Implementation Trends and Real-World Use Cases
Every finance leader is hearing the same pitch. AI will transform your operations. It will cut costs, improve accuracy, and free your team to focus on strategic work. Investors are asking about your AI roadmap. Vendors are flooding your inbox with tools that promise to automate everything from close to forecasting.
The problem isn't a lack of options. It's too many of them, with no clear framework for what actually works.
That noise was the backdrop for an F Suite Strategy Summit panel featuring Lee Taylor (CFO, OnlyFans), Kash Mathur (CFO, Bungalow), Joanne Cheng (CFO, Jellyfish), and Michael Belicose (CFO, Campus).
This article breaks down the barriers that make AI adoption difficult in finance, shares three real-world examples of how these CFOs are using AI to solve specific problems, and outlines how to take your first meaningful step in the next 30 days.
The Main Challenges of AI Use in Finance
AI implementation in finance runs into problems that other functions don't face. Marketing can test a new tool and pivot if the output misses the mark. Engineering can sandbox experiments without risking production systems. Finance doesn't have that luxury.
The stakes are higher because finance is the steward of capital. Every forecast, every close entry, every investor report has to be accurate. Trust but verify isn't just a principle; it's the operating model. And that creates friction when introducing AI finance tools and other AI-powered workflows to your organization.
Some of the main barriers to broader implementation include:
The hallucination problem. Language models can generate confident answers that are completely wrong. They can misinterpret data structures, apply incorrect formulas, or reference metrics that don't exist. For a function where a single decimal error can throw off board reporting or investor confidence, that risk makes adoption feel dangerous.
Tech debt and shadow AI. When individual team members start experimenting with AI tools without documentation or governance, you create shadow AI. The business becomes dependent on a process that only one person understands, the same vulnerability that happens with an overly complex spreadsheet model. If that person leaves or the tool changes, you're left with a black box that no one can maintain.
Technical proficiency gaps. Finance and accounting have historically been slower to adopt new technologies than other professions. There are forward-thinking leaders pushing experimentation, but many teams lack the technical fluency to get the most out of AI tools. The challenge is finding the balance between encouraging curiosity and maintaining the rigor that finance requires.
These barriers are real, but they're not insurmountable. As CFO Kunal Agarwal said in a panel on designing smarter finance orgs, "we don't fully know what's possible with these tools yet, so curiosity is the point." The finance leaders making progress aren't waiting for perfect conditions. They're testing carefully, documenting what works, and building systems that let them move faster without sacrificing accuracy.
General Trends in AI Implementations for Finance Teams
Before looking at specific use cases, it helps to understand the patterns shaping how finance teams are approaching AI. The panelists identified three themes driving adoption across their organizations.
Shift from cost reduction to capacity expansion. Early AI conversations centered on cutting headcount or reducing expenses. Now the focus is on enabling teams to take on higher-value work. AI handles the repetitive tasks so finance professionals can spend time on analysis and strategy.
Specialized AI leadership hiring. Companies are creating roles specifically to manage AI adoption. These aren't traditional IT positions. They're hybrid roles that understand both the technical capabilities of AI tools and the operational needs of the business.
Custom discrete solutions. Off-the-shelf AI tools promise to solve everything, but the most effective implementations are narrow and purpose-built. Teams are identifying specific friction points and building or customizing solutions to address them rather than deploying broad platforms that try to do too much.
These trends reflect a maturation in how finance leaders think about AI. The hype has given way to practical experimentation focused on solving real problems.
3 Real-World Examples of How To Use AI in Finance
The most effective AI implementations in finance start with a specific problem. Not just "how can we use AI?" but "where is manual work creating bottlenecks, errors, or capacity constraints?"
The following use cases show how three CFOs identified friction points in their operations and built AI solutions to address them.
These examples may not map directly to your business, but they demonstrate processes for finding opportunities and testing solutions that fit your constraints.
1. Prioritizing Human vs. Automated Tasks
Kash Mathur at Bungalow faced a straightforward problem. His property management business needed to control costs while maintaining service quality for thousands of tenants. The obvious target was maintenance coordination, a high-volume, repetitive function that consumed significant resources but didn't always require human judgment.
The team built an AI system to handle routine maintenance requests. When a tenant reports a leaky faucet or a broken light fixture, AI triages the issue, schedules the repair, and coordinates with vendors. Simple problems get resolved without human intervention. Complex issues that need judgment or escalation get routed to a person.
This freed the customer success team to focus on situations where human touch actually matters: conflict resolution, lease renewals, and tenant retention. The AI changed what people spent their time on, cutting time spent on mundane maintenance tasks by 85%.
Mathur's philosophy for identifying AI opportunities is simple. Look for tasks that are high-volume, rules-based, and don't require nuanced decision-making. Those are the candidates for automation. Everything else stays with humans who can exercise judgment and build relationships.
For finance leaders, the broader lesson applies beyond property management. Customer success, accounts receivable follow-up, vendor onboarding—any function with repetitive workflows and clear decision trees is a potential candidate. The question is whether automating it frees your team to do work that matters more.
2. Using Predictive Analytics To Improve Customer Retention
Michael Belicose at Campus runs a business built on human connection. The platform helps universities create online communities for students, and retention depends on engagement, belonging, and relationships. Those aren't things you automate.
But Belicose found a way to use AI to protect revenue without replacing the human elements that make the product work. His team built predictive models that analyze user behavior to identify students at risk of churning. When the model flags a student, it doesn't trigger an automated email sequence. It alerts a community manager who can reach out personally, understand what's happening, and intervene in a way that feels genuine.
The system works because it separates what AI does well from what humans do well:
AI handles pattern recognition. The model tracks engagement patterns, login frequency, participation in discussions, and other signals that indicate someone might be pulling away from the platform.
Humans handle the relationship. Community managers receive the alert and decide how to intervene based on context the AI can't interpret.
This approach reflects Lee Taylor's philosophy from the panel about human-first design with AI in finance. The technology should serve people, not replace them. In this case, that means using backend analytics to surface opportunities for human intervention rather than trying to automate the intervention itself.
For finance leaders, this model works in any business where relationships drive retention. Subscription services, professional services, B2B platforms—AI can identify the early warning signs that a customer is disengaging. What you do with that signal determines whether you keep them.
3. Tracking AI Spend To Avoid Inefficient Automation
Joanne Cheng at Jellyfish faced a problem that's becoming common as AI tools proliferate. Her engineering teams were adopting AI coding assistants and other productivity tools, but no one had visibility into whether the spend was justified. AI is supposed to drive efficiency, but without measurement, it's easy to overspend on tools that don't deliver returns.
Jellyfish's platform analyzes engineering productivity, so Cheng had the infrastructure to track this. Her team started measuring how individual engineers and teams were using AI tools, what types of tasks they were applying them to, and whether those applications were actually improving output or velocity.
The data revealed gaps. Some teams were using AI for tasks where manual processes were faster or more accurate. Others were paying for premium features they rarely touched. In a few cases, different departments were paying for overlapping tools that solved the same problem.
This visibility gave Cheng leverage in two ways:
Budget discipline. She could have informed conversations about which AI investments were worth continuing and which needed to be cut or consolidated.
Smarter automation decisions. The data showed when a different approach, whether a simpler automation or a manual process, made more sense from an ROI standpoint.
For finance leaders, the lesson is that AI adoption needs the same rigor you'd apply to any other capital allocation decision. Just because a tool uses AI doesn't mean it's worth the cost. Track usage, measure outcomes, and be willing to pull back when the returns don't justify the spend.
How to Start With AI in Finance in the Next 30 Days
If you're feeling behind on AI or overwhelmed by the options, the answer isn't to build a comprehensive roadmap or evaluate dozens of tools. Start smaller.
The panelists offered a practical first step: identify your highest friction spreadsheet. The one that takes hours to update, breaks when someone changes a formula, or requires manual data pulls from multiple systems. That's your target.
Spend the next 30 days working to automate it. This might mean building a discrete solution with help from your engineering team, or it could mean testing a purpose-built tool that solves that specific problem. The scope matters less than the discipline of focusing on one real pain point and solving it completely.
This approach does two things. First, it gives you a tangible win that builds momentum and credibility for future AI projects. Second, it teaches you how to evaluate AI solutions in practice rather than in theory. You'll learn what works, what breaks, and how to document and maintain automated processes so they don't become technical debt.
The finance leaders making progress with AI aren't the ones with the most ambitious visions. They're the ones solving specific problems, one at a time, and building confidence through repetition.
The F Suite helps you make better decisions, faster—backed by an active community of your peer CFOs.