Africa has leapfrogged earlier than. Thousands and thousands of individuals went straight to cell cash with out ever writing a cheque. I believe the finance operate now has an identical alternative with AI, and the CFOs who take it won’t be those with the most recent ERP. They would be the ones who cease ready for one.
The objection I hear most frequently from finance leaders is an affordable one: our programs are usually not prepared. The final ledger sits in an ERP that was customised a decade in the past. Subsidiaries in several international locations run completely different charts of accounts. Financial institution statements arrive as PDFs and far of the reconciliation lives in spreadsheets. The idea is that AI has to attend till all of that is changed.
I don’t suppose it does. In over a decade of constructing customized AI, a lot of it for finance groups, I’ve discovered that probably the most helpful programs have a tendency to sit down on high of the present property moderately than substitute it. Fashionable AI is nice at studying what individuals already produce: exports, statements, invoices and spreadsheets. That makes it a bridge throughout previous programs.
Listed here are three locations I’d begin.
1. Month-end shut: let AI do the studying, preserve individuals on the judgements
Many of the shut will not be accounting judgement. It’s accumulating, matching and chasing. Financial institution strains want matching, intercompany balances want agreeing and somebody has to clarify why a steadiness moved.
AI brokers can tackle a big share of that work. They’ll pull the trial steadiness export, learn the financial institution assertion in no matter format it arrives, suggest matches, flag the objects that don’t reconcile and draft variance commentary for an individual to verify. None of this wants a brand new ERP. It wants learn entry to the outputs the previous one already produces.
What the agent should not do is publish journals unchecked or signal something off. The rule I recommend to each finance crew is easy: the agent prepares, a named individual approves, and each proposal leaves a report of what the agent checked out and why it advised what it did. When the auditor asks how a quantity was reached, “the AI did it” will not be a solution.
2. Money forecasting: begin with the information you already belief
Money is the place many African CFOs really feel probably the most stress, as a result of cost behaviour, foreign money availability and buyer liquidity can change rapidly. I’d begin small.
Most companies already maintain the uncooked materials for a greater forecast: receivables ageing, provider phrases, historic receipts by buyer, payroll dates and mortgage schedules. A sensible first step is a mannequin that learns how every main buyer really pays, moderately than how the bill says they’ll pay, and refreshes the thirteen-week forecast each week. The worth will not be an ideal prediction. It’s a forecast that exhibits its working, so the treasury crew can see which assumptions drive the quantity and problem them.
Run the AI forecast alongside your current one for 1 / 4 earlier than you depend on it. If it can’t beat a cautious spreadsheet by yourself knowledge, you’ve realized one thing helpful.
Learn additionally: Mastercard and Interswitch land Google Pay in Nigeria as tap-to-pay goes mainstream
3. FX publicity: visibility earlier than cleverness
For a lot of companies on the continent, foreign money threat is a each day working concern, however the publicity is commonly scattered throughout contracts, income streams and intercompany loans.
AI helps right here first by discovering and organising the publicity. Studying contracts and buy orders to establish which commitments are priced through which foreign money, and after they fall due, is precisely the form of doc work trendy AI does nicely. Bringing that into one view, refreshed weekly, offers the CFO one thing many shouldn’t have as we speak: a dwell image of what the enterprise owes and is owed, by foreign money and by date.
The way in which cash strikes throughout African borders can be altering. Initiatives such because the Pan-African Cost and Settlement System, which permits cross-border funds between African international locations to settle in native currencies, give finance groups extra choices. A transparent publicity map makes these choices seen. Hedging choices stick with individuals; AI offers them higher info.
What to get proper earlier than you begin
First, knowledge safety. Finance knowledge contains salaries, buyer particulars and financial institution info, and regulators throughout the continent are more and more energetic, from the Nigeria Information Safety Fee to Kenya’s Workplace of the Information Safety Commissioner. Know the place your knowledge goes when an AI device processes it, and select instruments and contracts that preserve you compliant.
Second, possession. Each AI course of in finance wants a named proprietor who understands what it does, evaluations what it produces and may change it off.
Third, sincere measurement. Select one measure per use case earlier than you start: days to shut, forecast accuracy in opposition to actuals, or hours spent on reconciliations. If the quantity doesn’t transfer, cease or change course.
At Vital Future we now have seen that the finance groups who do nicely with AI are not often those with probably the most superior expertise. They’re those who begin with a selected, unglamorous drawback, preserve an individual accountable for each quantity, and construct from there.
The leapfrog
The companies that skipped landlines didn’t look forward to an ideal copper community. African finance features can do the identical: put AI to work on the paperwork and exports they have already got, preserve judgement and sign-off with individuals, and modernise core programs on their very own timetable moderately than treating that as a precondition.
The legacy programs will nonetheless be there subsequent yr. The true query is whether or not your finance crew will nonetheless be spending month-end retyping what these programs produce.
Concerning the creator
Adam Riccoboni is Founder and CEO of Vital Future, an AI technique and knowledge engineering consultancy specialising in finance operate AI. He’s the creator of “The AI Age” (2019), co-author and editor of “Engineering Arithmetic and Synthetic Intelligence” (Routledge, 2023), and lectures in synthetic intelligence at ESCP Enterprise College.
