AI for operations: killing the spreadsheet that runs your company
Every growing company has at least one: the spreadsheet that quietly runs something important.
It started as a quick fix. Now it holds the logic for how orders get routed, or how inventory gets tracked, or how the month closes — and one person understands it. When they are out, things slow down. When they leave, things break. Nobody chose to run a critical process on a fragile file maintained by hand. It just happened, the way operations debt always happens: one reasonable shortcut at a time.
Operations is where AI earns its place fastest, precisely because so much of it is this kind of work — repetitive, rule-shaped, high-volume, and reversible. The trick is starting in the right place and not skipping the unglamorous step that makes it stick.
What to systematize first
Data entry and movement. Information gets copied from an email into a system, from one system into another, from a PDF into a spreadsheet. It is slow, it is error-prone, and it is constant. A system can read the source, apply the rules, and move the data — accurately, every time, without anyone retyping anything.
Approvals and routing. A request comes in and has to get to the right person with the right context, then the result has to go somewhere. Most of that is logic — if this, route there; if that, escalate — wrapped around a couple of human decisions. Automate the routing and the chasing; leave the actual yes-or-no to the person who should own it.
Reconciliation. Matching two lists that should agree and flagging where they do not — invoices to payments, shipped to ordered, system to system. Tedious, high-volume, and exactly the kind of pattern-matching a system does tirelessly while a person reviews only the exceptions.
First-pass reporting. The recurring report that someone rebuilds by hand every week or month. Pull the data, assemble the numbers, write the first draft of what changed. A person still reviews and interprets — but they start from a draft, not a blank sheet.
Simplify before you automate
Here is the mistake that wastes the most money: automating a broken process.
If a workflow is tangled — full of steps that exist only because of an old system, or approvals nobody remembers the reason for — automating it just makes the mess run faster and harder to untangle. The discipline is to simplify first. Watch how the work actually happens, cut the steps that no longer earn their place, and only then build the system. Often the simplification alone recovers more time than the automation does.
This is why embedding matters more in operations than almost anywhere else. You cannot fix a process you have only heard described in a meeting. You have to sit inside the real work — the actual approvals, the real spreadsheet, the genuine exceptions — to know what should be cut, what should be kept, and where a human still needs to stand.
Keep a person on the exceptions
Automating the volume does not mean removing the people. It means moving them to where their judgment is worth the most.
The system handles the ninety-plus percent of cases that follow the rules. It flags the ones that do not — the unusual order, the mismatch that does not reconcile, the request that falls outside the logic — and routes those to a person. Your operations team stops spending its day on rote processing and spends it on the edge cases and the improvements that actually need a human brain.
The payoff compounds
Operations is the foundation everything else runs on, which is why fixing it pays twice. The first payoff is obvious: hours back, fewer errors, less dependence on the one person who knows the spreadsheet. The second is quieter and larger — a clean, systematized operation produces reliable data and stable processes, and that is the ground every other AI system you build later will stand on.
Kill the load-bearing spreadsheet. Simplify what is underneath. Then let the systems you build on top of it compound.
Want this kind of system inside your business? Start a conversation →