Finance department heads grapple with one dilemma that defines their careers: They're responsible for the economic efficiency of their organizations, yet they themselves constitute a non-revenue cost center.
They're exactly the kind of overhead they need to minimize, so they need to make certain no line-of-business executive complains that the company lost a major sale because of friction from finance.
This would be hard enough without disruptions from geopolitical chaos, supply chain interruptions and interest rate uncertainty. The customers won't concern themselves with the reasons why their order can't be fulfilled; they'll just take their business elsewhere.
And, as the world is finding out, agentic artificial intelligence is no magic wand.
"While AI undoubtedly has had impressive achievements, many of those wins are narrow, context dependent, and, frequently, hard to achieve," according to
The trick, then, is to get the maximum out of AI by giving it something highly specific to do. We can only achieve that by putting in place the appropriate tools and processes needed to guide business decisions, then bolt on the AI layer that will speed up the process by automating those decisions.
If AI is the key to streamlining financial decision-making, companies first need to evaluate how AI is being integrated across operations, and where the gaps remain.
Starting at zero
Fragmented data and manual processes are CFO kryptonite. Unfortunately, some corporate headquarters seem to be sitting on kryptonite mines.
We can describe these laggard companies as standing Level 1 of a 1-to-5 scale of fintech maturity. They're still piloting small-bore projects to figure out how technology can improve their financial operations. Level 2s are more aware of what's possible but still in the process of adopting appropriate tools and processes. Most of your competitors are probably 2s and will soon be 3s — those that are leveraging the tech that's widely available and proven in the marketplace to deliver measurable results. The goal of Level 3 is standard processes and, at this point, companies migrate to cloud-hosted solutions.
The sweet spot on this maturity scale for a company past its startup phase is to be a 3 on your way to 4. Level 4s have added data-driven decision-making procedures to their automated processes. This is all about having predictable data to work with, so this is where AI comes in. Again, it doesn't help to plant AI without all that arable soil beneath it.
And that's something people and bots have in common: We can only make incremental improvements. It's hard enough as a CFO to move your department from 2 to 3 or from 3 to 4. It's virtually impossible to make a bigger jump.
"While some businesses might be inclined to skip levels or aim for a maturity level far off from their current location,"
Level 5, the bleeding edge of technology adoption, is the money-is-no-object scenario. Few companies need this level of optimization in their core competencies, and fewer still in their headquarters' operations.
Table stakes
The CFO of a Level 3+ operation needs
- Data integration: This results from deploying such enterprise resource planning systems as SAP, which orchestrate solutions for receivables, payables and overall payment solutions.
- Visibility: What gets measured gets done, and we can't measure what we can't see. Real-time executive dashboards are straightforward enough — you could build these in Excel if you had to — but compliance-reporting tools will need to be automated with fit-for-purpose solutions.
- Governance: Ultimately, though, visibility is useless without firm governance. Your firm might rely on key performance indicators, objectives and key results, or some other actionable metric. Whichever you choose is a religious decision, but it's important that you choose something and stick with it. Success must be definable.
A fourth element — advanced analytics — is where AI integration adds the most benefit. Predictive cash flow models, quantitative risk analysis and scenario planning tools all fall under this category and are all necessary to mature the operation to Level 4. AI has the most immediate, direct impact on cash management, collections and credit risk, although it can also impact invoicing and payables. We're only just beginning to see how AI adds functionality to and improves the user experience of customer portals.
Real-world impact
Decluttering decision-making processes for financial optimization requires the replacement of obsolete manual processes with one-click access to real-time cash flow data — allowing leaders to make faster and more accurate decisions, and leverage existing ERP software suites to craft an actionable treasury operations blueprint.
Every company that undertakes this journey toward AI-assisted financial decision making has a unique starting point and its own set of goals, so every path is different. But across all journeys, there are key considerations CFOs must keep top of mind:
- It all starts with the payment system. If it isn't giving you error-free, real-time data, then everything downstream is suspect.
- Security is crucial. This is preaching to the choir, but encryption, authentication and SOC2 compliance ensure that only people who are making corporate decisions based on the data ever get the chance to see it.
- Everything changes when you go cross-border. Currency effects emerge. Fees add up. Clearing slows. Compliance becomes an exercise in trying to please everyone. And the proliferation of internal corporate entities adds layers of complexity. The ability to centralize financial and treasury operations, then, becomes a sine qua non.
- You don't have to throw out your whole current system. If your company has already moved on from its initial growth spurt, it likely has some enterprise-class systems in place. The trick is to determine where the disconnects, bottlenecks and error sources are. Those are the places where you need to find something better.
- AI is not the destination. Faster, more fact-based decision-making is. Your degree of automation should meet but not exceed your need for it. If all you need your system to do in the foreseeable future is to follow a script, then old-fashioned rules-based automation should be fine. If you need it to help you make better forecasts about anything from market conditions to a specific customer's likelihood of paying on time, then predictive AI might be called for. If you're prepared to improve efficiency by delegating planning and simple decision-making to a fit-for-purpose bot, then you've entered the realm of agentic AI.
- Finally, there is no magic wand. You can't just cut over from one system to another — and you're more likely to be cutting over from multiple systems to a more centralized target state. It's going to take time to install the infrastructure, acquire the licensure, develop the systems, test the systems, then train the end users and find out why they might be resistant to accepting the new way of doing things. You need a partner who doesn't just show you what's wrong with your current treasury operations and show you a vision of a higher standard — you need one who can actually lead you to that better tomorrow.
At the close
It doesn't matter whether you're one of the biggest companies in the world, or whether you're a startup on the verge of mainstream success. It doesn't matter whether you now have predictable data from a single point of truth, or if that's some distant dream.
As long as you set realistic short-term goals to go with your visionary long-term aspirations, you can take steps now to improve your company's financial decision-making. And when you make better decisions, your clients become more loyal, your cost of capital dips, and you can spend more executive time focused on lines of business and less on headquarters processes.
