expenditures it needs fluctuates based on its ongoing business with international suppliers. Previously, its finance teams would review all purchase orders to calculate whether the retailer had enough in its reserves or not.By applying machine learning, its system today not only reviews and reports on current reserves, but also uncovers patterns to predit how much foreign exchange it will need.

With more accurate forecasts, teams can make sure their reserves aren't too close to the baseline or, in contrast, aren't too high so any surplus can go back into the business. They can also determine how to minimize capital and share the insight with the company's stakeholders – going well beyond just financial reporting.

Another way AI can take on transactional work and elevate F&A personnel is on invoice exceptions in accounts payable. While robotic process automation (RPA) is effective at rules-based, high-volume automation, such as supplier invoice and receipt matching, there are exceptions where a bot can't finish the job. In these cases, you need intelligent, multi-dimensional matching.

Previously, a multinational retail organization had eight million exceptions to manage, which required 500 people to correlate invoices and receipts. Given the large volume of exceptions, the company couldn't address every case, possibly leaving money on the table. By using AI, the company could automatically reconcile its exceptions, generating new claims to recover lost money. It also freed people to focus on more valuable work. And having uncovered new patterns and intelligence, the team created a negotiating role to reassess contracts with suppliers.

AI is also transforming the traditional financial close process. Rather than spending five-to-10 days scrambling at the end of each month, you can close the books on demand and have access to real-time data for decision-making. There has been remarkable success in automating reconciliations and inter-company reporting where AI can automatically match 60% of records and draw patterns from past data to identify issues in reconciliations. AI now allows finance teams and CFOs to solve more complex problems and uncover opportunities.

Any car – autonomous or not – is only as good as what's under the hood. It takes a good engine, transmission, radiator, battery, and brakes to have a fine-tuned vehicle. AI applications have essential requirements before they can move forward, too (figure 1).

Figure 1: Key requirements before implementing AI

1. Have a focused purpose: First, identify where AI can really transform the finance function and deliver continuous value. If there are critical processes that consume people's time, involve lots of documents, or are too complex or variable for standard RPA, bring in AI. By analyzing structured and unstructured data, both internal and external, AI also surfaces insights that can make decisions more accurate.

2. Establish robust data management and governance: AI is only as good as the data that it has to work with. With a centralized data foundation, different functions and people work with the same, consistent data sets. But you also need people with data engineering and master data management skills to create and maintain the pipelines going into the lake so that your data is clean and comprehensive.

3. Eliminate bias: AI bias can creep in when decisions made by AI reflect the conscious or unconscious values of the people who designed it or data it's based on, for example, when finance teams make decisions on customers' credit or payment terms.

Our latest study, AI360, shows that 78% of consumers expect companies to actively address bias. Start by recognizing its sources: old data, underrepresented data samples, unwanted influence from business cycles, and assumptions in model selection. Make sure you use current, holistic data sets to create your AI models and understand the behavior of data-business cycles. Continue to monitor the models and test to see if assumptions hold true. And to avoid AI becoming a black box, ensure you can trace the reasoning path behind AI-based decisions.

4. Think through change management: For AI deployment to go off without a hitch, you need to manage the change with your F&A teams. Leaders can minimize bumps in the road by communicating how AI enhances their day-to-day jobs, in addition to enabling them to take on more important roles.

For example, for retail firms, predictive insights and intelligent recommendations help teams quickly and accurately understand the implications of price markdowns on revenue and profit margins so they can make faster and more effective decisions. CFOs can set the example by using AI-generated insights to guide their own strategic choices.

5. Find and nurture the right talent: Applying AI to F&A creates new demands for teams with both business and technical skills. People need industry and functional knowledge to provide essential context and review algorithms. Advanced teams are even hiring behavioral scientists and anthropologists. But they also need technical skills, such as forecasting, data scientists, and engineers, analytics, design thinking, and agile programming. Once you have the right people, they need the right infrastructure to work with. With easy access to intuitive technology at home, a workplace with outdated, clunky systems won't encourage them to stay.

And you need to nurture your talent. Teams once responsible for transactional work may require re-skilling in how to collaborate with AI and use its outputs when, for example, negotiating contracts. According to our study, 75% of workers say they are willing to learn new skills to take advantage of AI. So make sure you provide learning resources.

6. Accelerate your approach: Realizing the benefits of AI can take time, but you can speed things up. Rather than redesigning entire systems and processes, you can take a modular approach using pretrained AI accelerators. Find solutions that use insights unique to your industry and can plug and play into core business processes to improve experiences, accuracy, and efficiency at previously impossible speeds.

While autonomous cars have yet to become mainstream forms of transport, AI is a practical solution for finance functions today – and it can be a strategic asset with the proper considerations for data, bias, change management, talent, and acceleration. As transactional work no longer stalls efficiency, F&A teams can use augmented intelligence to improve decision making. AI-powered prescriptive engines that spot patterns and make predictions and recommendations allow finance teams to address strategic business questions. With AI, CFOs are helping their companies accelerate past the competition and secure continued growth.