Every CFO is being asked the same question right now

How many hours has AI saved you?

How many dollars has AI added to the bottom line?

If you are a finance leader, you are getting this question from the CEO, the board, and probably your peers. And if you have been honest with yourself, you know the answer is complicated.

Some of it is real. A lot of it is hype. And the pressure to pretend otherwise is growing.

This piece is for the finance leaders who want a clear, unhyped view of where AI creates value in FP&A, where it does not, and what has to be true of your planning architecture for any of it to matter.

Where AI wins in finance (and where it doesn't)

The honest answer is that AI is producing real wins in controllership and finance ops, and more modest wins in FP&A.

Controllership and finance ops: the clear wins

Reconciliation. Journal entry classification. Document processing. Invoice matching. Anomaly detection in transaction data.

"Repeatable, standardized, high volume, rule-based. AI is good at this work because the task has a clear pattern to learn and a clear definition of correct output." Those are the tasks Marc Culver, VP of Finance at Priceline, called the "golden tickets."

If your organization has not yet captured these wins, start there. The ROI is measurable and the risk profile is low.

FP&A: the more modest wins

AI in FP&A is not going to rebuild your planning model for you. Not this year. Probably not next year.

"AI is gasoline on a fire. But the fire has to be there first." Anthony Losurdo, Fintastic's head of solutions.

Practically, this means:

• Analysis that used to take an analyst a day and a half can now take three hours with a capable AI assistant
• Natural language queries against live plan data can surface insights faster than building reports from scratch
• Formula building and debugging can accelerate with AI-assisted tools inside the planning platform
• Sensitivity analysis can be kicked off conversationally instead of through manual modeling

That is real value. It is not transformational yet. It is accretive.

What AI in FP&A is not going to do (anytime soon)

Let's be specific about the limitations, because the marketing noise around AI in finance is making it harder to set internal expectations.

AI is not going to build your planning model from scratch

The hype cycle suggests that AI will soon be able to design planning architectures, generate calculation logic, and hand you a working forecast model. That is not where we are.

Anthony's framing is the right one: trust but verify. Finance professionals are not going to accept a model built entirely by AI without weeks or months of validation. The stakes on accuracy are too high. The audit trail matters too much.

AI is not going to rescue bad data

Garbage in, garbage out still applies. AI performs only as well as the data it operates on. If your actuals imports are inconsistent, your dimensional structures are messy, or your data flows are broken, AI will produce confident-sounding answers that are wrong.

There are AI techniques that help with data quality, particularly unsupervised anomaly detection that can surface suspicious values for human review. Those are useful. But they are tools for data cleansing, not magic that makes bad data good.

AI is not going to replace the judgment layer

The parts of FP&A that require strategic judgment, business context, and political navigation are not automatable. AI can surface patterns faster. It can accelerate calculation. It cannot decide how to present a difficult forecast to a skeptical board.

The prerequisite: AI needs connective tissue to work

The biggest single misconception about AI in FP&A is that AI is the constraint.

It is not. Execution architecture is the constraint.

An AI model is only as useful as the data it can reach. If your planning environment is fragmented across multiple disconnected models, bolted together with imports and reconciliation, then AI has no coherent picture of your business to reason against. It can only see what you manually feed it.

"In order for AI to really take advantage of that, you have got to have your connective tissue be your planning layer that stitches HR, your transactional systems, your booking systems, your ERPs together into a common language. Then AI becomes that accelerant." Anthony Losurdo, Fintastic's head of solutions.

This is the part most vendor pitches skip over. AI capabilities demoed in isolation look impressive. The same capabilities connected to a fragmented planning environment produce thin, surface-level answers because they cannot see across the model.

What to actually demand from AI in a planning platform

If you are evaluating a planning platform with embedded AI, the right questions are not about what the AI can generate. They are about what the AI can reach.

Does the AI operate on the live model?

AI that queries a snapshot of data is reporting. AI that queries the live planning model is analysis. The difference shows up the moment assumptions change.

Does it respect the same permissions as the rest of the platform?

AI that bypasses role-based access controls is a compliance risk. AI that inherits the same permissions is a productivity tool. Ask specifically.

Does it reason across versions and scenarios?

Useful AI in FP&A can answer questions like "what changed between version A and version B, and which drivers account for the difference?" This requires the AI to reach into isolated scenarios and compare them natively.

Can it trace its answers back to source data?

Answers without traceability are risky. Finance needs to be able to audit how an AI arrived at a conclusion. This is particularly important when the output is going into a board deck.

Does it accelerate building, not just reading?

The most useful AI for FP&A teams today helps build and debug formulas inside the planning tool. This is where time savings are most immediate. Ask for a demonstration of the AI assistant in the context of actual model building.

A realistic plan for getting AI value in FP&A this year

• Capture the controllership wins first: Reconciliation, classification, anomaly detection. This is where AI is mature and where ROI is measurable.
• Give your FP&A team access to capable general-purpose AI tools: Give your FP&A team access to capable general-purpose AI tools alongside their planning platform. The two work in tandem. Excel-based ad-hoc analysis benefits from an AI assistant. Strategic planning benefits from AI embedded in the model itself.
• Evaluate planning platforms on architectural fitness for AI, not on AI features: A connected, unified model is the substrate AI needs to be useful.
• Set realistic internal expectations: Position AI as an accelerant, not an autopilot. Push back on pressure to quantify impact in ways that are not yet real.
• Plan for the co-pilot era: The next 12 to 24 months will be defined by AI as a co-pilot for FP&A, not as a replacement for analysts. Design workflows that assume human-in-the-loop.

The Fintastic approach

Fintastic embeds AI natively in the planning architecture rather than layering it on top. The AI operates on the live planning model, with the same permissions and logic that govern the rest of the platform. This means answers are grounded in actual plan data, not in disconnected summaries or exports.

More importantly, Fintastic was architected so that the connective tissue AI needs is already there. Financial data, operational data, workforce planning, and revenue planning live in a single unified model. When AI reasons against that model, it has a complete picture of the business to work with.

The AI is not the differentiator. The architecture that makes the AI useful is.

Talk to our team

If your FP&A team is feeling the limits of what your current platform can do with AI, the conversation worth having is not about AI features. It is about what your planning architecture can support.

We will walk through your current setup, your scenario workflows, and where the constraints actually live. No demo. No pitch deck. Just a working conversation about what is possible.

Connect with our team