Let's have an honest conversation about AI in FP&A. Maybe I won't be invited to some conferences.
Most of what is sold as “AI-powered insights” is vaporware dressed up in machine learning buzzwords.
I've built and sold multiple SaaS companies. I sat on either side of the vendor table. And I've been following our industry journey for the past two years trying to convince CFOs that AI will replace FP&A analysts. All vendor sales pitches promise the same fantasy: “Our AI automatically generates predictions, predicts outcomes, and provides strategic insights!”
That's nonsense. That's expensive nonsense.
Everyone knows it (but no one says it)
What no one wants to admit on the industry panels we all attend is that AI cannot replace FP&A analysts. Today is different. Not within 5 years. Probably not in 10 years.
why? Because the hard part of FP&A isn't crunching numbers, it's understanding context.
AI doesn't know that the drop in revenue in the third quarter was due to internal political turmoil on the part of its largest customer, which delayed orders. We don't know which cost centers are untouchable due to management relationships that go back 15 years. You can't decide which of a thousand possible scenarios will actually matter to next week's board presentation.
AI models are probabilistic. Finances are deterministic. When presenting a $50 million capital investment plan to a board of directors, “about right” doesn't work. And “the AI said so” is no explanation. That's an excuse that will get you laughed out of the room.
What Vendors Won't Tell You (But I Will)
Here are some uncomfortable truths about AI-enabled FP&A that I've learned from actually operating in this space. A lot of it is advanced automation wrapped in the buzzword of machine learning.
I've attended hundreds of these vendor conferences. The next time one of you demos a “revolutionary AI insight,” ask:
- “Show me the architecture of your model.”
- “What is the training data?”
- “What is the confidence interval for this prediction?”
I'll tell you exactly what happens. They will squirm a little, then pivot to “proprietary algorithms” and “competitive advantage.”
Translation: They have nothing.
The vendor is teeth Does it use real AI? They are often solving completely the wrong problem. They are trying to automate strategic decisions (the hard parts that humans are good at) while ignoring the infrastructure bottlenecks where AI can actually deliver value today. It's like building a self-driving car that can drive in a straight line but doesn't know where to go.
The real problem (no one will solve it because it's not sexy)
Want to know what's really holding back finance teams from delivering strategic value? An infrastructure that locks talented analysts into tactical work.
The industry is busy chasing headlines like “AI analysts who think like humans.” Something spectacular. Meanwhile, real finance teams are working on:
- Implementations that take months to deliver any value.
- Endless data coordination between systems that need to talk to each other but don't.
- Manually fortify your diet for at least 20 hours each week.
- Talented analysts, people who graduated from good schools and have strategic minds, do the work that computers should do.
- Strategic questions that cannot be answered in real time because the infrastructure does not support it.
This is the real problem. There are talented people who sit around doing nothing or do monotonous work. And what is the industry's reaction? “But look, our AI can generate predictions!”
nice Who cares if the team can't make the infrastructure work well enough to trust its predictions in the first place?
Analysts are now even more valuable (and vendors hate this)
There's an irony here that should scare every software vendor. In the world of AI, strategic decisions matter. more Nothing less is precious.
As information becomes plentiful and cheap, asking the right questions becomes invaluable. If your model generates 1,000 scenarios and you find that 3 of them are actually important, that's what separates the insight from the noise.
FP&A Analysts are not data processors. They are context translators who understand what leaders actually need. They are strategic interpreters who distinguish between variables that they can actually control and variables that they only pretend to control. They simplify complexity into actionable options. They provide certainty when everything else is uncertain.
AI can't do this and won't be able to do this for years. Anyone who says otherwise is trying to sell you something.
One question every CFO should ask
Next time a vendor pitches you an “AI-powered FP&A solution,” ask them: “Does your AI understand my business context better than an analyst?”
It's not “Can you predict it?” It's not “Can you detect a pattern?”
Do you understand political dynamics? What about institutional knowledge? What are the strategic nuances that make your business different from the business next door?
If they say yes, they are oversold and you should walk away. If they say “not yet, but soon,” they're guessing and you should probably walk. If they say, “No, that's why your analyst is important,” they might actually be worth having a conversation with.
What AI should actually do (from the people who built it)
AI is powerful. I've seen it work beautifully. But only if you apply it to the right problem.
The most valuable applications of AI in FP&A are not the sexy ones that grab the headlines. These are infrastructure improvements that free up analysts to do what they do best.
- Let AI handle data integration. AI is great at that.
- Automate integration and reconciliation when reliable and auditable.
- Leave it to analysts to make strategic decisions where humans take the lead.
- Eliminate months-long implementation nightmares that halt projects before they even begin with better automation.
It's not revolutionary. Be honest about what works.
What does the real AI revolution look like?
I have seen a real AI revolution in the financial industry. Here's what it actually looks like:
Analysts spend less time wrestling with spreadsheets and more time answering strategic questions. They test scenarios in real time, rather than promising “I’ll get back to you next week” (at which point the decision has already been made without them). They have become the co-pilot that CEOs actually need. Not because they worked harder or got smarter, but because the infrastructure finally got out of their bloody way.
That's the promise. It's not an exchange. release.
The question is not whether AI will transform FP&A. It's whether vendors use AI to solve problems that actually matter to finance teams or that look good in pitch decks.
conclusion
Analysts are not the bottleneck. So is the infrastructure.
We have talented people who can do strategic work. You have a business problem that desperately needs to be solved. What stands between them is not a lack of AI, but a system that forces strategic thinkers to become data custodians.
Fix your infrastructure. Eliminate monotonous work. Then look at what your team can accomplish when they are free to actually do real work instead of fighting against tools that don't work when they should.
