The State of AI in Finance Report: High but Shallow usage; no discussion of trust
- Niv Nissenson
- 7 minutes ago
- 3 min read
A recent report on “The State of AI in Finance” by CFO Connect—authored by Luc Hancock and sponsored by NetSuite, Spendesk, and Remote. The report offers a useful snapshot of how finance teams are approaching AI today however while being directionally right but misses some of the most important issues.
Most Finance Execs are Using AI… But in a shallow way
One of the headline stats in the report is that 56% of finance leaders are already using AI. But when you look at how AI is actually being used, the picture changes.
Most of the use cases are still fairly lightweight—helping draft presentations, summarize meetings, assist with reporting, or pull together research. Useful, yes. Transformational, not quite.
This aligns with what we’ve been seeing for a while now:
AI is great at making individuals faster. It’s not yet great at running finance processes.
That distinction matters more than people realize.
Data Remains Top Problem
The report does touch on the challenges holding teams back—data fragmentation, security concerns, lack of training, and the constant pressure of reporting cycles.
The biggest issue that stands above the rest: data fragmentation. Because data isn’t really an AI problem. It’s an infrastructure problem that finance teams have been dealing with for years.
But using AI on scale requires good data so this age old issue becomes more difficult to ignore as finance executives try to keep up.
If your data lives across disconnected systems, stitched together with APIs and manual workarounds, you’re going to struggle no matter how powerful the AI layer on top is. A quote in the report from Subscript’s CEO that sums it up perfectly: "Nine out of ten finance leaders think their data is a mess."
The Thing No One Talks About
In the report there’s no real discussion of hallucinations.
In finance, that’s a big deal.
Because at the end of the day, finance doesn’t run on “probably correct.” It runs on trust.
If an AI system gives you an answer, you need to know:
Is it accurate?
Is it consistent?
Can I defend it in an audit?
Right now, the honest answer in many cases is: not reliably.
We’ve tested this ourselves with accounting scenarios, including revenue recognition logic. The results weren’t something you’d want to rely on in a real close process.
And yet, the report leans into examples of AI applying standards like ASC 606 and generating accounting entries automatically.
That’s an exciting vision—but it’s not the day-to-day reality for most finance teams today.
There’s still a gap between what AI can do in a controlled demo and what it can do consistently in production.
Tools Are Evolving Faster Than Workflows
One part of the report I did find interesting was the breakdown of tools.
ChatGPT has a commanding lead among finance professionals, which says a lot about how adoption actually happens: people use what’s accessible.
Meanwhile, more specialized tools are starting to emerge—platforms built specifically for finance workflows like reconciliations, variance analysis, and planning.
This is likely where the next wave of real progress comes from.
Not generic AI tools, but systems designed around finance-specific problems.
So Where Does This Leave Us?
If you step back, the report reinforces something we’ve been saying for a while:
AI is not being held back by lack of interest.
Finance teams are clearly eager. They’re experimenting, adopting, and learning.
What’s holding things back is more fundamental:
Data that isn’t clean or centralized
Systems that don’t talk to each other well
And a lack of full trust in AI outputs
Until those are addressed, AI will continue to sit primarily in the individual scaling mode.
A More Grounded View of What Comes Next
The report is a good reflection of where the industry is right now. We’re in a transition phase.
AI is already valuable. It’s already saving time. It’s already changing how finance teams work on a daily basis.
But the bigger shift, the one where AI becomes deeply embedded in financial operations, is still ahead of us.
And getting there won’t just be about better models.
It will be about better data, better systems, and most importantly trust.
