Save content
Have you found this content useful? Use the button above to save it to your profile.
The ChatGPT app icon on a mobile screen
iStock_it:Robert Way_AW_GPT

Ten ways accountants can use ChatGPT

by

Two speakers at the Festival of Accounting & Bookkeeping outlined how accountants can use ChatGPT to streamline or enhance their work, from internal agents and data analysis to client communications.

21st Mar 2024
Save content
Have you found this content useful? Use the button above to save it to your profile.

Since the launch of ChatGPT back in November 2022, generative artificial intelligence (GenAI) has been a red-hot marketing term, not least in the accounting world. However, accounting professionals looking to glean productivity gains or access hitherto unreachable data have repeatedly asked the same question: what GenAI tools are available now to help our day-to-day work?

Tucked away at the back of the hall at last week’s Festival of Accounting & Bookkeeping, speakers on the ‘Accounting for AI’ stage sought to answer this question for an enthusiastic audience of finance professionals. Discussion focussed mainly on the current market leader ChatGPT and several other alternatives from players such as Google's Gemini and Microsoft's Copilot.

Attendees’ experiences with the new wave of tools varied hugely, from accountants who’d built their own GenAI agents right through to those looking to get a bit more information before taking the plunge for the first time.

In separate talks, Billie Mcloughlin, technology lead for 2020 Innovation, and Heather Smith, Australian accountant, tech enthusiast and author of 'Xero for Dummies', outlined practical ways in which accountants and bookkeepers can use ChatGPT today to streamline or enhance their work, ten of which are listed below.

It’s worth noting that their sessions involved ChatGPT 4.0, the paid-for version of the tool, and both encouraged attendees not to put personal details such as full names, company numbers or UTRs, full addresses or telephone numbers.

Heather Smith

Interview preparation: Smith’s first example was to use ChatGPT to streamline or enhance preparation for job interviews you’re conducting as part of the hiring process. Starting generally, a user could enter a prompt such as: “You're an expert interviewer. Please give me five questions to interview an accounting intern.” The user can then drill down into more specific areas by adding to the prompts, for example, “we are particularly interested in their experience with inventory, please take account of this when putting together the questions”.

Changing email tone: For accountants more used to dealing with numbers, Smith stated that being empathic with clients or colleagues over email can take a lot of energy. As peculiar as it sounds, ChatGPT can help. She said that one trick she uses is to paste her emails into the system and ask it to rewrite them in a more friendly tone – or more direct if chasing an invoice. It can also help communicate with a more neurodiverse range of people, for example, for a client with ADHD, you could ask the system to write in clear, simple bullet points.

Research and Analysis: Smith uploaded a demo profit and loss (with client details such as company name and number removed) via the attachment tool in the paid version of ChatGPT and used the system to interrogate it. She prompted the system with the information that it was a cafe business in Birmingham and she was its accountant and asked it to review the statement and highlight five areas she should talk to the client about.

Excel training: The advent of any new accounting technology tends to bring out the harbingers of doom for Excel. However, for the most part, ChatGPT bucks this trend, with many accounting professionals combining their existing Excel knowledge with new GenAI tools to open up new areas of efficiency for the profession. In her presentation, Smith started by uploading an Excel workbook (without identifiable client data) and asking the system ‘how can I make this spreadsheet more efficient?’ It can also help identify the right formula for users’ needs, and while a Google search or YouTube video might do this job, ChatGPT’s speed of response and conversational interface can produce faster results.

Added value video service: Smith finished with an explanation of how ChatGPT and other GenAI tools could be used to offer an additional layer of customer service where an accounting user could upload a client’s financial statements and the system could create an automated script and video, talking the client through their finances. The user could then leverage tools such as HeyGen.com to translate this into different languages - citing her campaign for election to the ACCA Global Council as an example.

Billie Mcloughlin

Internal FAQ systems: Mcloughlin outlined how one firm she had recently spoken with had used ChatGPT to create a chatbot for new team members. The chatbot acts as an advanced FAQ system, helping new employees integrate faster into their firm and allowing for anonymous enquiries about office protocols such as where the coffee mugs are kept or who to go to with questions about certain subjects or clients. While she pointed out it was not capable of completing technical accounting work, in today’s remote working environment they could be a good way of helping to integrate and train new staff.

Work schedules: Another example Mclouglin flagged was using ChatGPT to manage capacity at audit firms. Starting with a spreadsheet containing details such as client company names, year ends, when the board meetings are, if it needs a stocktake, how many days it needs for juniors, seniors, managers etc, she combined this with staff within her fictional practice – how many days a week they work, what percentage of their work split is between audit and other jobs. She then fed this data into the GenAI system via the attachment feature and asked it to create a 12-month audit schedule to manage staff capacity throughout the year. This schedule can be easily changed and circumstances evolve and applied to other areas of the firm’s business.

Client meeting preparation: One use case Mcloughlin raised was to help with client meeting preparation. For example, before a meeting with a prospective client in the building trade, an accounting professional could ask ChatGPT for a list of issues that builders usually encounter when running their business. Or if the client has just started in business, a list of questions to run through to ensure they haven’t missed anything. She stressed that these lists are a good way to begin preparations

Content creation for marketing: While generating marketing content, including blog posts and social media updates, is one of the most popular ways accounting professionals are using GenAI tools, Mcloughlin urged attendees to use ChatGPT as a starting point rather than the finished article. Instead, she recommended using it to generate ideas for specific client content such as startups, builders or e-commerce businesses, and layer on the human touch afterwards.

Forecasting and analysis: While Mcloughlin was keen to stress that any personal information should be removed before entering it into a system like ChatGPT, she outlined how the tool can be used to analyse and drill down into client data. By pasting data into the system or attaching a spreadsheet, users could ask questions such as "what happens if my client’s turnover goes up by 5% but their purchases go up by 10%?"

Replies (2)

Please login or register to join the discussion.

Stepurhan
By stepurhan
22nd Mar 2024 12:00

Any way to share the entries and results?

Some of these don't sound any different from doing some basic online research, just using an AI bot instead of a search engine. It would be interesting to see the actual examples to demonstrate if it is actually more than that. Also there have been much publicised cases of AI producing misleading or incorrect results, so proof how it works in these cases would also be of interest.

Thanks (3)
Replying to stepurhan:
avatar
By johnjenkins
03rd Apr 2024 16:15

Who is to say that AI can interpret information better than we can?

Thanks (0)