Main content

Planful CEO Grant Halloran - the next ten years of AI will transform the role of finance professionals

Phil Wainewright Profile picture for user pwainewright October 30, 2023
FPM vendor Planful isn't rushing into launching new generative AI features, but nevertheless its CEO Grant Halloran foresees a huge impact on the role of finance professionals over the next decade.

Looking down at shoes facing a choice of arrows pointing to either habits or changes © natasaadzic via Canva
(© natasaadzic via Canva)

CFOs have a reputation for being very conservative when it comes to technology adoption. But the volatility of the past few years has persuaded many to invest in tools that enable rapid planning and forecasting. Despite the current economic uncertainty, that trend has driven growth for the likes of Planful, which provides cloud-based Financial Performance Management (FPM) software. I met Grant Halloran, its CEO, when he was passing through London last week. He tells me:

I often say to people, with the world so dynamic, you just never know what you're going to wake up to on a Monday morning.

It's hard enough running your own company — you've got your own internal issues. The best laid plans don't come off. That product launch is delayed, and whatever else. But then there's all these external factors. Again, we're in another war here that's creating controversy and concern for people. You just don't know what's going to be coming around the corner.

So you have to have these resilient financial systems, back-office systems, that enable you to make better judgements quicker, make faster course-correcting decisions. We've shown at least that, for our systems, demand has actually increased somewhat, because of uncertainty about the economy.

Technology is also raising the bar of what's achievable, in particular when adding Artificial Intelligence (AI) into the mix. For example, Halloran talks about Planful's predictive analytics tool. Tuned specifically to work with financial data, he says it now routinely creates projections that can be shown to be more accurate than those that finance professionals have previously created:

We're now doing full P&L forecasting with exceptionally high degrees of accuracy, compared to the human. We prove it, when we do the POC with a customer — we can prove their projections from a year ago, we run our models across that data, and we're always more accurate.

Not only more accurate than the traditional approach, but also far faster. He says the forecasting model can typically be trained on a company's data in just 20 minutes, and then it takes just 10 minutes to run the projection across the full P&L. This means that instead of spending weeks of effort to build just one projection, a finance team can achieve the same result in a couple days, or create multiple projections and compare the impact. Halloran continues:

If you're running a complex company where you're having to do a monthly forecast, that typically was, across the whole business, might be 1,000+ hours of effort. Now you can at least have a starting point in 10 minutes. So it's dramatically reducing the amount of corporate effort that goes into it, and also compressing that cycle time. You can now get down to maybe a two-day effort to run a forecast. You're more doing quality assurance and checking out, versus actually having to seed the data yourself.

The pace of innovation in this field means that core financials and ERP vendors are also introducing AI-powered FPM tools, but Halloran argues that Planful's specialist focus as a best-of-breed vendor means its capabilities will stay ahead of more generalist tooling. While it used to be a challenge even to connect different data sources together, that exercise has become "trivial" he says, and what's most important now is how the tool supports people in analyzing and interpreting the data. He explains:

It's the calibre of functionality that enables people to do stuff with it ... The concentration of focus of what the use cases require — to be frictionless, to be enjoyable, to add extra value to a company — it's hard for these behemoth companies to always compete with those best-of-breed specialist players [and] to innovate faster.

Generative AI roadmap

When my colleague Jon Reed spoke to Halloran in May ahead of the vendor's annual Perform conference, he was impressed by the absence of generative AI hype. But that's not to say Planful is neglecting the technology. There will be announcements at Perform next year, Halloran assures me, with more on the roadmap. But the vendor wants to take the time to get it right. He explains:

We didn't want to do anything trivial. We wanted to do something meaningful and impactful. And so we'd rather take a little bit more time to make sure that happens.

The first of three use cases he outlines is simply the ability to create a more conversational interface for Planful users, by training a Large Language Model (LLM) that can understand their natural language requests and guide them to the results they need. This in turn makes it possible for a wider cross-section of users to perform administration functions. Halloran sees this as a generational change in the user experience — the first generation was green-screen terminals, then came the drag-and-drop Graphical User Interface (GUI), and the third iteration saw the advent of cloud-based SaaS. Now AI will bring a fourth generation of software, where users will simply have a conversation with the system and it will figure out the best way to deliver what they're asking for.

The next use case is to harness a company's historical data to inform current analysis. Halloran argues that a company's financial data stores a largely untapped data resource that holds lessons about what worked or didn't work in the past. He explains:

If we make certain decisions today, one of the ways in which we can understand the potential impact of those decisions is to understand what the result of previous decisions was. A lot of companies, unfortunately, they'll make decisions, but they don't necessarily create structured data in which you can see the short- to long-term impacts of those decisions ...

Our system will be able to inherently capture the narrative context of decisions ... For instance, we already do have language in our system where analysts are constantly asking questions and commenting on financial data. As that builds up over time, the system is going to understand how the humans think about the data. It's going to understand what a decision was, versus just a question. Over the long term, that's going to be very useful.

He gives an example of how this might work:

For example, you're halfway through the year, and you're off plan for one particular product line, or division or whatever. You'll be able to ask the system, 'What should we do? Give me three recommendations.' And the system will be able to tell you, 'In previous situations in this business, where we've been off this amount or whatever, we've done these three things, and these were the results. In this situation, based on this context, we would recommend the following.'

That is incredibly powerful because ... if you look at all the financial data in our software, it effectively is an amalgam of all of the corporate memory of the company. Every decision that was ever made, every resource allocation that was ever made, is manifested in financial data. So I think it's a magical future where you'll be able to now bring the illusion of thinking in the system, and then narrative generation out of that, that makes sense to people. It's quite profound.

Humans will still be in the driving seat, taking the ultimate decisions about what to do, but those decisions will be far better informed. He adds:

We still think that the human in the loop is going to be really important ... At the end of the day, humans are going to have to make a judgement call.

Changing role of finance professionals

Nevertheless, this has repercussions for the role of finance professionals and the contribution they make to the business, he adds. If AI is able to interpret historical data and present alternatives based on prior experience, then the value of having 'been there, done that' may be lower in the future. He goes on:

In a large company, or a scaling company, any company with complexity, today, we're just so dependent upon the memory of the people running your business, running all the different parts of the business. This is why tenure is so important. We can extrapolate this out into the future of work. Tenure may become less of a predictor of success in the future because the system can effectively be your corporate memory.

The third use case — and now we're looking perhaps a decade ahead rather than in the next year or two — is where AI, having removed much of the friction of producing forecasts in collaboration with finance professionals, will be able to proactively build the forecast and interpret the data. He explains:

Financial performance management platforms like ours have already been on that journey of trying to reduce that friction, and largely succeeding in many areas. I think that the AI takes it to the next level. One, in terms of being able to do the hard things quicker. For instance, seeding a new forecast, or having virtually a limitless way in which you can ask the system to give you different scenarios — best case, worst case, Europe down by 10, whatever. And just being able to conversationally talk, in the systems in which you work, rather than having to always log into a system ...

I think that, somewhere in the future, let's just say 10 years plus or minus, the cost of running finance and accounting could potentially be an order of magnitude cheaper, with the pace at which this stuff is developing. Because while software has hitherto been more focused on automating low cognition activities, we're now going into this realm of almost instantly automating high-cognition activities, like creating plans, creating forecasts, generating report narratives.

In this future scenario, a finance leader would no longer need to read, digest and interpret a report, or have someone else do that for them and provide a summary. Instead, the AI will do that. He goes on:

If I'm looking at a report that has 2,000 lines, and I want to figure out what's important in this report, I have to basically eyeball it. I have to scroll through it and I have to look at things. I can use conditioning and colouring and things to help point these out. And you can use our predictive technologies to spot anomalies.

In the future, at least with Planful, you're just going to be able to ask it, what are the top five things I need to know about this report? However, we should not underestimate the challenge to get there.

My take

Some thought-provoking takes on the future from Halloran, albeit hedged with caution that there is still plenty of work to do before some of these outcomes become a practical reality. However it underlines the importance of heeding the warning he gave in his conversation with Jon Reed:

I think it behooves you to start learning quickly what this technology can do, being part of the conversation inside your own company.

If his conjectures about the future impact of AI hold true, then the impact on the role of finance professionals in any enterprise will be enormous. Much of today's more mechanical activities, such as laboring over spreadsheets, compiling forecasts, or detecting outliers, will be carried out by AI. Instead, the role will become much more focused on exercising judgement, especially in cases that require the kind of creative, outside-the-box thinking where humans still have the edge over AI. Communicating those judgements and collaborating with others across the business will also become increasingly important.

In thinking about how people may respond to these potential changes, I'm reminded of how IT professionals responded to the advent of cloud computing a decade or more ago. While some embraced it, many dismissed it as untrustworthy and flawed, remaining wedded to their existing ways of working. It would not surprise me to see a similar reaction from the finance profession to the advent of AI today. There are plenty of ways to find fault with this novel technology if you're not minded to adapt to the new ways of working it demands.

I suspect that Planful's customer base will skew towards those who are ready to embrace these new ways of working. It is already promoting a collaborative approach to planning and a 'continuous close' mentality when dealing with financial data — see my follow-up interview for more on this collaborative, dynamic approach to planning. It's also bringing other datasets within the ambit of FPM, such as workforce planning and marketing spend. When competing against larger core financials and ERP vendors who are also adding FPM to their product offerings, this openness to innovation is a significant advantage. Based on my experience with the transition to cloud computing, those larger vendors tend to be held back by the need to cater to those in their customer base who are reluctant to embrace change. Whereas a best-of-breed specialist like Planful can focus entirely on those enterprises that are ready to adapt to the new world.

A grey colored placeholder image