If you took a month off next Monday, what in your week would stop happening because the logic for it lives only in your head? For most Financial Directors I meet, the answer is the rolling cash forecast. This is the first in a short series with Ronan Rooney on how Irish businesses are putting AI to practical use, and I want to start with a growing manufacturer’s weekly cash forecast.
The Problem
The client is a multi-entity industrial and manufacturing group that has scaled quickly, with new product lines, customers, and supply routes. Like most mid-sized companies, it put the people in place first and told them to do their best. It has finance, operations, and an ERP system, but the systems do not talk to each other.
When I first sat in with the board earlier this year, I told them I knew a bit about their business, and that they knew it far better than I did. I asked them to tell me their problems and how they were using AI. Most of what they described, from operations to sales, funnelled into one person: the Financial Director.
Every week, he had to produce a cash flow projection. That meant pulling debtor and creditor ageing, production plans, ERP materials tracking, sales, budgets, payment plans, and outsourced pipelines. Logistics updates landed on a Friday and the production plan on a Monday, and none of it arrived in one place. He was ringing people and walking down the corridor to check invoices, whether a delivery had happened, and whether production volumes were real. The judgement calls were the harder part. Which customer or supplier sits on which line, when each will really pay, what to write down as doubtful, and which overheads were already in the budget. The dynamics and pace of the business meant that all of that lived in his head, and it took six or seven hours every week.
Nobody in that boardroom saw the hours as the problem, including the FD. They were used to the old process and comfortable working long hours, because that is how a small, fast-growing business runs. In my experience, people operate the way they think they should operate until they are given an alternative.
The Solution
We started with the output and worked back. I asked the FD to show me what he produced each week, and we traced each line back to its origin in the business. Only then did we ask which parts of the work AI could help automate, and how much we could take off his plate.
We used AI to map the process, prototyped quickly, and connected the core systems, so the forecast now reads live from the accounting system, with open invoices and bills on their real due dates. Some of the information was unstructured, for example, the scanned paperwork that comes in with the trucks, so we used AI to capture that too. If a connection ever fails, the system falls back to the manual file, so the forecast never stops.
So we had AI run over everything we had captured and draft his rules, then went back and forth with him three or four times until they matched how he thinks. After that, we shadow-ran his cash flow process with AI working beside him until he trusted it.
The system now works in three steps:
- It reads the incoming data and proposes where each new debtor or creditor should sit, when they are likely to pay, and how they should be treated.
- The FD accepts or overrides each proposal with a click, and a weekly inbox shows him only the exceptions.
- Decisions he makes repeatedly become rules, and he can see how often the system’s suggestions match his own past choices.
The judgment is still his, and what we did was capture it.
The Challenges
Three challenges come up in every project like this: security, sustainability, and educating the teams upstream whose data the process depends on.
Security. The first question is always about security. The platform uses enterprise models that do not train on the client’s data, and everything that passes through it is encrypted and secured in line with ISO and SOC 2 standards, regardless of the process we run.
Sustainability. The second question is, “will this run if I am not here?” Unlike a chat, this is a process that is an accessible part of the business’s infrastructure. It becomes an asset that sustains the business beyond the person in the seat.
Education upstream. Operations, logistics, and sales all feed information into the forecast, so each team needed to understand how their data would be used further down the line, and to see themselves connected as one business. That was mostly an education piece. What made the difference was having the founder on board as an advocate for an AI-native way of working, since in a mid-sized company the FD will always seek that approval first.
Timing mattered as well. When we first spoke earlier in the year, they were not ready, with a new product line launching and board reports due. You need a bit of patience with a small business, and once they were ready, they went at it full bore.
The Result The weekly build is now done before the FD opens it. Work that took him six or seven hours a week to chase and compile now takes about fifteen minutes to review, which is roughly a day a week back.
For me, that is the real headline, because a Financial Director should be thinking about better use of the business’s cash and assets, not assembling a spreadsheet.
The forecast is also current. The old one was only as fresh as the last manual build, and now it is as fresh as the last run, so the board sees this week’s position. It exports to Excel and flows straight into the board pack, in whichever format each audience needs.
Takeaway
Pick the one thing in your week where the logic lives only in your head. It is usually the one everyone comes to you for, and the one you would hate to hand over if you took a month off. Start with what it produces, trace each part back to its source, and capture the judgment calls you make along the way. If the business can own those rules, you have taken key-person risk off the board for good.
If you would like to talk through a process in your own business, I would be glad to hear from you. You can reach me at ronan@parklanehq.com, and I am always happy to meet for a coffee.
Ronan Rooney is Co-founder and Director of Park Lane, an AI transformation firm, and its platform company Oak & Nine (see here>>>)
