In this blog, I want to discuss the application of Chat GPT and other AI tools by financial advisers, accountants, and mortgage brokers to enhance practice efficiency across various advice and business-related tasks.
I must admit to having steered clear of this topic for advice businesses until now.
I try to avoid jumping on bandwagons. Sometimes it feels like the latest “shiny” new thing comes out in the financial services industry, and a wave of consultants, marketers, and salespeople appear madly spruiking the best thing since sliced cheese.
History is littered with the “next big thing” that proved to be a waste of time, particularly for fintech.
Tools – chatbots spring to mind as a prime example – that ultimately prove to be ineffective in delivering the utility promised, distracting practice owners away from things that actually deliver like generating leads, servicing clients and building a systemised practice business mode.
That’s why I’ve taken my time to explore first. However, after a few months of experimenting with Chat GPT and its counterparts, I’ve found several applications specifically for advice, accounting and broker that I think are worth sharing.
Let me begin by outlining what I’ve learned about using Chat GPT— let’s call them the “musts” and the “don’ts.“
The Number One Tip for Practice Owners When Using Chat-GPT
The most important lesson I’ve learned is; the quality of your output is determined by the quality of your input.
In short, if you provide vague instructions like “Write me a blog on financial planning,” or “Give me ten reasons to get a good accountant”, I can almost guarantee that the output you receive will be equally generic and vague.
When you invest a little time in clearly defining what you want, you can achieve some impressive results.
Let me provide you with a live example. When I first started using Chat GPT, my input command prompt might have looked something like this:
“Copy edit the following transcript for me.”
Now, after understanding what the tool can and can’t do, that prompt looks more like this;
“Could you please proofread this for me, make changes to align with Australian English grammatical standards, and ensure it’s written in a friendly, informal, and personal tone? Write as if you are me, having a one-on-one conversation with the reader, who is a practice owner who is either an accountant, financial adviser, or mortgage broker. Additionally, suggest formatting improvements for better readability and recommend section headings to break up the text”
Much better result.
(I’d also recommend avoiding asking it to imitate an Australian voice unless you want a copy filled with ‘G’day mate’s and ‘shove another shrimp on the baar-bee.’
Don’t believe everything it says
Another crucial point to understand is that while Chat GPT may talk a good game, some of its output can be not only inaccurate but downright false.
This ranges from simple errors – like when I asked it to summarise webinar transcripts into chapters, and it skipped over chapter 2 completely until reminded that it had missed one – to more extreme instances where the tool invents complete fantasies, such as Brazil winning the 2018 World Cup.
No advice professional wants to be caught out saying things that aren’t actually correct.
If you plan to rely on Chat GPT as a source of truths, verifying and fact-checking the information you receive is vital. It’s far from infallible.
What it excels at
If I had to describe what Chat GPT can do for practices, the simplest way to put it is that it removes 80% of the “grunt” work that can flesh out many tasks that often bog advice professionals down.
If you’re an accountant, it will save you a bunch of time. Same deal with advisers. Mortgage brokers too.
I know there will be different opinions – and I’m just sharing my personal experience and how I extrapolate its utility for my clients – but when I think about the scope of many tasks, you’ve got:
- The first 10%, where you work out what you need to do and come up with idea. Deciding what blog posts you’ll write based on your knowledge of your clients and the subject matter (rather than relying on Chat GPT to generate titles and then just writing what it tells you to) is an example of this.
- The middle 80%, which could be all that laborious copy editing, grammar checking, formatting and a host of other tasks that you might be too busy, distracted, or (as is the case with me) lack the attention to detail to do consistently well. This is what I mean by the “grunt” work.
- The final 10%, which might involve reviewing and refining content to ensure it sounds genuine and aligns with your personal style, fact-checking, or polishing out your final product to a state where you’re comfortable putting your name and brand on it.
From my perspective, Chat GPT is a tool for the middle stage, and I can’t see myself suggesting any practice owner completely rely on it for initial creation or final validation and checking processes.
Practical uses
With that stated, let me outline some specific use cases I’ve discovered for Chat GPT in practice operations:
- Writing how-to guides: Chat GPT can provide detailed step-by-step instructions of the kind you’d put into your operations manual. For example, I asked it to create a guide on inputting client information into Xplan, and it did a commendable job. I’ve also used it to give me guidance on landing page structure, conducting keyword research and a host of other things I wanted others to do for me, but didn’t have the time to write out every single step involved myself, as well as clean up process guides that I hadn’t got around to finishing. If you’re trying to onboard new admin staff, or just remove some of that expert knowledge from your head faster, this will help.
- Summarizing meetings: As someone who values dictation, I recognise that raw meeting output filled with grammatical errors or lacking structure can be hard to work with and time-consuming to fix. Chat GPT excels at copy-editing and organizing these transcriptions into user-friendly formats with bullet points and headings, summarising the key points, action items, and decisions made during client meetings, to provide the start of a decent meeting file note.
- Transcription. Alternatively, if you need a long-form version that is actually readable, it can do it very quickly.
- Writing emails: It’s also a quick way to obtain first drafts of emails for various purposes that can be tweaked to suit your specific situation, or to have an existing one re-written in a different way.
- Summarizing long-form content: One of my favourite prompts – TLDR, which stands for “Too Long Don’t Read” – can be applied to text to provide a version that condenses the content into the smallest possible form it can take for quick consumption. Even if not using that prompt, it’s still possible to have lengthy reports or research papers summarised into executive summaries so you can grasp the main points quickly. I haven’t applied it to any ASIC publications as yet…
- SEO/Keyword optimization: This is work that can eat up hours the old way. Assuming you’ve done the work upfront to identify the keywords you want to target, Chat-GPT can optimize website content for search engines, produce meta descriptions and tags and even suggest relevant keywords to help drive more organic traffic.
- Explaining concepts in simple terms: As an expert, explaining complex topics in simpler language can be challenging. Chat GPT can give apply a layperson’s perspective and make it easier for you to go back to the ground floor and explain intricate topics in a more accessible and understandable manner.
- “Quick and dirty” research: While Chat GPT’s access to internet-based information is limited to pre-2021 data, it can still assist in streamlining your research. In one case, it was able to help identify 30 relevant white papers on a specific topic I should check out, drawing from various sources worldwide. Whilst not all of them were still available online or completely relevant, it still turned over stones I wouldn’t have thought to look under.
- Coding: Tech professionals have been some of the most vocal about Chat-GPT’s utility, as it’s pretty good at doing all sorts of codey things. At a less technical level though, it can also help with simpler tasks like crafting tricky Excel formulas or compound calculations that most of us forgot not long after we took out last maths class.
These are just a few examples. Undoubtedly, more will become apparent as this rapid leap forward takes further shape.
While I wouldn’t recommend making a tool like Chat-GPT the epicentre of your advice firm or replacing your entire admin & paraplanning team quite yet, the early signs are it can help expedite tasks that take us advice-focused individuals considerably longer or we’re just not wired to do well.
Now all that’s left to solve is the energy issue, because we’re a long way from a Chat-GPT that can do what we can do on nothing more than a cup of coffee and a slice of toast.
What’s your experience been? What have you been able to do that hasn’t been mentioned? Keen to hear more about the practical uses of this tool – email me at help@audere.com.au and let me know.
