Guides / Operating gaps
Before a vendor invoice gets paid, three documents are supposed to agree: what you ordered, what actually showed up, and what you're being billed. Done by hand, that check happens at the end of a long week, against a stack of paper. A tool can check every invoice against the other two and bring a person only the ones that don't add up.
The invoice says $4,180. The purchase order says $3,900. The packing slip says nine pieces came in, not twelve. Somebody has to notice that before the check goes out, and the person doing the noticing is usually the same person doing payroll, answering the phone, and closing out last month.
I ran a machine shop for a decade. The rule that you don't pay an invoice you can't tie to something is right. It's also the first thing that slips when the week gets busy, because nothing bad appears to happen when you skip it.
It's an old idea with a formal name, and the name is worth knowing because it's what your accountant or auditor will call it. It's the same control whether you make what you sell or distribute it. You match three things before you pay:
If all three agree, pay it. If any two disagree, somebody needs to look before money moves. That's the control. It's also why a two-way match, comparing only the order and the invoice, is weaker: nothing in it proves the material ever showed up.
The hard part is doing it on every invoice, not just the ones that catch someone's eye.
The failures aren't dramatic. That's the problem with them.
Nobody sees them unless someone lays the three documents side by side.
Here's the one I built and tested at a mid-size manufacturer. Vendors emailed their material invoices to a shared inbox. AI read each invoice, whatever layout the vendor used. Then ordinary software pulled the matching PO and its receiving record out of the ERP and checked the three against each other:
Each weekday morning it sent one email to the person handling payables: what was ready to enter, what needed a closer look and why, and what was on hold waiting for material.
The AI in a tool like this is narrow, and worth being precise about. Vendor invoices arrive in every format there is, so reading the document and pulling the numbers off it is the part AI does well. The comparison itself is not AI. Matching a price to a price has to be right every time, so that part is ordinary software doing arithmetic, which is what you want when the output decides whether money moves.
A tool like this doesn't pay anything. Nothing moves without a person, on purpose. It doesn't replace your bookkeeper, and it isn't an accounts payable platform with a vendor portal and its own subscription on top of the accounting system you already pay for. It runs on your ERP and your books as they are.
The version I built also didn't type the invoice into the ERP. It did the checking, and a person still keyed each one. If you want the clean ones entered for you as well, that's added work, and I'll scope and quote it as its own piece rather than assume it.
And it can't see what isn't recorded. If material gets received into the building and nobody logs it, the invoice will look unmatched, and the tool holds it and says it couldn't find the receipt rather than guessing. That's on purpose: a system that quietly assumes is worse than no system.
Work it out for your own operation. Take the invoices you paid last quarter that nobody matched line by line, and ask how many of them you'd bet money were exactly right. If the answer is all of them, you don't need this.
Volume decides the rest. With a handful of vendor invoices a week, the person keying them can check them too, and this isn't worth building. With enough of them, the checking is where somebody's week goes, and that's where a tool like this belongs. For a smaller operation, the version worth building is the one that also enters the clean invoices, so nobody keys those by hand.
No, and the difference matters if you're comparing prices. Most accounts payable platforms bundle approval routing, payment and a vendor portal, on a subscription of their own on top of the accounting system you already have. This is one job, done against your existing systems: make the three documents agree, and show you the ones that don't. If you want the full platform, buy the platform.
It needs the receipt recorded somewhere it can read. The one I built read the receiving record straight from the ERP. If receiving lives on a clipboard and nowhere else, there's nothing to match against, and the honest answer is that the receiving side has to exist somewhere before anything can compare to it. That's worth finding out on the first call rather than after.
Some bills never have one: a utility, a subscription, a service. The one I built handled material invoices against a PO in the ERP. Anything without one was set aside and labeled for a person, never forced through a match it could never pass. How your operation should handle those gets scoped for your setup, not assumed.
No. It runs on top of what you already have and reads from both. It doesn't ask you to move your books or your purchasing into a new platform, and it doesn't charge per seat.
I build it, host it, and keep it running for you on managed infrastructure, working on top of the systems you already run, so nothing moves to a new platform. It's a Build that's quoted up front, so you know the price before it starts. After that an ongoing care plan keeps it running: hosting, monitoring, bug fixes, and small adjustments as your work changes, with term and price by agreement. No hourly billing and no per-seat license; the ongoing cost is simply what keeps the tool running and improving.
It starts with a paid half-day, $2,500, at your place or over video. AI reads, drafts, and remembers what your people know. It is not exact, and it does not run your systems. So we go through the jobs eating your week and sort them: the reading, writing and know-how AI can genuinely help with, and the parts that need a tool built because the answer has to be right every time. Your people leave able to write skills themselves. A skill is your way of doing a job, written down so the AI follows it instead of guessing. Whether AI, a small build, or nothing at all is the honest answer, I'll say so. Written document within two working days. If a build is the answer, I confirm your ERP and accounting setup fit and pin down the scope before quoting. From there it's a Build that's quoted up front, the number's set before any code is written.
The same problem, one step earlier. Everything above happens after the material arrives. The check that happens when the order goes out is covered in vendor PO acknowledgment tracking, and the one on orders coming in from your customers is customer orders, keyed in by hand. If your books and your ERP don't talk to each other at all, start with JobBOSS and QuickBooks. On the selling side, the same kind of check runs before you bill a customer: the billing check holds each shipment against the price on the customer's own PO.
If invoices are getting paid without anyone matching them against the order and the receipt, send a note. First call's free. About 30 minutes, a straight conversation about how your purchasing and payables actually run, not a demo. If there's something worth building, I'll say so; if there isn't, I'll say that too.