005 – ChatGPT and the Finance Professional
(and more Microsoft Build)
Hey Friends!
I’m currently sitting in a hotel room in Las Vegas, getting ready to present today on how AI will impact the financial planning industry. It’s my first big conference and so I’m trying to figure out if I should be a little nervous, extra-nervous or ALL THE NERVOUS.
…honestly, I can’t help but be excited to share…
First of all, to address the question some of you might have – “Did ChatGPT write your presentation?”
Nope. Actually, I structured my presentation to have as few words as possible. When you have technology like this, do you really need words?
So, if you don’t mind, I’d love to dig into AI and the financial planning professional and pull in some Microsoft Build videos that I think will blow your mind (that I didn’t get to share at the conference but I want everyone to see)
Financial Planners have a different practice than Tax Professionals who have a different practice than Accounting Professionals…
…but they all have very similar problems…
1) Email as the main source of communication for EVERYTHING
2) Administrative tasks (necessary but repetitive and boring)
3) Different pieces of information “trapped” in different software applications
4) Increased complexity
Okay, so how would AI solve this? ChatGPT is great but its isolated with the exception of a few plugins and BingAI isn’t isolated but its sometimes a little “cranky”.
What I saw at Microsoft Build would address #1 and #4 but also give us the tools to address #2 and #3.
Addressing #1 – Taming Email Overload
I currently have one or two (or thousands) of emails sitting in my inbox. Fortunately, I have a third-party software that captures those emails but here’s the challenge:
If its not a meeting, its an email and those email chains get long and branched. So, if I need to go back and summarize, it can get complicated.
A client might need an answer to question and I *know* they mentioned something about it six months ago – I just need to find that email…. Or was it seven months ago?
A project goes over a longer period and have multiple email chains – collecting and combining those into a project summary is a big task and often gets put off because its also a somewhat exhausting tasks.
I shared this last week , but at Microsoft Build – a woman (Arachana Saseetharan) had CoPilot pull and summarize emails from Outlook *while she was in Microsoft Teams*. And then she also had CoPilot pull in documents Viva Sales. And then she had CoPilot craft a response email. Her part of the presentation starts around minute 10 – you can find it here. Meeting prep that would have taken at least 30 minutes, possibly an hour was done in 1-2 minutes.
How much easier would it be to advise clients if preparing for our discussions and pulling together their information was as easy as asking CoPilot to go collect and summarize everything for us?
Addressing #4 – Cooling the Complexity in Our Practices
Has anyone else’s tax law changes gone bonkers in the last six years (2017 was six years ago…) or is it just me. 🙋🏻♀️
One of the things I often hear from financial planners is *how much* of their practice includes tax planning. And let’s be honest, good tax planning is an art that takes years to master. There are sooooooo many strings - if you pull one, it pulls three more.
…and then Congress goes and upends the laws. Not just once, not just twice, but every year for the last six years…
AI will help us with this as well!
I know what you’re thinking – “I asked ChatGPT about Roth 401(k)s and it gave me a bad answer!”
I imagine it did. It’s not a tax-technical mastermind in the least.
Think about how ChatGPT was trained – it was trained on the *visible* internet. And only up to September, 2021. And then think about how many well-written articles there are about Roth 401(k)s vs how many well-written articles there are about 401(k)s in general. And then think about all of the terribly-written articles that ALSO trained the model.
The model is taking your input and statistically determining which rabbit hole would be the best rabbit hole to go down. I’m sure when it sees “401(k)”, its like “Oooh buddy, we’re going to talk about pre-tax retirement contributions today!”
But, what if we didn’t have to rely on ChatGPT (or the GPT4 engine, really) to know everything off the top of its head? What if we could bring in tools that it could use for researching a response?
And that’s where plugins come in….
Plugins bring the third-party tools we use now INTO our software. So, instead of having to stop what we’re doing, go look something up, scrunch brow while trying to understand the topic, take a bunch of notes to make sure we’re understanding it and then craft an email or a memo…. Just send the robot to do it!
In that same video (here), around minute 35:00, a fictional attorney takes a master agreement template, populates it with the prospect data from an earlier call and then PULLS in Thompson Reuters tools to tweak and help review the contract. Without ever having to leave Word.
How would that help us with our tax law changes? Well, Thompson Reuters is the publisher of Checkpoint, PPC Deskbooks and Quickfinder – all tax and accounting research products in different formats. They also publish annual tax updates and have Continuing Professional Education on all sorts of topics.
So, if your ChatGPT (in this case CoPilot) could access all of the annual tax updates and ALSO have access to your client profiles…? Can you just imagine sitting in Word and having CoPilot create a Tax Planning Memo for a client in 30 seconds based on everything it sees about that client?
I’m going to share next week about solutions for #2 and #3 above. Of all of the applications that you use in your practice, which one would you be most excited about bringing into your workspace?
Happy Chatting!