Almost every sponsor and board member has given the same mandate: The CFO needs to prioritize AI: now. And yet, the finance function has been slow to adopt.
On the face of it, it may seem a puzzling disconnect, but it’s not. AI deployment is daunting, especially for CFOs who aren’t necessarily technologists.
And it may be even more daunting for the CFOs who know a little something more about tech. Or rather, for the CFOs who understand that effective AI requires a strong data environment. It’s not that they don’t know where to start with AI; it’s that they don’t think their data environment is ready for it.
We’re here to tell you: you’re letting perfect be the enemy of good-enough.
You don’t need to wait for a perfect data environment to kick-start AI—nor should you, if you want to meet sponsor and market expectations. Instead, you can start with more discrete steps. Here’s how.
Pinpoint priorities.
There’s no easy AI button; we can’t just snap our fingers and AI-ify all the less-than-efficient processes. What CFOs can do, however, is identify select areas that would most benefit from automation—and then start getting them AI-ready in digestible phases. This means focusing on specific, high-impact use cases where AI can drive real value. Look for a select few areas: a) where manual processes dominate b) where business units are already data-aware or tech-forward or c) where a quick win could demonstrate AI credibility and build internal momentum.
For example, the finance department processes hundreds of vendor invoices manually per week. That involves data entry, verification, approval routing and other tasks. It’s time-consuming and error-prone, and delays the month-end close. Deploying Visual-AI capability with automated processing can be a major win with business impact. Now, it’s not a one-size solution that will address all your manual bottlenecks—but it doesn’t have to be.
Clean (just enough of) your data.
Yes, AI is built on solid data foundations, but you don’t need to solidify your entire universe of company data right now. You’re looking at the data that powers the high-impact use cases you’ve identified (such as invoice processing)—and asking yourself questions that will inform how you organize, cleanse, and leverage that data: Do I have access to the correct data? How complete, consistent and trustworthy is it? Where are the gaps?
The goal here isn’t perfection. Addressing these questions helps produce a right-sized data set that’s more than “starter” data but less than a multi-year investment. That often means mining for foundational data domains like customer, project/SKU, pricing and transaction-level financials—data that’s high-value, yet more manageable to organize and clean, and can lay a solid foundation for effective AI readiness.
Note, too, that you can use AI to empower AI. Gen AI accelerates the process by cleaning, transforming and enriching your data.
AI-align your people.
AI readiness initiatives can’t succeed in isolation. They touch cross-functional workflows, impact decision-making and reshape how people work in their day-to-day roles. That means your people need to be part of the process.
Get them on board early. Align priorities with business leaders, making sure they understand what’s changing and why, and address their concerns proactively by framing AI as a job enabler rather than a threat. Securing people’s buy-in is the key to making your AI successful and sustainable.
Make AI part of your processes.
Your data and people are secure and aligned. Now, it’s time to integrate AI into how your business operates. Too many AI projects stall at this stage; CFOs are often paralyzed by the numerous business processes that require efficiency gains. The good news, though, is that you’ve already targeted the key use cases in which to embed AI—ones that have passed the test of business relevance, feasibility and data readiness.
Now you just need to operationalize. Take, for example, the identified invoice processing use case. Operationalizing here means:
- Designing workflows that incorporate AI outputs in a repeatable way—here, automating invoice intake, extraction and routing;
- Building feedback loops as users engage with the AI (accepting, correcting or overriding its recommendations) that capture feedback to refine your model’s logic and improve accuracy over time; and
- Securing more than just buy-in from your team—they need training to understand and trust the system.
Don’t let tech (or governance) get in the way.
Technology and governance matter. But they’re there to support rather than steer your business. That means the tech and governance health check should come after you’ve solidified your AI goals.
For the tech stack, ensure you have core systems in place (e.g., ERP, CRM) to support the AI services that align with your use cases. Additionally, enable a seamless data flow across systems and utilize flexible, cloud-based technology that can scale as your organization evolves. Your organization (and AI goals) will evolve, so you shouldn’t invest in implementing a huge, custom tech platform before you know where your AI journey will take you.
When it comes to governance, the reality is that many boards or sponsors won’t let anything AI-related move forward without it. But compromise exists; you don’t need a full-fledged governance model to get the show on the road. Start with “light touch” governance: have secure, enterprise-grade tools with clear usage guidelines and a process to prioritize use cases based on business goals.
The AI race is happening now. Here’s what we know: winning doesn’t mean deploying all the AI, all at once, but losing means not getting started at all.
