Skip to content
News

Here’s how AI really makes your finance department more efficient

•

9 min leestijd

AI makes your finance department more efficient in data-heavy, rule-based work such as matching invoices, flagging anomalies and forecasting cash flow. AI prepares; approval, judgement and decisions on working capital stay with your people.

Your team spends valuable time analysing contracts by hand, working through cash flow scenarios and preparing reports. As CFO, you know that time would be better spent on strategic work. First, Erry de Boer, CEO of Eqeep, shares his view of AI in finance. We then show what IFS Cloud can already do and how you stay in control.

Why does AI in finance start with reliable data?

Without reliable data, there is no reliable analysis. That is my firm belief. Without a solid foundation you get nowhere, as we wrote in an earlier blog. Only when your data is sound and clean can you use AI to automate process flows.

How does AI help with your cash flow?

AI-driven systems spot cash flow bottlenecks early and help you resolve them. Finance is everything. You can make millions in profit, but if your cash flow is not in order, you can still go under.

Let me give an example. Every seasonal business needs more working capital in some periods than in others. Traditionally, that means hours of juggling payment terms and cash flow forecasts. Where your finance team would normally spend days analysing contracts, AI does it in seconds.

Say you pay all your suppliers within 35 days by default. The system sees that with certain suppliers your contract allows 45 or even 60 days. By using those longer terms wisely, you free up extra working capital in peak season, exactly when you need it most.

AI flags where liquidity dips are coming and suggests which suppliers you could pay later. The system calculates the effect on your cash flow at once.

So AI takes over the digging and the number crunching. That means fewer errors, more consistency and time to set the course of your business.

Start with the foundation: make sure your data is in order. Only then can AI really add value to your finance department. The added value of AI does not stop at finance, of course. In later blogs I talk about AI in operations and in sales and marketing.

Erry

Can you use every longer payment term?

Not always, which is why we add one caveat to Erry’s example. In the Netherlands, large companies must pay SMEs and self-employed suppliers within 30 days, according to the Dutch government’s Ondernemersplein. Between other businesses, 60 days is in principle the maximum.

A company counts as large if it meets two of three thresholds on two consecutive balance sheet dates. Those are 250 employees or more, net turnover above €50 million and total assets above €25 million. If you are a large company, any term over 30 days is already too long for an SME supplier. That includes the 35 days in the example, let alone an AI suggestion to pay after 45 or 60 days.

Belgium generally caps terms between businesses at 60 days, as our KPI guide explains. Have your legal adviser check your situation. If you pay after 35 days while the contract says 60, the contract terms are probably missing from your supplier records. You decide, supplier by supplier, which terms to use and what that does to the relationship.

Which finance tasks suit AI?

Repetitive, rule-based tasks with an outcome you can check.

  • Matching: linking a purchase invoice to its order and receipt, or a bank statement to open items.
  • Flagging anomalies: duplicate invoices, changed bank details or costs that slowly drift.
  • Forecasting: when customers will actually pay and what that does to your cash position.
  • Reading documents: turning invoices and contracts into data you can check.

Approval, review and anything that needs judgement stay with your people. Think of an anomaly that turns out to be correct, a customer on a payment plan or a supplier you want to keep. Our article on financial reporting covers the close and reporting from one system.

Which finance AI is already in IFS Cloud?

At least two, according to IFS: a forecast of when customers will pay, and AI that reads supplier invoices. Customer Payment Delay Prediction predicts for each invoice instalment how many days late or early a customer will pay. Switch it on in a cash planning scenario and the expected cash flows move by those days.

That forecast covers the receivables side. The supplier side of Erry’s example starts with the payment terms in your supplier records.

IFS recommends training the model on at least 24 months of invoice data. Predictions are made only for project invoices, customer order invoices and instant invoices, which have no link to stock. Lobbies, the dashboards in IFS Cloud, show whether the prediction matches what customers actually paid, before you act on it.

According to IFS, machine learning requires an IFS.ai subscription. If you run IFS Cloud yourself as a remote deployment, you first connect it to the IFS AI services. The technical documentation requires version 25.1.1 or later and outbound traffic to IFS. If you are still on IFS Applications 10, first read how the move to IFS Cloud works.

Training and predictions consume tokens, the usage units of IFS.ai. Include them in your cost estimate, alongside the other costs of IFS Cloud.

IFS Cloud reads supplier invoices with AI, after which a user checks the data and transfers the invoice. To automate accounts payable, read our article on TRAILD. IFS describes how IFS Intelligence lets you chase overdue receivables in one conversation. The chat prepares the change, and nothing is committed until you approve.

How do you stay in control of what AI does in finance?

With the permissions, approvals and audit trail you already have. IFS describes IFS Intelligence as working through the Industrial AI Harness. Every decision is logged, including declined recommendations.

AI works within those existing permissions, which is also where the risk lies. If one person can both create a supplier and approve a payment, the AI works within that same conflict.

IFS Cloud has the Segregation of Duties analysis for this. It shows users with access to areas that your own rules say conflict. A conflict is information only: access changes only when you revoke or split permission sets. Clear those conflicts before AI prepares finance actions.

You set which actions AI may prepare and who approves them. Agree that set-up with your auditor in advance.

Does the AI Act cover AI in finance?

Mostly not under its strict rules for high-risk AI. For finance, Annex III of the EU AI Act lists AI that assesses the creditworthiness of natural persons or sets their credit score. AI used to detect financial fraud is explicitly excluded.

Invoice matching, cash flow forecasting and reading documents are not on that list. If you use AI to assess the creditworthiness of natural persons, such as consumers or sole traders, look further. According to the European Commission, the rules for those uses apply from 2 December 2027.

Building your staff’s AI literacy is already a duty for every organisation that uses AI. The closing blog of this series explains what that means.

Where do you start with AI in finance?

With one frequent task whose result you can measure, such as forecasting customer payments. First check your data: contractual payment terms in your supplier records, 24 months of clean invoice history and a log of changes to bank details.

Sometimes AI in finance has little to offer yet. With few customers and invoices, or less than two years of usable history, cleaner master data and a fixed process pay off first.

Then measure how accurate your current forecast is and compare it with the prediction after a few months. Our guide to KPIs and metrics shows which measures help.

Questions about AI in finance

Who approves what AI proposes in finance?

Someone you appoint, not a person who can also create suppliers. The Segregation of Duties analysis in IFS Cloud shows such conflicts.

Which AI features for finance does IFS Cloud have?

According to IFS, IFS Cloud includes Customer Payment Delay Prediction and AI that reads supplier invoices. The first predicts for each invoice instalment how many days late or early a customer will pay.

How much history does a payment forecast need?

IFS recommends at least 24 months of invoice data to train the payment delay model. Retrain it regularly after that, so it reflects recent payment behaviour.

Which finance task would you hand to AI first?

Tell us where you stand with IFS. After one conversation, you will have a clear next step. Bring the task that costs your finance team the most hours today. Discuss your IFS question.

Stay one step ahead with IFS. Get release news and practical examples from IFS Specialized Partner Eqeep in your inbox. Free, and you can unsubscribe at any time.

Yes, keep me posted