ditor's Note: This blog post is adapted from a transcript of our Manufacturing Minute podcast episode featuring Randy Lowman of Lake Turn Automation. The content reflects the conversational nature of the original recording.
During my Manufacturing Minute podcast conversation with Randy Lowman, I asked a question that I know many CFOs and controllers are thinking: "What's a safe first step to experiment with AI?" His answer was both practical and encouraging—there are many landmines out there, but there's also a clear path forward.
Start Small and Start Where You Already Are
Randy's first piece of advice? Don't jump into the deep end of the pool. Many organizations want to tackle their biggest, hairiest process first. That's exactly the wrong approach.
Instead, start with tools like Microsoft Copilot, which is now embedded in Excel and Outlook. While it's not perfect, it's getting better, and it already operates inside your existing security framework. There's training available, and the risk is manageable.
Not Every Task Requires AI
Here's something important that often gets lost in the AI hype: not every task requires AI. You want to use the most cost-effective solution to meet your objectives.
Many repetitive, rule-based processes in finance—like reconciliations, report generation, and AP triage—can be automated with simple RPA or workflow tools. Randy emphasized that workflow automation and RPA are great stepping stones before you layer in artificial intelligence. They carry minimal risk and build team confidence that automation isn't going to replace jobs, but rather augment capabilities and productivity.
Choosing Your First AI Use Case
When you do test AI, pick a well-defined use case. Look for something that is:
- Small in scope
- Clearly defined
- Has clean, digitized data
Finance teams are generally good about having clean data, which gives us an advantage. Some possible first experiments include:
- Using AI to summarize monthly results
- Automating spend categorization
- Generating a first draft for variance analysis or explanations
Set Up Success Factors Upfront
As finance professionals, we love ROIs—and that's exactly what you need here. Every AI proposal should answer three fundamental questions:
- What problem does it solve?
- How much time or money does it save?
- How will we measure it?
If you can't answer these three questions, why are you doing it? Define your success metrics before you start. Think about time savings, risk reduction, and error reduction. Put numbers on those benefits, then measure during and after implementation. Iterate and repeat.
Build an AI Council
One of Randy's best suggestions was to create what he calls an "AI council." Gather curious team members together to explore, stay safe, teach each other, and experiment. Getting a team together to roll up their sleeves will do more than months of planning ever will.
More importantly, this approach increases team buy-in along the way. When people are involved in the exploration and decision-making process, they're much more likely to embrace the changes rather than resist them.
Evaluating AI Vendors and Solutions
The noise around AI can be overwhelming. Between new startups and established enterprise solutions, the pitches are incessant. This is where finance professionals naturally shine—we're good at asking for the numbers.
A Structured Evaluation Framework
Randy shared the structured methodology his firm uses to help clients build an AI framework. When developing your AI strategy, consider these four pillars:
Vision: What's your vision for AI? What role do you see it playing in your organization?
Value: What value do you see in AI? What specific benefits do you expect?
Risks: What potential risks do you see in AI adoption? How will you mitigate them?
Comfort Level: What's your organization's comfort level with AI adoption?
When making decisions about specific AI projects, evaluate each opportunity using four filters:
Impact: What's the impact on your organization? Consider operational lift, sales potential, or reduction in operating expenses.
Feasibility: Can your team handle this? Do you have the technical capability, the right people, and the necessary resources?
Risk: Can you manage the risks? This includes data security, implementation challenges, and potential business disruption.
Cost: What's the total investment required, including not just software costs but implementation time, training, and ongoing maintenance?
Score all your opportunities against these criteria. If a project doesn't score well across these filters, it's probably not worth pursuing.
Be Cautious with New Vendors
There's a new AI vendor with a new "gizmo" seemingly every day. Randy warned that you have to be careful with newer vendors because they often operate too close to what he calls "the blast zone" of major players like OpenAI.
These smaller vendors face real risks. They might get acquired, become obsolete, or simply vanish. Even worse, you may not know where your data goes when you use their solutions.
Randy's recommendation? Stick with established providers—companies like Oracle and CCH. Yes, they might not roll out the flashiest features as quickly as startups, but from a privacy and safety perspective, you're better off with someone who has a proven track record.
The Cultural Component
Getting started with AI isn't just about technology—it's about culture. You need to create an environment where it's safe to experiment and even fail. This can be challenging in accounting, where we're used to everything balancing perfectly and where failure traditionally equals bad outcomes.
But with AI, many experiments won't work, and that's okay. That's how learning happens. That's how you discover what AI can and can't do, so you can make informed decisions or test again in twelve months to see if the technology has improved.
Safe experimentation with guardrails is key. It builds both confidence and capability across your team.
My Takeaway
The path to getting started with AI doesn't have to be intimidating. Start small, use established tools, involve your team, and measure your results. Build on small wins and learn from experiments that don't work out as planned.
Remember, the goal isn't perfection—it's progress. Give your teams the freedom to explore, learn, and adapt. That's what a future-ready finance team looks like.
Separating hype from value isn't about rejecting innovation—you need to embrace it. It's about discipline. With a clear strategy, alignment with business objectives, measurable impact, and trusted providers, you can start your AI journey and see real business value.
-1.jpg)
• 27 min
AI in Finance: From Hype to Real Business Value
Manufacturing Minute
Play