Will AI Agents Take Over CFO Offices?

Welcome to the first edition of 2025! As we step into this transformative year, we’re diving headfirst into one of the most pressing topics in finance: AI agents. With their rapid evolution and growing presence, these tools are poised to redefine the CFO office and reshape how we approach finance operations.

In this issue, we’ll start exploring how AI agents are changing the game. From understanding their capabilities and limitations to examining real-world applications, we’ll cover everything you need to know to prepare for their inevitable integration into your workflows.

Additionally, I’m thrilled to share updates on upcoming events where I’ll be speaking about the future of AI in finance. Let’s set the stage for deeper discussions and practical insights to guide you in the year ahead.

Let’s dive in!

Announcing strategic partnership with Octopus AI

I’m excited to announce my collaboration with Octopus AI, an innovative AI-powered budgeting tool created by Lena Levin. Over the past year, I’ve been experimenting with AI in budgeting, and I know there’s so much potential here. AI can completely change how we forecast, plan, and manage budgets, making it faster and smarter.

That’s why I’m thrilled about Octopus AI. It combines Lena’s deep understanding of the challenges finance teams face with cutting-edge technology to make budgeting more efficient and insightful.

If you’re curious about what AI can do for your budgeting process, I’d highly recommend booking a demo with Lena. She’s not only incredibly knowledgeable but also passionate about helping finance teams unlock their full potential.

Also, don’t miss her LinkedIn newsletter—it’s full of practical tips and trends for using AI in finance.
 

The Balanced View: What Finance Professionals Need to Know about AI Agents

Many experts predict that 2025 will be the "Year of the AI Agents." Over the past year, we’ve seen this technology evolve, maturing enough to tackle traditionally conservative finance processes.

Here’s my take: regardless of your role in finance, you’ll encounter AI in one form or another in 2025. And if you’re in a management or executive position, chances are you’ll be responsible for integrating “AI agents” into your workflows and have it among your goals.

AI Agents vs. Traditional Automation

Let me make this clear: AI agents are not workflows, not RPA, and definitely not just software programs you design, deploy, and forget about. They are artificial workers with true agency—requiring management, supervision, and feedback, just like any other employee in your company. Unlike traditional automation tools, AI agents are dynamic, adaptable, and capable of learning from every interaction, making them indispensable for handling unstructured, unpredictable tasks in complex finance processes.

We must also prepare to address new types of Principal-Agent challenges that these intelligent workers will introduce. Supervision, direction, and control will need to evolve. Working with these agents means adopting new management strategies, as they are no longer pre-programmed software but sophisticated, evolving contributors to your team.

So, let’s start by breaking down what AI Agents are, how they work, and what you need to know to make them work for you.

What is an AI Agent?

An AI agent is not just another software program; it’s a highly capable system designed to handle complex tasks that demand decision-making, adaptability, and creativity. Here’s what sets them apart:

• Environmental Perception: AI agents analyze data and contextual information to make decisions.

• Tool Utilization: They leverage various software tools to execute tasks efficiently.

• Adaptive Learning: They improve their performance over time based on outcomes and feedback.

• Strategic Planning: They manage multi-step workflows to achieve specific objectives.

Unlike traditional automation tools, which rely on predefined algorithms and static rules, AI agents are dynamic and adaptable. We will truly delegate tasks and processes to them.

What Do AI Agents Look Like?

AI agents come in many forms, tailored to different organizational needs:

• Software-as-a-Service (SaaS): AI-powered platforms that handle a task or an entire function, for example, Octopus AI for budgeting, Hyperbots for accounts payable, or Numeric for accounting close.

• Prebuilt and Customizable AI Agents: Prebuilt AI agents specifically designed for finance are gaining traction. For example, 913.ai offers agents for document automation, and Zbrain provides a marketplace of AI agents for tasks like scheduling client payments and monitoring loan covenants.

• Custom Automations: Tailored integrations connecting your ERP, accounting system, or databases with public or private AI models. You will likely need an internal team or an agency to develop these custom solutions, but the result will be fully tailored to your specific needs and workflows.

• Integrated AI Components: AI-powered features embedded within existing software. For instance, Bill.com’s “invoice capture” function reads invoice data, validates and verifies it, and inputs the extracted information into an ERP or accounting system. This is an AI agent in action, even though it isn’t explicitly labeled as such.

Technological Building Blocks of AI Agents

AI agents rely on several core components:

• Generative AI Models

• Integration Tools

• Task-Specific Tools

• Data Handling Systems

A system's AI component may be highly visible or almost invisible, and this level of visibility significantly affects users' engagement with the technology.

Understanding where AI functionalities begin and end within these configurations is critical to successfully managing, supervising, and controlling AI systems. This task requires a significant shift from traditional automation frameworks. Unlike static tools, AI agents require constant evaluation, feedback, and fine-tuning to ensure optimal performance.

AI Agents in Action: Account Payables

Accounts payable processing, critical in accounting, is often manual because of diverse data formats and the expertise needed for accurate expense classification. Previously, this variability hindered automation, but AI agents now transform it with intelligent workflows that manage diverse inputs and ensure accuracy.

Traditional Accounts Payable Workflow:

1. Invoices received: Invoices arrive in various formats, such as PDFs, emails, or paper copies.

2. Manual data entry: Staff input invoice details, such as vendor names, amounts, and dates, into the accounting system.

3. Invoice review: Accountants check invoices for errors, missing information, and inconsistencies.

4. Expense classification: Accountants manually classify expenses and record transactions in the system.

5. Compliance checks: Larger and public companies conduct regular audits of booked transactions to ensure compliance with accounting policies, company guidelines, and contract terms.

AI-Enhanced Workflow:

1. Data Extraction: AI agents automatically read and extract key details from invoices, such as vendor names, invoice amounts, and due dates, regardless of the format or medium.

2. Expense Classification: Using advanced pattern recognition and historical data, AI categorizes expenses accurately and efficiently, reducing manual input errors.

3. Compliance Verification: AI agents cross-check invoices against contracts, accounting rules, and company policies, ensuring compliance before posting.

4. Anomaly Detection: The system flags incomplete or unusual data entries for human review, allowing accountants to focus on higher-value tasks.

5. Automated Posting: Once validated, invoices are seamlessly posted into the accounting system without further human intervention.

The new AI-enhanced workflow increases the speed and efficiency of the accounts payable process and minimizes the need for post-audits, as all the necessary checks are performed at the time of booking.

Additionally, AI can automate payment scheduling by analyzing payment terms, ensuring timely disbursements, optimizing the position of transactions in the payment queue, and enhancing cash flow management.

Account Payables should no longer be a manual process in 2025! Depending on your company’s size, a pre-built AI agent or a SaaS solution like Hyperbots can and should streamline this workflow, saving time and reducing errors.

Do you need help choosing the right solution or guiding your implementation? I’m here to assist! Feel free to book a free consultation, and we can discuss how to make this transition seamless for your business.

Account Payables is the ideal process for an AI agent. It is repetitive, highly manual, and challenging for traditional automation methods.

There are multiple other processes in the CFO office that are a great target for AI agent implementation. They all have several key characteristics:

• High Variability: These include tasks where the inputs vary widely and unpredictably. For example, fraud detection often involves analyzing diverse transaction patterns to identify irregularities that traditional systems might overlook.

• Data-Intensive Workflows: AI agents excel in workflows that require processing and analyzing large datasets quickly. For instance, an AI agent can sift through years of financial data to predict cash flow trends or identify performance anomalies in real time.

• Complex Scenarios: Processes that involve adaptive learning and pattern recognition, such as customer segmentation or dynamic pricing strategies.

While AI agents bring transformative potential, there are scenarios where they add little to no value:

• Tasks with Limited Variability: Processes such as basic bookkeeping or calculations that follow static, well-defined rules. Traditional automation is more effective in these static cases.

• Low Data Availability or Quality: Scenarios where historical data is insufficient or inconsistent, such as predicting trends for a newly launched and untested product category.

• High-Level Strategic Decisions: Activities requiring nuanced human judgment, such as crafting corporate strategy or resolving complex stakeholder disputes.

• Interpersonal Dynamics: Situations involving direct client interactions, conflict resolution, or mentoring.

The potential of AI agents is enormous. They can help scale operations without increasing team size, achieve higher efficiency, reduce errors, and provide better insights for decision-making. However, successfully implementing AI agents, managing them effectively, and maintaining their performance is not straightforward. 

In the next newsletter, I’ll share a comprehensive framework for successful AI Agent integration and a step-by-step guide for implementing, maintaining, and troubleshooting these powerful tools.

Announcements and updates

Two events that I invite you to attend if you are a finance professional interested in AI: 

1. My co-webinar with the Caprus.ai team on January 30th, where we will be discussing AI Beyond Chatbots. I am looking forward to this session and expect to learn a lot about the technical side of AI implementation in finance.

Sign up here

2. On March 27th, I will be on the panel at the FInEX summit, discussing how AI will Transform the CFO's office. The event is free, and my fellow experts are amazing, so it's one you can’t miss.

Sign up here

This is going to be a dynamic year! I can’t wait to see what’s coming!

We Want Your Feedback!

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation! 

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina 
AI-Powered CFO and AI Implementation Consultant