The Art of Prompt Engineering for Accountants: The Latest Skill You Didn’t Know You Needed

With the rise of artificial intelligence tools such as ChatGPT, accounting professionals need to add a new skill to their toolbox: prompt engineering. But don’t worry, it doesn’t have much to do with engineering, and the learning curve is quite achievable for everyone. The best part is that the payoff can be game-changing.

Prompt engineering delves into a world where numbers meet language. While it might sound like a far cry from balance sheets, ledgers, and tax forms, it’s an easy concept to understand. Let’s dig into the basics and benefits of prompt engineering for accounting professionals.

Prompt Engineering 101: A New Era in Accounting

Prompt engineering is a process that fine-tunes how we ask questions or provide instructions to AI systems. It involves crafting precise queries to help these machines deliver the exact information we seek. Let’s use a simple analogy to break it down.

Imagine you’re trying to instruct a new intern to calculate gross profit. If you simply say, “Calculate profit,” the intern might get confused – do you mean gross profit, net profit, operating profit, or some other variant? However, if you specify, “Calculate gross profit by subtracting the cost of goods sold from total revenue,” the intern will now have clear guidance and likely deliver what you asked for. Prompt engineering works in much the same way when communicating with AI.

Using Prompt Engineering in Accounting

Everywhere you turn, there are applications in accounting that you can enlist ChatGPT and other large language models to help you save hours each day. From writing just about anything you need (including parts of this article!) to generating financial reports, cleaning up QuickBooks transactions, and providing tax saving ideas, ChatGPT can do it all with the right prompt. You have plenty of opportunities to practice building your prompt engineering skills simply by looking at your to-do list.

Components of a Great Prompt for ChatGPT

When you first start to use ChatGPT, Bard, or Bing AI, your prompts (or input commands) are likely to be quite shallow. We might mimic getting to know a person where we start by saying hello and talking about the weather. With AI models, we need to dive in deep right away. We need to give it a specific, detailed command in order to get a great output.

Some things to include in every command:

  1. Goal: Let it know what you want as output. Use an action verb to start with.
  2. Tone: How do you want the model to reply? Include adjectives such as engaging, professional, playful, or stern.
  3. Length: If you’re asking it to write something, let it know how long it should be. Specify word count.
  4. Define the audience the answer will be for. Give it background information as much as possible.
  5. Format: You can ask it to create text, tables, programming code, numbered lists, csv formats, and others that are ready to copy and paste.

There are hundreds of prompt techniques that you can employ to get the best answer.  While we can’t cover them in this article, we do have a course available that can speed your prompt engineering learning curve.

Looking Ahead with AI

As more accountants embrace prompt engineering, the profession stands to gain a powerful ally in AI. Harnessing the full potential of this technology can streamline operations, improve accuracy, and allow more time for strategic tasks that require a human touch.

New Course on Prompt Engineering

Want to see ChatGPT in action along with dozens of ideas on how to save time in your accounting firm?  We have a brand new CPE course right here: