If you spend any time reading about AI in business, you usually see two extremes. On one side, tech evangelists tell you to automate your entire business tomorrow. On the other side, skeptics warn you about hallucinations, data privacy, and compliance nightmares.

If you are a founder or a proactive finance pro, both extremes are useless. You just want to know how to use the tool to fix your operational bottlenecks without breaking your business.

I recently co-authored a peer-reviewed academic paper published in the Journal of Social Impact in Business Research alongside Associate Professor Habib Khan from the University of Canberra. The paper is titled: “Treat GenAI like an employee: a conceptual framework for the use of GenAI in small business.”

But I didn’t write this just to sit in an academic journal. I wrote it to codify the exact framework I use to build automated financial engines for my clients.

Here is the strategic blueprint for adopting AI safely, efficiently, and highly profitably.

The Problem: Why Small Businesses Stall on AI

When we look at why small businesses hesitate to adopt GenAI, it almost always comes down to a lack of trust and high perceived risk. Specifically, there are five ethical barriers slowing people down: Fairness, Accountability, Transparency, Accuracy, and Autonomy.

Business owners are (rightfully) afraid that an AI will send a wrong number to a client, hallucinate a tax rule, or compromise confidential data.

To solve this, we developed the EthAI-SB Framework. The core concept is slightly disruptive, but highly practical: Stop treating AI like a piece of static software (like Excel). Instead, humanise the technology. Treat GenAI exactly like you would treat a human employee.

The AI Promotion Track (Intern to Manager)

You wouldn’t hire a 21-year-old intern on their first day and hand them the keys to the company bank account. You would give them low-risk tasks, review their work, and slowly build trust.

You must do the same with AI. In our framework, we break this down into three levels of engagement:

1. AI as the “Intern” (Passive)

At this level, trust is low and risk is high. The AI’s only job is cognitive engagement, such as summarising data or drafting internal notes. It operates with zero autonomy.

Example: You feed the AI 1,000 messy support emails and ask it to categorise the top 5 complaints. It saves you three hours of reading, but it doesn’t actually do anything with the data.

2. GenAI as the “Assistant” (Interactive)

As trust grows, you promote the AI. It now has limited autonomy and can generate new outputs, but still requires a human signature before execution.

Example: The AI reads the support emails, categorises them, and actually writes the draft responses. A human manager still has to click “Send.”

3. GenAI as the “Manager” (Proactive)

High trust, low perceived risk. The AI is now allowed to act autonomously based on a strict set of business rules that you control.

Example: (Like the system I broke down in my post on Automating vCFO Meeting Prep). The AI automatically pulls the Xero data, analyses it against the client’s historical context, flags cash flow anomalies, and drops a completed advisory brief into your folder while you sleep.

The Missing Link: Onboarding Your AI

The biggest mistake I see business owners make is opening ChatGPT or Gemini, typing a vague two-sentence prompt, and getting frustrated when the output is garbage.

If you hired a human employee, you wouldn’t just yell a task at them and walk away. You would onboard them. Our research proposes a specific, multi-step onboarding process for GenAI. Here are the critical steps you need to start using today:

• Provide Business Context: Don’t just ask a question. Feed the AI your company mission, tone of voice, and website URL. Tell it who it is acting as (“You are a strategic vCFO for an Australian transport company…”).
• Assign a Coach: Give the AI rules and protocols that must be followed. (e.g., “Never provide financial advice, only summarise the data provided.”)
• Define Responsibilities & Metrics: Explicitly state what a “good” output looks like. Upload templates of your previous work so it can mimic your formatting.
• Schedule Check-Ins: AI isn’t set-and-forget. You need to review the outputs and provide positive or corrective feedback so the system learns your preferences.

The Strategic vCFO Takeaway

Being a “lazy” accountant or founder doesn’t mean doing less work; it means being highly leveraged.

AI is not going to replace the strategic CFO. It lacks empathy, nuance, and the ability to understand human context. However, a CFO with AI will absolutely replace a CFO who is still manually exporting CSVs and color-coding spreadsheets.

By treating your AI like an employee, you can build an invisible workforce that does the heavy lifting for zero ongoing software subscription costs. This includes onboarding it properly, setting boundaries, and promoting it from Intern to Manager as it proves its accuracy.

Want to dive deeper into the academic theory? Read our full peer-reviewed paper here.

Are you tired of doing the repetitive grunt work your software should be doing for you? Let’s have a conversation about automating your finance function.

Let’s Talk