Planful Perform 2025 - if FP&A is a ticket to burnout, can AI change the finance narrative?

At the Planful Perform 2025 opening keynote, CEO Grant Halloran opened with a reality check for FP&A (Finance, Planning and Analysis) professionals: If we stick with business-as-usual, then we're heading in the wrong direction.

The work has never been easy... Long hours, complex domains, high stakes. 

But over time, it's only gotten harder: more pressure, fewer people, tighter timelines, no slowdown. Expectations keep rising - resources don't.

Not the happiest of pictures - and the talent shortage isn't helping. Halloran unfurled the stats: 

According to CFO magazine, we have a shortage of at least three million finance and accounting professionals globally... And while this is a staggering talent deficit, the problem is, frankly, only going to get worse: 75% of CPAs are within a decade of retirement.

The "staggering FP&A talent deficit" - where do we go from here? 

New graduates aren't going to be filling the talent gap. Halloran: 

You would think, 'We'll just plug it with all these new graduates that are coming through.' Well, unfortunately, the number of graduates that are taking the CPA exam is at the lowest point in nearly two decades, and has no sign of reversal. In fact, CPA graduation graduation rates since 2015 have dropped 20%. That was tracking at one-to-three percent decline each year, and then it went to eight percent last year. So the problem seems to be getting worse.

Planful cited similar stats in the workforce planning area: 

• Out of 450 FPA executives, 23% said they had trouble getting real time workforce data.
• 52% said that they had trouble predicting labor cost and needs.
• 64% said they had trouble aligning workforce planning with the company growth strategy.

Cobble together FP&A on the fly, give spreadsheets all your perseverance... The world has changed. Halloran: 

Frankly, in finance and accounting, these teams are running on fumes. We hear it every day from our customers. Job satisfaction is, frankly, at an all-time low in finance and accounting, and attrition is the highest that it's ever been. The stress that's been placed on you and your teammates is, frankly, overwhelming and unsustainable... The pressure is continuing to grow. You're grappling with these evolving regulations, geopolitical uncertainty.

Meanwhile, the expectations from executives and boards have gone up: 

At the same time, the expectations, the demands from your leaders and from your boards is increasing, and frankly, you just don't have the resources to cope with it. We hear this every day.

Halloran's warning: 

We're not going to be able to overcome this sort of deficit just by doing the same thing.

Putting Planful's answer to the sense test

So Planful must now respond: how will FP&A teams overcome - and even thrive? Yes, AI has something to do with it. But how? What's different today? AI, in a sense, is not an entirely new set of technologies either. Halloran declared the mission: 

Our mission is to enable companies to achieve peak financial performance irrespective of the conditions that they're operating in... We also want to give our users, the human beings that we serve - we want to give you folks a major win also. 

We want to reduce, and perhaps eventually eliminate entirely, a lot of that mundane, repetitive, low value work that you have to do. We want to give you time back to do high value, or strategic value, creating the sorts of things that I imagine you were thinking about when you decided you wanted to go into finance, and put all that work into your studies, and cultivate your careers to do those fun things. We also want to give you more time back, frankly, for your personal pursuits and your loved ones.

A bit dreamy? Perhaps. But as grouchy as I can get, I'm not opposed to dreamy visions of better workplaces - as long as they include a realistic plan to get there. Basically, Planful's plan must pass the sense test. There are two main parts to my test: 

It better not just be about AI. 

And, on the AI part, you better have a big dose of honesty about the pros and cons of today's AI - and a human-centric narrative on what the future of work looks like. Who said we have to cede the AI narrative to the exaggerated capabilities of autonomous agents, while our fragile livelihoods circle the drain, our untapped talent languishing? 

"Our goal is not to displace people" - Planful's AI ambitions

In my February 2025 interview with Halloran, I detailed the core of Planful's AI intentions: Our goal is not to displace people" - Planful CEO Grant Halloran takes a stand on the future of AI for finance leaders. But Perform 2025 advanced those plans - and provided a contrast with customer expectations. Halloran conceded the AI hype factor. But he says this time, it's different: 

My industry is notorious for overhyping new technology paradigms, right? This AI thing - there's one thing you can take away from me at this conference. This is not hype. I've never seen anything so transformative. It's going to be extremely powerful. I just want to also be really clear that while there are software companies out there that tout one of the big benefits of AI is job elimination, that is not our mission for Planful. 

I want to be unequivocal and unapologetic about that, given the numbers that I shared earlier. Our mission is to simply make it possible to execute all these processes that you guys run. Make it possible for you to do that without having to hire more people. We have to succeed. We must drive huge productivity gains into finance and accounting. We must do it, otherwise we are just not going to overcome this problem at the more micro level, inside your business.

For those who might be a tad cynical about the 'AI is going to solve our finance woes' narrative, we should note: this news-heavy event was not just about AI. Planful issued other announcements that made a good impression on customers:

• Power BI integration, with Tableau integration on the roadmap.
• Snowflake integration, with other "lakehouse" partnerships likely in the future.
• Workforce Pro, a premium enhancement to Planful's popular Workforce Planning Solution. 

Planful's AI relevance - customer Rocket Software shares field views

For now, let's look at Planful's AI roadmap from the eyes of customers - in this case, from Rocket Software's Luis Martinez Luna, Director FP&A. I asked Luna to react to Planful's AI news - specifically, the addition of persona-based AI assistants to Planful's existing AI/ML tools. Planful says its new assistants are "designed to aid and augment human effort, generate insights, and improve decision-making." Three new personas are available to early adopters:

• Analyst: Provides variance insights, narrative generation, report analysis, and scenario building to support financial analysts.
• Planner: Supports financial planners with forecasting and scenario optimization.
• Controller: Provides compliance reviews and automated adjustments to support controllers.

Other upcoming assistants include Business Analyst, Modeler, and Administrator. 

The use of AI at Rocket Software goes back to Predict:Signals. As Luna told me, Signals' anomaly detection helped his FP&A team punch above their weight to serve business users: 

On the corporate side, we're a light team, so we can't possibly go look at all the key intersections of data to go assess variances, to see where we have risk. That's where Predict was a natural answer for that - to say, 'Okay, where are my blind spots? Where am I not looking? I can look at reports top-down. That only gives me so much.' What we've been able to do, within my team, is we've built much more detailed reports, and then we just use Signals to flag where we might need to go ask some questions of the broader team, or our accounting colleagues. 

So how do the new Planful AI Assistants resonate? Luna sees an immediate use case, to support the broader FP&A user base: 

As far as our extended FP&A user base, we're more excited about AI assistants when it comes to their usage of it, because naturally, they're already looking at things on a much more detailed basis - so they know what's going on within their scope of responsibility already. Something like anomaly detection doesn't add as much value to them, because they're in the weeds already; they're deep down into journal entries and following much more closely hands-on in the close process. So it's more value-add to us.

Luna says Rocket Software's Planful use is more conservative than an aggressive early adopter. As a long time participant in Planful's Customer Advisory Board, Luna has a clear sense on where his peers stand. For Rocket Software, Luna says he's not yet ready to roll out AI-based chat to a range of business users. But it's in the fast lane, and he intends to do some roll outs later this quarter.

I tend to wait until I think they're at a mature enough level where the team will actually use them. I think we're hitting that threshold now with the AI assistants, and so that's going to be the next step for us, to roll that out and have them actively engaged in that.

Luna sees another big payoff to Planful AI: as Planful bears down on AI, this saves Luna from the overhead (and challenges) of an internal data science team: 

About two years ago, I did a webinar with Carl Seidman on the topic of AI in finance. At the time, weren't quite sure what it was going to look like - where we stand here today. But our assumption was generally that the technology vendors were going to make it so much of a seamless experience to leverage AI that we wouldn't necessarily all need to become experts in Machine Learning, statistics, math and AI to implement it. 

We're now seeing that come to fruition exactly as we assumed it would, which is: the vendors that have the capability and the expertise in the technology space, they're just embedding AI into the user experience. What that does for us is we no longer have to worry about building a data science team, and we can keep our functional experts, whether it's marketing or finance or whatever organization, focused on what is core to them. They don't need to go necessarily go learn a new skill set; they just need to learn how to interact with these new capabilities.

And so that is a little bit of a game changer, because we don't need to go and number one, spend financial resources to go hire teams of data scientists. And number two, it's there's no implementation effort required. Something like a Planful solution is just rolling out the capabilities as part of their monthly update cycle. 

Luna says this includes security - a key aspect of Planful CTO Sanjay Vyas' AI keynote segment. Vyas surprised me by going deeper under the AI hood.. I've argued that customers want to hear those details, including the security aspects vendors tend to gloss over. Vyas went into detail about how Planful's AI maintains all security parameters, and "you can't use a chat experience to bypass any of that." Third party audits, compliance and AI accounted for - it's what customers want to hear from the keynote stage - especially CFOs - but rarely do. When you see AI demos alongside SOC 1,2,3, HIPAA, BAA, TrustARC Privacy Seal, GDPR and CCPA compliance, those are the kinds of acronyms CFOs like to see. 

My take - on AI dreams and finance realities

FP&A teams have a chance to change how they serve business leaders, and it isn't just about AI. But as Halloran said, 'do more with less' is also in effect. Will AI have the profound difference Halloran anticipates in finance, without the dark sides of job losses? That's a debate for another time, and it's one worth having (example: I do see large AR and AP teams impacted by automation/AI). 

My current position: AI may be a bit less impactful than we hope in its current state, but modern FP&A solutions like Planful are very impactful in their own right - and are now underestimated in the face of AI innovation. Moving out of rogue spreadsheet addiction and onto a trusted data platform is a substantial win, not a small one. It's a big reason why Planful will be able to deliver a relevant, specialized type of AI in that context. You could say that the two are now inseparable. A big payoff for getting your finance data into a modern planning environment is the AI you'll have access to - and you won't have to build, test, and protect it with your own team.

Planful customers are on board with how Planful's AI assistants will bring finance data to business users in fresh ways, helping them identify top actions and decision points. Planful CPO Steve Welsh told me that on the demo floor, customers who gave the new assistants a spin were in 'I want it now' mode. 

Readers know this is the point in the piece where I start carrying on about AI accuracy issues, and why probabilistic LLMs have inherent weaknesses. Well, I can go easy on that point here. Planful isn't overdoing it with 'autonomous agent' hyperbole. Every customer I talked to had a grasp of the issues here, perhaps due to their own ChatGPT experiments. If you are cognizant of this problem, and keep your thinking cap on, you are a long way towards useful AI. Example: customers made a distinction between running instant scenarios with AI to compare plans, versus picking a plan and implementing it - which calls for validation outside of AI.

Meanwhile, CTO Vyas took us under the hood during the keynote, sharing an AI architecture that looked as effective for keeping LLMs on track as any I've seen to date, including the use of an external calculation engine for running numbers, verifying those numbers against actuals, and utilizing LLM output in the form of "domain specific instructions." No, this doesn't ensure that LLMs will behave at all times - whoever figures that out will earn themselves a Turing award. But it is an innovative architecture that warrants further on the record discussions, which I hope to bring to you from Planful going forward. 

Sharing these AI details on stage isn't getting too technical; it's a step towards the AI literacy we're all going to need. Why most vendors hold back on this is baffling; silly me, I thought AI adoption was about trust. Trusting AI is totally different than trusting your iPhone.

Obviously, Planful is doing something right. They increased their workforce by 35% last year, a welcome contrast with so many tech vendor layoffs. Meanwhile, Planful continued its international expansion, on the way to "record breaking" 2024 sales. 

As for how transformative AI will be for finance teams, I think we'll have a vivid answer - once finance teams have access to fluid, intuitive AI tools and start pushing into what's possible for themselves - and working with vendors to make it so. Those stories will be truly exciting. Planful is on track here; their AI was built in close contact with their customer advisory groups. 

I'm stubbornly hard to impress with AI; what would impress me here? I'd like to see Planful's total integration of external data sources into finance planning and AI interactions - be it weather data, labor data, inflation data, or any other macro-economic data set a customer might need. That's the best thing about AI: it's pushing us to think about what data we need at the point of inference, and what we may be lacking. Yes, you can pull some of that data into Planful now, and yes, from the keynote this is clearly on the roadmap. But we can all be impatient about something. That's the one I'll pick. 

AI is mature enough to offer value now, as Planful's Predict customers have noted. But I'm impressed by so-called time-to-value. Today's modern software - however impressive - is still, by and large, too time consuming to implement for the circumstances companies now face. Planful talks about getting notable results in months, and sometimes weeks. Call me an old fashioned crank if you want, but I'll talk about that kind of time-to-value all day long. Get that right, and the AI results will follow.