In accounting, the word “just” usually hides a lot of manual work. For a long time, whenever someone asked how our financial model stayed updated at Latch, I would say, “I just pull the figures out of Xero into Google Sheets.” On a checklist, that sounds like a simple transfer. When our accounting team actually mapped every step, it was a 15-minute routine that typically required over 200 clicks: setting date ranges, exporting CSVs, deleting unneeded periods, writing lookup formulas, wrapping values in error checks, and more.
That doesn’t sound like much. None of those steps are difficult on their own, but together they formed a fragile chain of manual effort that is repeated every week. By applying AI to act as a bridge between our tools, we were able to collapse that manual 200-click workflow down to around 5-10 clicks. Most finance leaders know AI will reshape accounting, but figuring out where to start can feel overwhelming. You do not need to tear down your software stack or implement complex enterprise platforms to see immediate results. The best place to start is actually right inside your current close process.
By giving AI proper context into our workflows, we were able to optimize away organizational pains and eliminate the rote, repetitive drag on our close cycle.
Bridging the Talent Deficit with Existing Systems
Finding practical ways to eliminate manual friction is becoming mandatory for growing finance teams. According to the AICPA’s 2025 Trends Report, higher education institutions awarded 55,152 accounting bachelor’s and master’s degrees in the 2023–24 academic year, a 6.6% drop from the previous year. While degree numbers may stabilize, the supply gap remains significant: the accounting industry is growing roughly seven times faster than the undergraduate student population it draws from for new hires.
Finance departments can no longer throw extra headcount at growing transaction volumes. Efficiency has to come from system design.
Fortunately, your existing software is usually enough. Applications like Xero, QuickBooks, and Excel handle accounting logic well on their own. The main source of operational drag is the empty space between these applications: the manual copying, pasting, and reformatting required to make systems talk to each other.
AI functions best as connective tissue between these tools. It extracts, formats, and routes data across existing software automatically. Our team hasn’t shifted in terms of accountability: human accountants still review the numbers, evaluate variances, and approve the close. The technology simply removes the mechanical repetition required to prepare the data.
Five Questions to Ask Before Automating
Before attempting to automate any financial task, look past the high-level checklist and capture the specific decisions behind each action. Ask these five questions to evaluate whether a process is ready for automation:
- Does it repeat? Tasks performed weekly or monthly yield immediate time savings and provide frequent opportunities to test and refine the workflow.
- Are the inputs and outputs clear? The system must know precisely where source data lives, where it belongs, and how completion is verified.
- Can the rules be explained clearly? If you cannot articulate why a specific step occurs or what logic dictates a mapping, the process needs clearer documentation before introducing technology.
- What exceptions require human judgment? Define specific thresholds, unexpected zero balances, missing data, or changed account mappings that should pause the workflow for human review.
- What audit trail will the reviewer receive? The workflow must generate clear evidence showing what data moved, when it moved, and any anomalies detected so you can verify results quickly.
Start Small to Build Momentum
Broad mandates to “transform finance” often stall because they try to solve too much at once. A single recurring process provides a far better environment for learning. Pick one tedious, rules-based workflow that an experienced team member knows thoroughly and would gladly stop doing by hand. Record the exact steps in real time as the work happens. Define the review checkpoints, test the automated flow with live historical data, and compare the results against your manual baseline.
Measure success by the manual touches removed, error rates, and the capacity reclaimed during busy close windows. Reducing a single 15-minute, 200-click routine to 5-10 clicks reclaims hours of high-value focus across the year for forecasting, variance analysis, and operational support.
Every accounting process sounds simple until you look under the hood. As finance professionals, our job is to examine the mechanics beneath the summary figures. Pick your most repetitive routine, ask the right questions, and good luck getting your own click counts down.
===
Stefan Kalb is cofounder and CEO of Latch, which has raised $8M to help businesses automate the manual work inside their day-to-day processes. He previously founded Shelf Engine, a Y Combinator company that raised more than $60M before selling to retail data company Crisp in 2025.
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
