Much like every other industry, the finance sector is already seeing massive AI deployments across its functions. Active use of AI within finances more than doubled between 2024 and 2026, rising from 30% to 75%, according to KPMG’s 2026 Global AI in Finance report. The pressure to keep pace has heightened, and the productivity case for AI in accounting and financial operations is well-documented.
What gets less attention is what happens when AI is introduced into finance workflows before those workflows are ready for it, and the particular risks this poses in the month-end close.
Why the close is uniquely vulnerable
The financial close is one of the most complex, interdependent processes in any organization. It involves reconciliations, multi-system data pulls, judgment-driven adjustments, and documented decisions that carry regulatory weight. It also tends to accumulate years of informal workarounds like spreadsheets that were meant to be temporary, manual steps that compensate for systems that don’t integrate cleanly, and reconciliation logic that exists either in process documents that haven’t been updated in years, or in the institutional memory of long-tenured team members.
When AI is layered onto that environment, it doesn’t magically create a clean process. It accelerates an existing one, with the good and the bad that comes with it.
Because AI executes consistently and at speed, errors that a human reviewer might eventually notice and question can instead become systematized. A reconciliation mapping that has been slightly off for months can be replicated across multiple periods before anyone flags it. A misclassified account can propagate through the close cycle faster than your exception handling process was designed to catch it.
This is a predictable consequence of deploying automation into processes that were not properly audited beforehand.
An unsolved data quality issue
The research is consistent on this point: most finance organizations are not data-ready for AI, even as they blindly adopt it.
KPMG’s report also highlights that data quality ranks as both the most cited barrier and the most cited opportunity in AI deployment, with 36% of organizations identifying it as their greatest vulnerability. A separate 2026 survey by Coupa found that while 63% of CFOs believe they have full visibility into their spend data, only 5% actually do, a gap that points to how significantly finance leaders can overestimate the reliability of their own data environments.
These figures describe the broader enterprise scenario. In the context of the financial close, the implications are more acute. Close data is drawn from multiple source systems (i.e. ERPs, subledgers, bank feeds, intercompany platforms). These often have inconsistent formatting, incomplete mappings, and varying levels of reconciliation discipline depending on who manages each system.
Feeding AI into that environment without first resolving those inconsistencies does not eliminate data quality problems. It embeds them more deeply into the output and will very likely make it more difficult to identify and fix by the finance professionals.
What “AI-ready” actually means for a finance team
AI-readiness in a close context has a specific meaning that goes beyond having the right software in place. It demands that reconciliation processes are well documented as they actually run today, not how they were designed.
It also means that data from source systems is clean, consistently formatted, and correctly mapped before it enters any automated workflow; that exception handling procedures are defined so the team has a clear protocol when AI flags an anomaly, and that a baseline audit of the current close has been completed, giving you a known quality standard against which to measure AI output.
The difference between automating a good process and scaling a broken one
There is an important distinction that tends to get lost in the momentum of AI adoption: AI that automates a well-designed process and AI that accelerates a flawed one produce very different outcomes. The problem is that the difference is not always visible in the short term.
Organizations that report significant financial returns from AI are, consistently, the ones that redesigned their end-to-end workflows before selecting the technology to power it (not after). When that order is reversed, the result is often an audit finding that looks like an accounting problem rather than an AI problem, which makes it harder to trace and more expensive to remediate.
The path forward
None of this argues against AI in the close. The efficiency gains available to well-prepared finance teams are substantial, and as the technology continues to improve, so will its outcomes. I’m making a call for teams to do the pre-work before deploying new automated processes, rather than hoping the technology compensates for the preparation that was skipped.
That pre-work won’t be the highlight of your workday, but it’s necessary. It’ll involve auditing your current close end-to-end, documenting the process as it actually runs, cleaning source data, and establishing clear quality baselines. It is the kind of foundational investment that doesn’t have a line item in most AI roadmaps, and that rarely makes it into board-level presentations about digital transformation.
But it is what determines whether AI accelerates your outcomes or accelerates your problems.
As we enter the second half of the year, I call for finance teams to make decisions that go beyond wanting to be the fastest adopters. Be the one who takes the time to understand what you’re looking to automate before doing it and I can almost guarantee it won’t be ‘too soon’ for you.
==
Shagun Malhotra is CEO of SkyStem, a financial close management software company serving mid-market and enterprise organizations. She has over two decades of experience in finance and accounting technology.
Sign in to get access to this free resource, and all of our whitepapers and reports.
Download this content today!
Register Now Already registered? Click here to Log In
