AI in FP&A: What does the future hold?

AI is helping finance teams generate more accurate forecasts, automate workflows, and get real-time insights. But implementing AI isn’t as simple as paying for a fancy AI tool and sending your team a memo about it.

Before using AI in FP&A, you need to know the challenges related to data, skills, and trust. In this guide, we look at the future of AI in FP&A, how you can start using it, and ways to overcome common challenges.

AI in FP&A: Where are we today?

AI adoption has become a key part of the boardroom agenda, but adoption is still uneven.

A 2024 Gartner report suggests 58% of finance functions have used AI to some degree.

Translation? Almost half the businesses aren’t rushing to adopt AI in FP&A. But that’s just one of the surveys.

Another recent survey suggests adoption is even lower, and only 23% of FP&A professionals use AI regularly. The ones who do use AI in FP&A prefer to be careful about it.

As a user mentioned in a Reddit thread, many companies prefer to pilot with tools like GPT for VBA scripting or auto-generating variance commentary.

But not everyone is taking things slow. Some companies with a higher level of AI maturity move to more advanced use cases, such as agentic AI and ML-based forecasting. In fact, Gartner predicted in 2023 that by 2026, 80% of large enterprises will run internally managed generative AI platforms and train them using proprietary financial data.

Challenges impacting AI adoption and how to overcome them

The predictions above may or may not materialize because there are several limitations to deal with before you can effectively use AI in FP&A. Here are some of the most common challenges to AI adoption in finance:

• Data gaps: Fragmented data silos are the biggest problem most adopters face. Silos usually result from data quality problems and a lack of integrations available for legacy systems.
• Skills: Most teams aren’t well-prepared for AI deployment. Some teams have a data fluency problem, while others lack the technical skills to use AI tools.
• Culture: Finance teams are often risk-averse. They rely on established processes, and introducing AI makes them rethink workflows and decision-making hierarchies, which can trigger them to resist.

How to remove barriers to AI adoption

While there are challenges you might face during adoption, they’re not impossible to deal with. Let’s look at how you can overcome these challenges.

Improve data quality and integration

The most common bottlenecks? Legacy systems and fragmented data.

If you have a data problem, start by centralizing your data architecture. This includes consolidating data from ERP, CRM, treasury, and other systems into a single repository.

Next, standardize naming conventions, schemas, and formats. This allows models to make sense of data that they pull from the repository.

Once your data is ready, you can feed it to a tool like Drivetrain to analyze it. For example, you can create an expense analysis dashboard on Drivetrain to monitor it in real-time and flag anomalies before they pose a greater risk.

Upskill for the future

Not every analyst needs to become a data scientist for your AI implementation to succeed. But they do need to become well-versed with tools like Power Query and AI-enhanced dashboards to automate repetitive tasks and get insights from data faster.

Remember, your team doesn’t just need technical skills to fully tap into AI’s potential. They also need business acumen and collaboration skills. They should be able to translate AI’s output into actionable recommendations and explain what’s driving changes in numbers.

So focus on both technical and finer skills when investing in training programs. For example, you could mix technical training with mentoring and peer learning.

Help your team work through their resistance to change

When you first mention AI, some of the people on your team might become worried that AI tools are going to replace them. This is natural, and it’s why the way you talk about AI matters. You need to make it crystal clear that AI is a supportive tool, not a substitute.

To keep your team’s anxiety at bay, start the implementation process with a few team members open to trying new tools. Let them use the tools for a while and experience their benefits.

When your team reduces the time taken for month-end close from four days to one, they’ll tell their colleagues about it and make the benefits more tangible for others.

This approach also makes your team feel more involved than if you just roll out AI from the top down. However, you still need to be patient with your team. Offer them support and communicate openly as they become more accepting of the new AI tools.

How AI is reshaping FP&A

As companies gradually become comfortable with AI, a few noteworthy trends are emerging that give us a preview of the future of FP&A.

• Predictive analytics and forecasting: Machine learning augments your team’s capability to analyze data, which now saves FP&A teams hours they otherwise spent on building complex models. It can build scenario models in seconds and help you establish relationships between drivers like sales and macroeconomic factors.
• ERP and cloud integrations: AI tools are mostly cloud-based and integrate with all the popular ERPs, CRMs, and data lakes. This eliminates the need to manually move data, which can often lead to errors.
• AI Copilots and self-serve analytics: AI copilots like Microsoft Copilot and Power BI Q&A are making it easier to integrate natural language analytics into FP&A workflows. They allow your team to run ad hoc queries like “What drove the spike in Q2 cost of goods sold (COGS)?” without requiring technical details.

AI will introduce new workflows into your day-to-day FP&A

AI transforms how your team accesses data and extracts insights from past data. That means it will change three things in your team’s day-to-day workflow:

Forecasting

AI tools automate forecasts and make them more accurate. This means they allow you to skip the boring parts of the forecasting process and jump straight to analysis.

Picture this. You sit at the desk and open your laptop screen. Instead of updating a spreadsheet while sipping on morning coffee, you see an auto-updated spreadsheet.

Your AI tool has refreshed it with overnight data from sales and other sources. At the bottom, you see flagged variances and suggested actions in order of impact.

There are also a few scenarios the AI tool prepares for you to analyze. You click into a scenario, “10% rise in raw material costs,” and the AI tool instantly shows how it will impact gross margins and make the budget shift.

Now, take a minute to think about how quickly this tool will help you get things done.

Self-service analytics

With an AI tool available, your non-FP&A colleagues as well as your team will be able to get answers quickly. Instead of wasting time finding information, the AI tool will scan the data and provide an answer within seconds.

For example, if your colleague wants to know which business unit had the higher cost per lead last year, they can ask this question to your AI system, which will search for an answer in your internal database and respond to the query in plain English.

This democratizes analysis and allows teams to move faster. It also enables collaboration; budget owners and department heads can evaluate what-if scenarios independently.

AI-augmented decision-making: striking the balance

AI can make data-driven decisions, but doesn’t understand risk, and you can’t hold it accountable. The best way to manage risk and ensure accountability is to go the hybrid route, where AI supports strategic decisions, but there are certain decisions that always require human sign-off.

Where CFOs draw the line

Most CFOs would agree that even with sophisticated AI models, there are decisions that require human sign-off. And it’s easy to see why. You can’t let an AI model decide if you should acquire a company based on a model it built automatically.

Users have shared various experiences of auditors being skeptical of signing off on AI decisions. For example, a user mentioned in a Reddit thread, “We installed the automations, but the problem is that we are now required to have so many controls to review the automation (SOX and legal require human sign off)...”

This reflects how AI can streamline many things in FP&A, but finance leaders and teams are defining clear boundaries on what it shouldn’t do. Like everywhere else, a hybrid model is the best way to use AI in FP&A.

Risk, regulation, and responsible AI in finance

Regulators and internal governance bodies understandably demand higher transparency around AI-driven decisions in finance. If a bot is making budgeting or capital allocation decisions, regulators need to know.

In Europe, the EU AI Act of 2024 and the GDPR require explainability for high-risk automated systems like financial models because they can impact legal and significant business outcomes.

Regulators in the US have also signaled their growing interest in explainable AI (XAI). SEC Chair Gary Gensler spoke at length about the risks of AI-driven decision-making in his 2023 speech at Yale Law School, making clear his expectation that companies using AI must ensure transparency and accountability in how models are trained and governed.

While rules are clearly beginning to emerge, regulatory standards in general have not kept pace with AI’s rapid growth. This means that CFOs need to understand the concerns about ethical implications of AI and establish accountability before they implement it in their decision-making.  

The key takeaway here is that CFOs will need to be able to show auditors and regulators how the AI models and tools they use arrive at their conclusions. So before you start using your first AI tool, remember that your team must be able to answer not just what the algorithm recommends, but why. 

Future risks few discuss

Have you noticed that most of the discussions you hear about AI focus on automation and efficiency? Several risks never find a mention in AI conversations because they sit outside the tech itself.

Over-trust in AI outputs is one such example. As AI automates forecasts and recommendations and finance teams gain confidence in them, they might stop questioning those results.

That’s dangerous. Historical data can’t predict geopolitical shocks or black swan events. Over-reliance can dull your team’s instinct to challenge AI’s assumptions.

Data lineage decay is another under-discussed issue. The original source and its reliability can get lost as models train on data layered from multiple systems. And establishing accountability is tricky without a clear data trail.

Then there are model monopolies. Think about a future where a few dominant vendors control financial decision-making logic across industries. If these models themselves are flawed or biased, the ripple effects could be systemic.

There are no immediate solutions to these challenges, but staying close to AI governance conversations is a good way to prepare yourself for these risks.

An AI action plan for forward-thinking finance leaders

You don’t need a 12-month roadmap to get started. Instead, you need a pilot that helps you realize value as quickly as possible. The process of realizing value may look different for every company, but here’s a general guide to get you started:

1. Pick a low-risk with a high potential ROI

Risk is a top-of-mind concern for many CFOs when it comes to AI adoption. In the early stages of your AI adoption journey, it’s best to identify a few low-risk projects to start with. 

You won’t have to look far for ideas. There are several generative AI use cases that finance teams can adopt now, and more are sure to emerge. 

Make sure to keep a human in the loop for every project, even those where risk appears to be minimal. If risks emerge that you didn’t see before your implementation, this will help catch them early. It will also be very important for learning and adapting your implementation along the way. 

ROI is important, too. The higher the ROI, the more visible your project will be, which if successful, can encourage broader adoption. 

Find a process where the level of manual work involved is high. Implementing AI for this kind of work offers a greater potential ROI in terms of time saved but also in reduced opportunity costs associated with those kinds of processes. (Imagine what you can do when you get those hours of your analyst’s time back.) 

2. Assess your team’s readiness

The success of your pilot will be highly dependent on how prepared your team is going into it. As with any kind of project, you need to clearly define roles and expectations.   

You can expect to encounter resistance along the way. Identifying internal champions from finance and IT can help to effectively manage it.  

To ensure everyone on your team is ready, provide training on the necessary skills, such as data literacy, AI prompting, and how to evaluate AI results.  

3. Evaluate tech and data infrastructure

Automation is a key component of implementing AI successfully. So, identify where all the data you need for your use case resides and whether those systems will integrate with AI-powered FP&A tools (or vice versa) to facilitate automation. You also need to evaluate your data to determine if it is structured and reliable. 

‍FP&A software is innovating quickly to help deliver on the transformation that AI in finance promises. Evaluate carefully, though, as platforms are using different approaches to incorporating AI into their products. 

Some are “AI-enabled”, meaning they were not designed for AI but have been  adapted to incorporate AI capabilities. Others, like Drivetrain, are “AI-native” systems, designed from the ground up with AI as a foundational element of their underlying architecture and capabilities. 

Whether to choose an AI-enabled or an AI-native platform ultimately boils down to what you can and can't do with a tool. However, given how quickly AI is advancing, AI-native tools may be able to innovate more quickly.  

4. Define what success looks like

Set measurable goals before you start to avoid guesswork later. This will not only help you show the ROI of your AI project but also identify ways to improve it along the way. 

You can track time saved, errors reduced, or days taken for the month-end close and establish a baseline to confirm if the pilot was successful. 

Start monitoring your KPIs as soon as you implement (weekly is ideal), and collect feedback from your team early and often. 

5. Launch fast, learn faster

Keep the pilot brief and fast-moving, and be ready to scale based on results. 

Approaching AI adoption with this mindset will help you stay agile and able to adapt more quickly as new innovations happen and new use cases emerge. 

The future of AI in finance is unfolding quickly. The time to begin is now.