Linda · Invoice Checker
Finance · Custom implementation
The AI Invoice Checker
Checks incoming invoices, flags discrepancies and prepares them for processing.
Connects toExactRetyping invoices, sending reminders, chasing receipts and, at month end, chasing everyone for their paperwork. Much of the work in finance is the same every month, and a lot of it is still done by hand.
An AI employee for finance does the groundwork in your own accounting system: reading, matching, coding, reminding and flagging. What gets posted, what is out of line and what the figures mean stays with your controller or accountant.
Linda · Invoice Checker
Finance · Custom implementation
Checks incoming invoices, flags discrepancies and prepares them for processing.
Connects toExactRob · Collections Assistant
Finance · Custom implementation
Follows up on unpaid invoices, sends reminders in the right tone and flags what needs a call or a debt collector.
Connects toExactAFASLisa · Expense Claims Processor
Finance · Custom implementation
Reads receipts and expense claims, checks them against your own policy and only passes on the exceptions.
Connects toExactAFASMicrosoft 365Jeroen · Purchase Order Handler
Finance · Custom implementation
Turns purchase requests into orders, tracks approval and keeps an eye on what has been ordered but not yet delivered.
Connects toExactAFASMarijke · Month-End Close Assistant
Finance · Example application
Gathers what the close needs, flags what is missing and prepares the recurring checks.
Connects toExactAFASMicrosoft 365Tim · Subscription Billing Assistant
Finance · Example application
Keeps recurring billing on track, processes contract changes and flags what has not been invoiced.
Connects toExactAFASThese AI employees sit with another department, but touch the same work.
AI in finance takes over the recurring preparatory work: reading and checking invoices, processing expense claims, chasing debtors and preparing the month-end close. Your controller still decides on exceptions, payments and everything that goes to the accountant. The rule is simple: the machine prepares, a person signs off.
What does that look like in practice? A purchase invoice arrives in the shared inbox. The invoice processing employee reads it, matches it against the purchase order and the goods receipt, and suggests the ledger account and cost centre. If something doesn't add up, it goes to a person with the reason attached. At the end of the month, the month-end close assistant chases missing submissions and runs the standard checks. Dutch packages such as Exact Online, AFAS and Twinfield stay where they are. If your question is more about figures and reports from separate systems, see an integrated data platform for a financial organisation.
My advice: always have a person approve payments, as well as exceptions and the conversation with a customer who pays late every time. What AI can and can't do in bookkeeping is covered in AI bookkeeping. How an AI employee works as a colleague in your team is explained in what an AI employee is.
What we would not start with: letting AI loose on accounts that are already a mess. You just get the mess faster. Sort out your chart of accounts and your approval steps first. Where it then really saves time is what we work out in the AI Opportunity Audit.
This is the starting point for each task. Exactly where the line sits for your team, we agree together during setup.
Runs automaticallyReads invoices, matches them to the order and goods receipt, and proposes the ledger account and cost centre.
A person decidesApproval, and every invoice that does not match. What may be posted automatically is set per account and per amount.
Runs automaticallyTracks outstanding items and sends reminders in the agreed tone and sequence.
A person decidesDisputes, payment plans and collections. Large amounts have an approval step by default.
Runs automaticallyReads receipts, checks them against your expenses policy and posts what falls within it.
A person decidesClaims outside the policy. They go to the line manager with the reason attached.
Runs automaticallyCompletes requests into an order, collects approval according to your authorisation matrix and tracks open orders.
A person decidesThe approval itself, and anything outside the matrix. Ordering on its own only happens if you choose it.
Runs automaticallyKeeps the checklist, chases missing submissions and runs the recurring checks.
A person decidesThe postings and the review of the figures. Those stay with the controller or accountant.
Runs automaticallyCompares contract terms with what was billed, reports differences and prepares a billing proposal.
A person decidesChecking the proposal before it goes out, and the decision to apply indexation.
You stay in control.
Not every decision should be made automatically. For each workflow, we set out:
The AI employees on this page work with these packages, among others. Is your system not listed? We can connect any system, even if there is no standard integration.
The tasks that run on fixed rules: reading and matching purchase invoices, sending payment reminders, checking expense claims against your policy, handling purchase orders, keeping the month-end checklist and checking recurring billing. In every case, anything out of line goes to a person with the reason attached.
No, and that is not the aim. An AI employee takes over the gathering, reading and preparation. Postings outside the rules and the review of the figures stay with your controller or accountant. What may be posted automatically is set per ledger account and per amount.
Yes, through the API or the import route of your package. We only retrieve what is needed and write nothing back to your ledger unless that has been explicitly agreed.
By never approving anything silently. Invoices that do not match the order or goods receipt go onto an exceptions list with the reason: price difference, quantity difference, missing order or unknown supplier. A person decides on those.
I advise against it. AI in finance can prepare payments and flag exceptions, but have a member of staff approve every payment. Mistakes in payments are expensive and hard to reverse.
A firm has the same preparatory work, only across several sets of books: gathering documents, reading invoices and tracking submissions. The profiles here are written from the perspective of an in-house finance team. Whether the pattern fits your firm depends on your packages and the way you work, and we look at that case by case.