← AICPA Town Hall
AICPA Town Hall
Tech trends, talent pipeline, and BOI
01:00:15
Topics include: DC and profession update Talent pipeline AI trends and impacts on critical infrastructure Evolving technology trends across the profession Key technical...
Loading summary
Transcript86 lines
- [00:01]
A
Welcome to the AICPA Town Hall Series, your resource for the latest news and updates on pressing issues facing the accounting profession.
- [00:12]
B
Good morning and welcome to the AICPA Town Hall. We're coming to you live from Digital CPA in Denver. Lisa, it's been an amazing conference. So much to discuss. We probably have one of the largest audiences ever. We've got over 1000 with us here in the room and we probably got close to 10,000 online. So let's just kick it off, Lisa, and talk a little bit about some of the key themes. I mean, we've had countless discussions on the power of AI and how it's really, you know, transforming so many of the solutions that are driving the transformation of the different practice areas. Some great keynotes and we're actually going to have a good discussion today on the town hall about where AI is and where it's going. And it's been all about the evolution of firm business models, something I know, Lisa, you think a lot about. And then digging in to the transformation of the different practice areas. One big area that we've been thinking a lot about is the CAST area.
- [01:14]
C
And I was lucky enough to get to participate in a session yesterday that took a dive into the results of the CAST survey. They're incredible. We reported the survey participants reported 17% year over year growth in CAST services. 17%. That far outpaces the average growth that we're seeing of around 10 to 11% for overall firm services. And then a remarkable 29% growth in net client fees per professional, which is a standard benchmark that a lot of firms use to look at productivity and effectiveness. So we're going to be giving you a deep dive webinar into those results. And also you can go ahead and download that survey and start digging in and analyzing your CAAS service line against these benchmarks for results. But overall, great progress in continuing to mature this role in the profession and to really see remarkable growth.
- [02:15]
B
Yeah, I mean, we kicked off digital CPA 12 years ago and we started with great focus around client accounting services, called client advisory services by many now due to the cloud computing movement. It really enables firms to offer this new service in a very strategic way. And this benchmark survey has so many insights. It's over 40 pages, lots of lots of data. You highlighted some of the powerful numbers. Here's another one. For the firms that have over 50% of their revenue focused on vertical areas niches, Lisa, they're growing at over 20%. So a lot of good insights for you to leverage and the other big one other big topic related to the client accounting services area is this very, very important Update related to SARS21. I've received a lot of positive feedback about this. We've talked about this on previous town halls, but we just want to cover it quickly. Again, here are the key points and really here's the down arrow. When financial statements are prepared as a byproduct of client advisory service engagements or your virtual CFO service or whatever you may call it, accountants are not required to apply SARS 21 Section 70, it's your option to have these engagements be performed in accordance with the consulting standard. And Lisa, we kicked off on Monday with our Client Advisory Services Advisory Council discussing this and I've pretty much I've heard unanimously that everyone is going to be moving their CAST practice under the consulting standard. But this is still an exposure draft and it will be looking to get this finalized in 2025.
- [04:04]
C
And I've gotten the same feedback that this that CAST leaders see this as a huge step forward in making the CAS services more efficient and effective while maintaining that focus on the public interest protection. If you are in the CAS line, we'd really love your feedback on this. And you'll see a link in the slides as a reminder. If you're new to town hall, download those slides because we pump these slides full of links that take you to the valuable information that you're going to be looking for. So take a look at that and evaluate the standard as it applies to your practice. And we'd love your feedback by December 20th.
- [04:42]
B
So we talked about many other things here@digital CPA over the past couple of days. A lot of talk about audit transformation as well as what's happening in the tax area. So now let's just kind of review what we'll be getting into. On today's town hall. We're going to kick things off with a DC and Profession update with Rachel Dressen. Then we're going to get into an AI Trends and opportunity discussion with Pascal Fannette and Avni Desai, and I'll be introducing them shortly. And then we've got our vice chair here, Lexi Kessler, who's going to give us an update on the Talent Pipeline initiative. And then we've got, Lisa, you're going to be closing out with the technical updates. So let's get right into the DC and Profession update. Rachel, I understand Congress is back in session.
- [05:27]
A
They are.
- [05:27]
D
They're back in session. And the big question is what are they going to do about government Funding. I know we've talked about this a lot on the town hall in the past. The government funding deadline is next Friday, December 20th. So they only have a little over a week to determine what they're going to do to continue to fund the government. It looks right now like they will pass a continuing resolution or, or a CR which would continue current spending levels until the date they determine would be the expiration. They are talking right now like it will be sometime in March. We understand, as we've said before to the town hall that, you know, this could have some concerns if there is a potential government shutdown come sometime in March, whenever the deadline might be, if they aren't able to come to an agreement which would fall right during tax season. So we are very aware of that. But right now that's kind of the timing that they are looking at. We'll know more about this actually later today. The speaker of the House has said that he plans to come out with some more of the parameters around government funding. So we should be learning more in the next few hours and couple of days.
- [06:48]
B
And also I understand some developments related to tax legislation in 2025.
- [06:52]
D
Absolutely. So everyone wants to know what's going to happen with tax reform. We all know that next year will be a big tax reform year. And so the biggest news that has come out in the past couple of weeks is the idea of having two budget resolutions. So the thought is, is the way that they will pass tax reform is through a process, through budget resolution, meaning that it would not need the 60 vote threshold in the Senate. It would only need a majority vote. The idea originally was that they would do one budget resolution, but now it looks like they might do two budget resolutions and the first one would be more focused on things like border funding and not include tax. And that first resolution, they are talking about trying to get that one passed immediately within the first 30 to 100 days once President Trump is inaugurated. But then the second resolution would be the one to include tax reform. So that means it would be punted farther back in the process than had originally been thought. The original thought was maybe they would try to do it in the first 100 days. So if they do this, and it's not certain, but if they do this approach of two budget resolutions, then tax reform is something that we will likely see later next year rather than earlier next year.
- [08:25]
B
We'll be discussing this quite a bit in future town halls. So we've got a number of very important cabinet nominations that have occurred.
- [08:34]
D
Yeah. So within the past Two weeks since the last town hall, President Elect Trump has announced that he plans to have his treasury secretary be Scott Besant. Scott Besant is a hedge fund manager who is the founder of Key Square Group. He is not very familiar in D.C. because he doesn't have a government background. So a lot of public policy issues, those in D.C. are not as familiar with him because of that. But he has said that one of his top priorities will be to extend some of the TCGA provisions or to make some of those tax cuts permanent. He's also talked about tariffs and using tariffs as a sanction tool, but he has mentioned doing it in a more gradual approach. He's also raised concerns with the budget and talked about, you know, some spending cuts. So those are, you know, some of the things that he has discussed, you know, should he be confirmed by the Senate? He's having meetings with senators right now in the process. The Senate confirms. The House is not involved in this process. So he's having those meetings and all things right now. It looks like this should easily go through unless something comes up in his confirmation hearing, which will take place within the Senate Finance committee, likely before January 20th. So typically with some of the Cabinet positions, they try to do the hearings before the inauguration, and then even the day of the inauguration, the Senate will go in and vote on some of those. And Secretary Besant, or soon to be Secretary Pessant could be one of those nominations. And so we'll see what happens there.
- [10:34]
B
So we got a couple other ones we can click through here. But Paul Atkins, he's someone who does have background in the government.
- [10:41]
D
Yes. So we're very familiar with Paul Atkins. He is a former commissioner at the sec. He has a long history. He's testified before Congress over the past number of years. And it's interesting. So the two Republican commissioners at the sec, Hester Peirce and Mark Yoeda, they were actually staffers for Paul Atkins when he was at the sec. So that kind of gives you an idea of his ideology. We expect, should he be confirmed, that he will take a look at some of the rules the SEC has come out with in the past four years and maybe revise them or he may try to scrap them entirely. And so we could see kind of a change in some of the tone at the SEC where under Chairman Gensler, he has talked about some of the enforcement and some of the fines have been record breaking. We could see a bit of a change to that under a Atkins sec.
- [11:46]
B
And finally we got the IRS commissioner nomination.
- [11:49]
D
Yes. So this is the Newest one. So former Congressman Billy Long from Missouri has been nominated to be the IRS commissioner. And so he will also have to go through the confirmation process. He was a congressman from 2011 until 2023, when he ran unsuccessfully for Senate. Since then, he has been a certified tax advisor and business advisor, focusing on ERC work. So he'll also go through a Finance Committee hearing and then the confirmation vote.
- [12:28]
B
Okay, plenty to discuss as we move into 2025. Another big announcement that came out earlier, earlier last week.
- [12:36]
C
Talk about plenty to discuss. Yes. If you have been a regular on town halls, you know that we've spent quite a bit talk quite a bit of time talking about beneficial ownership information reporting, which I'll call boi. This is an anti money laundering provision that is part of what's called the Corporate Transparency act, cta. And we had some big news last week when a district court issued an opinion, a decision that basically said that CTA is likely unconstitutional because it is outside of Congress's power to enforce. To say that these are the requirements. Excuse me. It is so dry here. I apologize.
- [13:19]
B
A mile high.
- [13:19]
C
Yeah, we're a mile high. It also includes, and this is really important, it includes an injunction that applies nationally that prohibits the enforcement of CTA in the BOI reporting rule by the Financial crimes enforcement network, FinCEN. So under that injunction, FinCEN cannot enforce BOI filing requirements while the case is pending. While the case is pending is important because there are several other decisions across the country that say that BOI is constitutional and that we should be filing BOI reports. So what to do? The Department of Justice filed an appeal on behalf of FinCEN on December 5th, and we expected that they would. But the important thing for all of us to understand is that either the Fifth Circuit Court Court or the US Supreme Court could stay that district court opinion. That puts the injunction on filing. And so it's all very fluid right now.
- [14:27]
B
And so town hall, there aren't questions coming in. Lisa, so what should we do?
- [14:31]
C
Well, let's dig in a little bit more on what FinCEN's response is, and then that'll kind of guide us as to what, what to do next. It's important to note that the decision out of the Texas case, one of the largest, shoot, one of the largest people that were involved in the suit is the National Federation of independent businesses, NFIB, and they have 300,000 members nationwide. So that's why the court decided that this should be a nationwide injunction. So that's all very confusing and very fluid. Then you've got FinCEN coming back and responding, saying that because the injunction is nationwide and is in place, you are not required to file a BOI report with FinCEN right now. You're not subject to liability for failing to file your boi, but if you want to, you can continue to submit BOI reports. So that leaves our CPAs and our members in industry who are filing these BOI reports with a huge question mark, what to do. Our advocacy team has been in touch with FinCEN, the AICPA Member Insurance Program. Our friends at AON and CNA have all worked together to produce a risk alert that you'll find in your resources. And so I wanted to pull some of the best practices out of that risk alert to give you some guidance on what to do next. Basically, if you've been talking to your clients about filing these reports on their behalf, then make sure that you're informing them about the injunction and about FinCEN's decision to appeal so that they fully understand what their options are. The client needs to decide, and I'm going to repeat that. The client needs to decide if they would like to continue with the BOI submission process. Make them tell you that in writing. Okay. That's not something that you want to just.
- [16:28]
B
And you can give advice, though.
- [16:29]
C
Yeah, give advice.
- [16:30]
B
Give advice here.
- [16:31]
C
And here's the advice. The advice is this, this could change quickly.
- [16:34]
B
Right.
- [16:35]
C
And so at least consider going ahead and collecting the information so that if a decision comes out that reverses the Texas opinion, or if this could go all the way to the Supreme Court. So if that decision comes out, then you're ready to file. We fully expect that. We don't know. We don't have a guarantee, but we expect that FinCEN would give some time after, if, after January 1st this gets reversed, then they'll give us some time to get those filings in. But be ready, as you said.
- [17:05]
B
But nfib, there's a lot of people that are going to keep eyes on this, pressure on this, make it reasonable, make it practical.
- [17:13]
C
Yeah, agreed.
- [17:14]
B
Potentially with a delay, if this is then overturned again.
- [17:19]
C
Yeah. And we continue to advocate for the delay because this decision has created even more confusion in the, in the atmosphere. So I think, Rachel, if you want to tell us again what we're doing around advocating for this delay, I think that would help.
- [17:32]
D
Yeah. So what we're doing is the AICPA has sent a letter to Congress asking for a delay and we would like to see this include. I mentioned the CR earlier, which is a government funding bill to Include it in the cr, which is the last must pass bill in Congress. We've also been working with the state societies and together with the state societies we sent a letter to FinCEN asking for a delay. For that CR. There needs to be a four corners agreement, which means the four corners are the Republican and Democrat leaders of the House and the Senate to come to an agreement to include it in the government funding bill.
- [18:12]
B
Well, thank you, Lisa. Thank you Rachel. We going back, we'll look at the questions. We can come back to this during the technical update in Open Forum session as well. So now want to move to this discussion, AI trends and opportunities. We've got a veteran of digital CPA with us, Pascal Fannette. He's a veteran of the town hall. He is one of the leading innovation thinkers in America in the world. He counsels many, many companies and he's been just a great partner for us doing a lot of primary research. So Pascal, it's a pleasure to have you and looking forward to you're going to have a mini keynote right now in the town hall. So take us away.
- [18:56]
A
Yeah, thank you so much for having me back. And I just want to acknowledge a fact because we actually did a keynote at DCPA last year. So just 12 months ago and we had a room very similar to this, full of people, about 1,000 people back then and you might remember we launched something called the CPA.com Generative AI Toolkit. Remarkably, this resource held up to this day. So if you haven't downloaded it, it's still time to download it. But here's the interesting thing. We did a workshop. If you were in the audience you might remember this. But we did a workshop and we asked the participants, the thousand people in the room, who has used ChatGPT or any of the other large language models and at the time only about 20% of you raised their hand and said yes, I have used it outside of dabbling with it, as in I've opened it, I've tried to do any work related tasks in it. I am pretty sure and we will not do this today, but I'm pretty sure if I were to ask you today who has here used ChatGPT, Gemini, whatever system you want to use. I think we're probably at 100%. So a lot has happened in this just one year and I want to very briefly touch upon where are we today and then also talk a little bit about what do you or should you consider as we are going into 2025. So again it's been a Blistering pace. We're still moving forward incredibly fast when it comes to these large language models. Generative AI you might have seen. So the latest version of OpenAI's model is called GPT4O. This just came out of there, what they called preview version. You now can pay $200 to get access to a fairly powerful new model, GPT4O. But it's not just OpenAI. We have models from companies like Anthropic, which is Claude. Google is in the arena with Gemini. You can run Meta's latest model, Llama, which is an open source model. Their latest model, which only came out what, like three days ago maybe or four days ago, you can now run on your laptop. So if you have a somewhat beefy laptop, you can actually run a GPT4 class model completely on your laptop. So no data ever goes into the cloud. So really, really exciting things happening as well as just as a quick shout out to our friends in France, Mistral seems to be the only European contestant in this area. And then of course, Elon Musk's Grok is there and many other systems, a couple Chinese systems as well. What's remarkable about these systems, and this is an interesting industry trend just for you to observe these systems, all of these large language models are essentially the same. So the performance benchmarks, they differ by literally like 1% or 2%. So it really doesn't matter which of these models you're using. Some of the models are a little bit better in specific tasks. For example, Claude is very good at writing text. It just has a very nice way to express itself. Whereas other models are a little bit better with data, for example. But by and at large, it really doesn't matter which model you're using, at least for now, these models are incredibly good. We're also seeing still the trend that these models are getting still bigger. So one of the reasons why these models are getting so good is they ingest more data. And the more data you ingest, the better the models are getting. This is a trend which still continues. It's somewhat slowing because when you look at the frontier models, they have by now ingested pretty much all of the data which is available on the public Internet. So these companies are now tapping into more private data pools. The other big trend we're seeing is we're seeing models getting more specific, Meaning we take models and we train them on vertical data sets to make them very good at a particular industry or vertical. So, for example, DeepMind, which is Google's research organization, has trained models on medical data, protein folding, particularly to do drug discovery. And then the other thing we're seeing is we see these models getting larger memory windows, which means you can upload more data to these models. We are now able to upload to some of these models literally dozens of volumes of Harry Potter. So not just like a single volume of Harry Potter, you can upload dozens of volumes of Harry Potter and ask questions about these, the data contained in there. And that's a really exciting and very important trend because it allows us to do really crazy stuff with these models. Other thing I want to point out is if you have used Google in the past or Bing or whatever your search engine of choice is, those days might be over. We're moving very quickly from a world where you asked a question of Google, for example, and they respond with essentially 17.9 million links and you have to figure out what the actual information is contained in it. We're now moving into a world where using large language models to sit on top of this data to provide us answers. And that is really exciting because it makes information retrieval incredibly easy. A good example for this is in the public and broader space, a company called Perplexity AI. Highly recommend, just try it out. It's free to use. I have no direct connection to this company, so this is not an advertisement, but it allows you to do text research. For example, speaking of text research and talking about more vertical models, one of the vendors who is here, a company called bluejay, does a specialized. Has a specialized large language model which goes very deep into the specific data pools you need as a tax professional. So I also highly recommend checking those out. So again, information retrieval becomes incredibly easy. I want to give you a quick, like, hint at where this crazy world is going. There's an experiment which came out from Google called NotebookLM. It's completely free to use. It's an experiment. Please also note, this is on the public Internet. So any data you upload to these models will most likely be used to train the model. So be careful with any private data. But NotebookLM allows you to upload documents to a large language model, NotebookLM. And then it summarizes this, which is something you've seen large language models do fairly, consistently, fairly well. It allows you to ask questions about it. But it has a very neat little and interesting feature. And that feature shows you a little bit of where this crazy world of AI is going. So as a little test, I asked Casey to give me a document which she was particularly interested in. She gave me this proposed law Here with the absolutely. As a non accountant, the absolutely riveting title of certain partnership related party basis adjustment transactions as transactions of interest. I have to say I have no idea what that even means. But I took this document, so I took the whole webpage, essentially just the URL, and uploaded it into NotebookLM. So I basically said create a new NotebookLM, you know, file with this information. And then I asked NotebookLM not just to summarize this for me and allow me to ask questions about it, but they asked it. And this is a unique feature to create a podcast out of it. So what NotebookLM does is, and you will see this and hear this in a second, it creates a podcast with two podcast hosts, one male and one female. Make no mistake, these people do not exist. They have no idea what they're talking about, and yet this is what they came up with.
- [26:49]
B
Hey everyone, and welcome back for another deep dive. Today we're going to be wading into the always fascinating world of partnership taxation.
- [26:58]
E
Right?
- [26:59]
B
And let me tell you, the IRS has, they've really cooked up something that has the accounting world buzzing.
- [27:06]
C
Yeah, that's right.
- [27:07]
F
Essentially what the IRS is doing is.
- [27:09]
A
So you get the idea. And this might be, by the way, Eric, I just want to out point, point this out. This might be the end of AICPA Town hall. There might very well be a future version of all of us.
- [27:23]
B
You're getting complaints already here.
- [27:24]
C
Perfect.
- [27:25]
B
There you go.
- [27:26]
A
So this was just to show you where this crazy world is going. Now, one thing I want to make sure that you always remember large language models. And this is really, really important to understand. So you understand what the edge cases are. They do not answer questions or chat with you. All they do is they predict what the next word in a sentence is. So even this crazy example you just heard with the two podcast hosts, which do not exist, talking about, who do not have an opinion on anything, yet talk about very convincingly about something they don't know anything about, is purely a outburst of very, very advanced statistics. It is wild that this even works, but it does. Ethan Mollick, who's a professor over at Wharton Business School, one of my favorite pundits on this topic of artificial intelligence. He talks about this idea of, and this is important to understand the jagged frontier. So he points out that when you look at an AI system, there are certain things it can do really, really well, and there's certain things which are related and you would expect it to do really well and it can't do them at all, or it hallucinates or it can't do them in a particularly good style. And this makes it really hard for us as practitioners because it's not a binary yes, you can do this with an LLM or no, the LLM can't do it. So it is on us, and this is my invitation for you. It is on us to figure out where does this frontier live at any given moment in time. Which makes it hard for us as practitioners because we have to be out there and trying these things out and testing them. And I know that we'll talk later a little bit about how do you actually do this in the practical world? Simon Willinson, who's a programmer, very famous in the Python world, pointed out that they're actually these LLMs, despite the fact that they look easy to use because they have this text box and this friendly manner to them, they're actually really difficult to use because you really need, need to understand how they work. So my invitation for you is spend, continue to spend time investing into trying things out, playing with these products, creating proof of concepts to understand. Where is this frontier? Where are they actually helpful for you and where are they not helpful for you? Briefly, what are we doing with these systems today? This is a report which came out from Menlo Ventures, which is a big venture firm in Silicon Valley. This report came out about 10 days ago and they looked at enterprise users and asked enterprise users, what are you using LLMs actually for? And you can see the predominant use case, not surprisingly, this being a Silicon Valley firm, is code generation. So actually using LLMs to generate computer code or helping you generate computer code. But when you look at all the other categories, pretty much all of them are in the still in the text arena. So anything from support chatbots to helping me write emails, to summarize text, et cetera. So LLM still to this day, largely really good at anything which has to do with text. The stuff which is non text, particularly workflow automation. We'll get there in a second. Still on the frontier, still not quite there yet, but that's something we're gonna see in the next year. So very briefly, what's happening next year, huge topic you cannot avoid. It is something called agentic AI or AI agents. So the idea of can I take an AI and actually have it do things for me? So not just respond to a question and give me an answer and then I as a human now have to do something. Say, for example, you know, what is the cheapest way for me to fly from Denver to, you know, Orlando and the AI can go out there on the Internet and find the data. In the future, it will also be able to book that flight for me. So that's a huge area of interest for the industry. Tread cautiously. A lot of this stuff can be done and it's really good. And some of that stuff is very much in the world of hype. A lot of AI companies currently, and this is my personal opinion, need to just justify their eye watering valuations. And if your company is valued at billions, tens of billions or hundreds of billions of dollars, you better make sure that the world still believes that your productivity gains you're giving to the world will be equally as high tech. Summarization is nice, but it probably doesn't justify, you know, billions of dollars in valuation. So there's a lot of hype around this idea that hey, we've got these AIs which will do things for us, they will become incredibly productive again. Some of this will happen, some of this is probably in the space of hype and not quite there yet as you're thinking about AI. Just as a quick note here, here is six areas we like to look at. How can you use AI? So one of them clearly easiest low hanging fruit, productivity and efficiency enhancements. I think we all starting to do this. But then you can look into more advanced use cases of AI. And I invite you to really think about your strategy in all of those six buckets. So things like can it give me a competitive advantage and make my field defensible, my company defensible, can you use it to do new product development? Here we're seeing really exciting insights, for example in the manufacturing industry, new material discovery. There was recently a study from MIT, about a week old, which showed that companies, industrial Companies which use AI to do material discovery have on average 44% higher output in terms of new materials, 39% more patents and 17% higher downstream productivity, which is massive. But also think about how can you enhance your customer experience? But also, and this is interesting because a lot of people don't talk about this, how can you enhance the employee experience? And then lastly, of course there's this question around the complete transformation. So an interesting exercise you can do maybe on your way back from the CPA is we like this to call this the killer competitor. So the question of what would the company look like, which would be founded today without any legacy, so any of your legacy systems, your legacy thinking, which would put us out of business, what would a company look like today if you were to found the company as an AI think first business. And it's a really interesting exercise because it's very eye opening to like show you where could you actually invest Pascal?
- [34:13]
B
Just a bunch. As we kind of wrap this up and bring Avni in, just a quick question on just Copilot. You did the LLMs, but just a little bit about comment on Copilot.
- [34:24]
A
So Copilot is an interesting product. I think it's still somewhat for me personally stuck in the Microsoft Clippy world. So it's kind of like it's there, it's nice, but it's not fully actualized. I think if you want to see what Copilot could be in the future is if you look at the original Copilot, which was the GitHub copilot, the coding assistant Copilot. And that is a generally useful tool because it's inside of your workflow and helps a developer to do better development. And I expect Copilot to become much more integrated deeper into the system rather than like the little clippy function which is. Hey, I'm seeing you writing an email. Shall I help you write this email?
- [35:05]
B
Okay, so we can move finish this.
- [35:09]
A
Up and then we'll bring in so very, very briefly. And I just have like literally 30 seconds left for you here. Three things I want to point out to you. I think they're important. There's three facts we see with all of our clients that I do think. And we see that still very few organizations seriously account for the possibility of continued AI improvement in their strategic planning. Please don't do this. AI will get better. You have to take this into account. Fact number two, you actually don't need to know what happens next to realize that you should be planning for multiple contingencies. Nobody can actually tell you what AI will look like in a year from now. Nobody. But you should take into account that AI gets better. And you need to take this into account into your strategic planning. And you want to build for multiple contingencies. And lastly, like it or not, AI will not go away. It is disruptive already today, enough that you should make your decisions about it today and incorporate it into your into your strategy. Even if you believe the technology will stall, it will not move any further. Your challenge is that you have to do this today and at the same time need to plan for this crazy future which is ahead of you. And here's the most important thing you need to learn. This AI you're using today is literally the worst it will ever be. So if you're not happy with it today, Just wait a little bit. It's going to get better. And with that being said, I think we want to have a good conversation about how do you use it.
- [36:37]
B
Well, that was fantastic. Tremendous comments coming in. And one thing, we are using AI. We get 500 to 800 comments for every town hall and we use an AI tool to kind of help digest them. Post town hall. So, Avni Desai, it's great to have you. You've been on town Halls before. CEO of Shellman, top 50 firm, great leader in the profession for us. Two years ago, you were on the town hall in January, just after the announcement of ChatGPT and you actually had already built a bot. So how's the journey going? And welcome.
- [37:13]
F
Well, a lot's changed, I say, in my career, I've been in cybersecurity or security for over 20 years and this is the first technology that I have seen that everybody wants to get on. So I remember cloud computing and people like, well, I want to stay on prem. It took them three to five years, CISOs, but this is the first technology that. I mean, how many millions of people got on ChatGPT? It took Netflix months and years to get there. So a lot's changed. We spent a lot of time building our own LLM and then Chat GPT came out and I was like, oh, my God. Gosh, I can't believe that's what we did. But the good thing is we were able to really. It was a productivity multiplier for us when ChatGPT came on. So what we initially did two years ago was put together a chatbot. We called her Shelley after Shellman. And what the chatbot did was it answered questions from our talent team. So how much PTO do I have left? Where do I find our maternity policy? How do I fill out a form if I need to go on disability? And so forth. And it actually saved multiple hours from our HR team to do that. Once we were able to have this LLM, we have really significantly increased it. We are using it for coding now, we're using it for learning. We're actually putting all of our training and awareness into this chatbot. And anyone at the firm can say, what's the new version of. Of PCI 4.0? And how is that different from 3.0? And now they have it and then they can use that conversation to talk to their managers and so forth. We're also using it in marketing. You know, our. We have a mascot called It's a Koala for our quality and we ask it to build us pictures. And it builds us pictures with the Shulman logo on it. And that has saved again, significant amount of time from our team. But I think where we're really excited and I agree with you is really Agentix AI. We have used a bit of AI to start testing controls. So we do SoC reports and ISO reports. And one of the controls that you test is our password parameters in adherence to the Information Security Policy. And most of that is always structured data. It's in a PDF or it's in a Word document. We built a model which will go through hundreds of Information Security policy, pull out what the password parameter would be and say, yes, the password parameter. But we want to take a step further, right? We want to actually say, can we build a digital worker, a digital agent to actually test the next level of controls? So can you test a control that says have all terminated users been terminated in a timely manner based on the Information Security Policy, and then pull the evidence associated with it? The error rate's way too high. I mean, my error rate on the password parameters is pretty low, the error rate on this. So we have a long way, a long way to go. But I'm really excited about what we're seeing. I mean, but there are a couple of things that I think we have to think about. One, expectations, and I know it's not a security, but you know, expectations I call, I call AI the corporate Ozempic. Everyone thinks it's going to cut the fat right overnight, but it's not. It's going to take a lot of time. It takes, it takes a lot of effort. It's very difficult to build and these chatbots seem like it makes it really easy for us. So you have to plan like you plan any large technology implementation, even if you are using an LLM like llama and so forth. The second, which is near and dear to me is when anytime you put in a lot of data, data breaches is, it's something that has to be top of mind. Especially in our profession. We have client information, we have third party information, we have information from our employees. And when you have all that data, if you don't have good content management, if you haven't trained people of what you can put into the model and what you can't, making sure that you spend time to anonymize the information before you put it in, there are these little extra steps that you're going to have to take. You're going to have to do that. I worry sometimes if you put in client information and Then the model gets trained on that client information. Will you have a data breach that was unintentional. The second also is compliance. What type of information are you putting in there? Are you putting information from someone who lives in California or in the eu? Well, now you're going to have to focus on CCPA or the eu gdpr, for instance. And I think the last one, again, I think it's too much reliance. And we were talking earlier, like I call it augmented intelligence, because in our world we believe in trust but verify. And that's what you have to do. And what I've started seeing people do is the output. They think it's okay. This is the output and I'm going to provide it to the client or this is the output and I'm going to post it. But you have to verify it because error rates are still high, hallucinations still high happen all the time. So those are the things that there's so many, there's so many concerns that you have from a security perspective, but these are the ones that I would focus on, especially if you're dipping your toe into it.
- [42:20]
B
Thank you. I mean, that was a great summary. A lot of people are asking for even more discussion in AI. We do have a one hour webinar that we put on recently that we can put out in the newsletter. And Avi look forward to having you on potentially a future podcast or town hall. And we'll get back to this discussion in open forum. But what I want to do now is move to another very, very important discussion. And actually I think AI is probably playing a role here to help with the talent pipeline. But welcome, Lisa.
- [42:50]
C
Thank you. So let me introduce Lexi Kessler. Lexi is the chair of the National Pipeline Advisory Group, which we've talked about on town hall before. So she's going to quickly revisit that with us. If you are not familiar with what we call npag, it was a group of volunteers from all across the profession that came together to dig into research around what was impacting our talent pipeline. So Lexi did great work there along with all of the volunteers. And we've touched on that the in the past. We're going to dig in a little more. Lexi is also the vice chair of the aicpa, which is wonderful. And next year she'll be taking on the chair role in her day job. She is the Mid Atlantic leader for apprio, a top 100 firm. And I've had the pleasure of working with Lexi since I joined the the AICPA 12 years ago. So thank you for joining us, Lexi. Let's dig in and talk very quickly about about the themes that NPAG uncovered in its research and then a little bit of your experience as a firm leader, as you were working through this research. Sure.
- [44:00]
E
Thanks, Lisa. And one thing I do want to just to add clarity to is the NPAG mission or charge from Council also in basically coming up with some proposed solutions. It was also about preserving mobility. And that is just equally a very, very important thing. And I just wanted to highlight that that was part of the things that we were focused on. But you can see the six themes, they're all related, it's not just one, they're not in order of priority. Tell a more compelling story. All we talk about with that tends to be the inputs, how many hours, not the outputs. Those are the results of what we did. And it's important that when we're talking on LinkedIn, when we're talking at dinner, when we're talking with our friends, when we're at the parties on the golf course, whatever it is, celebrate the successes that we do, we make a difference in the lives of our clients, the lives of our organizations. I bet here that are in person or that attended the digital cpa, I bet you that every single person got two to three nuggets, maybe that you want to take back to your organization and you're going to take it back to your organization. You work really hard to try to make change happen and to implement it. And while it may not be an ozempic result, it may be half ozem ozempic result. So focus on that, celebrate and share the results of the hard work that we are doing as a profession. It's just as important to do that. So I challenge every single person in this room to try to do that. Addressing the time and cost of education that goes into where the 120 and 150 is. That was very clear as a result of our surveys and the data research that we did and came out with the recommendations that will have a link or a QR code where you can see more in depth the results of the recommendations that we have around that. But basically it's going to a model that focuses not so much on hours, but on knowledge and on competency based and taking steps towards that. Not going there as a straight path, but really beginning to change our mindset in our North Star. Because we had some three aha moments, one of which was really around that there are 13 universities across the United States that have changed the qualifications for graduating with a bachelor's degree from 120 credits to 90. So that was an aha moment of what are we going to do? 90 plus 60, then that doesn't make sense. So we really took steps back in coming up with the recommendations that we got there. I think it's important to have that context making the academic experience more engaging. Retooling introduction to accounting. Retooling introduction or retool intermediate accounting where we're not weeding out students, we are pulling them through. It's a different time period from when it was when I was in college. So we need to treat our students differently and how we encourage them to come through and follow through on a profession. Because currently right now only one in every nine students that start in the business school will go on to become a CPA. Imagine if we were able to get 2 out of 9 or data dream 3 out of 9, just get a bigger share of the piece of the pie for us. But we have leakage points in the college system. Well I should say the college system, but really in the funnel we called it, of how when a student enters the business school, what is it that's taking them away from accounting? So that's where telling the more compelling story comes into being and talking about the employee experience. We're not competitive with other professions. We need to raise starting salaries, be competitive with engineers, with architects. We need to commit to manageable workloads. So if we adapt this technology and we're able to save productivity, that means we can improve the employee experience with a work life balance and we'll have more profit to be able to share with our employees and raise starting salaries. So there you go. It's not just 120, 150 providing better support to CPA exam candidates, allowing them to take time off to study for the the exam, possibly even maybe there's something that can be done when students pass the audit section in college that they're able to sit for the audit section of the exam and not wait three years to be able to refresh their memories on the nuances of audit and just as important, expanding paths for underrepresented individuals. Our profession does not match our population and we need to work on that. The data shows that it's important to have diversity in the workforce. It's important to have diversity at the table with the conversations. So what can be doing to scale up college immersion programs helping to transition students from two year college to four year.
- [48:18]
C
Before we move on, I want to really quickly address a couple of questions that are coming in asking where is compensation on this slide? I want to be really clear about this. Compensation is the most important part of that. Enhancing the employee experience. It all starts there. And the work that CPA.com is doing along with PCPs around helping firms transform their business models is focused on compensation, workload flexibility and providing career opportunities. But compensation is embedded in that, enhancing the employee experience. We are not competitive with other career opportunities and we have to challenge how we run our businesses to be able to be more competitive.
- [49:00]
E
And PCPS has a great toolkit for changing the business model because it absolutely has to change.
- [49:05]
C
Thank you. Did I also mention that she's a former chair of pcps? All right, so really quickly I want to move on because Lexi mentioned the Pipeline Advisory Group report. It is full of amazing insights. If you are involved in this profession, you need to download this report and read it. It has just incredible insights about what's causing some of our shortages in the pipeline and what we as an entire profession can do. Lexi, one thing I know that's important to you is the Pipeline Pledge. Can you hit on that really quickly? Sure.
- [49:37]
E
Please take out your phones, scan the Pipeline Pledge. What this is, it's asking each person to do two things in a year. That's it. And we have 600,000 members in the AICPA. If we could get one in four, can you imagine if we had 150,000 people making a difference, going to a community college, going to a high school, changing how you share what you do, talking differently about what you do. So please take the pledge and let's make a difference together. Because remember, we are in this together. We all want the same result. We just need to work together to get there.
- [50:10]
C
Thank you, Lexi. Really quickly, this will be in your slides. I want to remind everyone about an approach that we have to help students get between 120 and 150 hours at a very reasonable cost of $150 per credit hour. You've got more information on there. The deadline for the spring semester is January 1st, so just wanted to make sure I called that out for your attention really quickly. I'm going to scoot through the technical update. You will have these slides. So if I don't. Don't hit on everything. Just know that you've got links and good information available. One of the most important updates for this other than our BOI conversation is good news around 1099 K filings. This was a big topic last year about the same time. And as a reminder, this all relates to legislation that was enacted as part of the American Rescue Plan act of 2021 that basically said that transactions that were running through an entity like PayPal or Venmo would fall down to a threshold of $600 regardless of the number of transactions. That was a substantial difference from the Pre Arpa of $20,000 and at least 201 transactions. So huge shift in the reporting requirements around proceeds being sent to someone from one of those third party pay platforms. In 23, they gave us a temporary pass and lowered that threshold to a more reasonable number. And what we're looking at for 24 is another transition period. So in 24 for tax year 24, the threshold is $5,000 regardless of the number of transactions. They're giving us another year to transition into the $600 requirement. So you can see the numbers there. We have a new resource for members that I've given you the link to as well, and you've got a JVA summary if you want more information. ERC employee retention credit, not a whole lot to bring to your attention, but I know it's, it's such a hot topic and we'll, we'll see if we can get any additional updates for our next town hall, which is next week. But what I wanted to point out on all of this is that the voluntary disclosure program closed on November 22nd. We know you're still getting inappropriate denial letters and our team continues to talk with the IRS about that. So again, more to come on that one next week. I always have to remind you about these amazing resources that our tax section team offers. And this one is a shout out to April Walker, who hosts a reimagining your tax collection practice roundtable series. I've called out the next two because a one is around a PCPS toolkit on right sizing your client base, which I think we're, we're super excited to make sure that you know about. And then one is about using AI and automation to elevate tax workflow. And Ashley Francis, who was here in the room with us today and presented earlier, will be leading that conversation. So you can register for those individually or as a series. And one other shout out to our member insurance program. They've provided a tax season survival guide that gives you a lot of checklists and tax tips to get ready for busy season. All right, there you go.
- [53:33]
B
We're on time.
- [53:34]
C
We're efficient.
- [53:35]
B
Amazing. Amazing. A lot of questions have come in. So Avni, a question related to just AI trust the opportunity that you see for the profession here because a lot of the questions coming in are related to concerns about this. And I know this is on your thinking a lot about how could and I know there's firms and there's a lot of the technical committees thinking about the role the AICPA can play and the members can play in kind of driving trust in AI.
- [54:08]
F
Yeah. Well, I'm going to quote something that Barry said. We own the trust industry. And I think when you take a look at what people are concerned about in AI, it's responsible AI, is it safe and can I trust it? So I do think there is a huge opportunity for the AICPA and the profession to play here. What we've already started seeing is a lot of companies like ERP systems or, you know, payroll providers are overlaying AI over their system and we're doing a SOC report on the system.
- [54:35]
B
Right.
- [54:35]
F
And that is providing them some level of assurance and trust around the controls associated with that. I think as AI grows and we get into like agentic AI, there's going to have to be a little bit more. It's not just going to be cybersecurity controls, it's going to be controls around hallucination and ethics and who has access and security. And so we're going to have to just make that a little bit more robust. But I'm very confident that the profession, I mean, we own the trust industry and that's what people want to do is they want to be able to trust the platform.
- [55:05]
B
Yeah. So a lot more to come on that. Lisa, I know you're looking at the questions BOI attacks. What?
- [55:13]
C
There's a common BOI question that I'll direct to Rachel and it is around what could the change in administration potentially mean for boi? Filing a request requirements.
- [55:25]
D
So there is a possibility that in the new administration that they could make changes to boi, that they could, you know, implement a delay. We don't know. It's uncertain. So right now you should plan accordingly as the law is written. But there is the possibility that there could be something done in a new administration.
- [55:54]
C
Thank you. Lots of questions about the pipeline and how to create work life balance situations in firms. And I will call out a podcast that we've recently relaunched called Transforming Tales of Business Evolution. In that podcast we talk with firm leaders who are working hard to change the culture of 80 hour weeks during busy season or 60 hour weeks all year long. So if you can go to your favorite podcast streaming platform and get that Transforming Tales of Business Evolution, you'll get insights from firm leaders across the profession of all firm sizes who are making these changes so that their employees and themselves can have that better work life balance.
- [56:41]
B
Well, thanks, Lisa. I think we'll move on to cover some of these closing resources, but first want to thank the panel. Pascal, Aveni, Rachel and Lexi, thank you very much for the participation in today's town halls. A little round of applause there. So Pascal Already referenced the CPA.com Generative AI Toolkit. People are asking how they can access this. You can click on the link if you download this program presentation and we'll also be highlighting this in the news, the town hall newsletter that will come out early next week. We do AICPA Town hall rebroadcasts now on Mondays so you can join us 1pm on Monday Eastern time. And Lisa, it's early on the West Coast 7am but we still had close to 10,000 with us today, so probably over 11,000 plus when you include the Digital CPA community. So information on how to subscribe to the AICPA Town hall, scan this and register and we'll get calendar invites out to you. Greatly appreciate the participation of the town hall community as we appreciate the participation of the digital CPA community. We learned from you and this is a very, very valuable time for us. So we're going to close out 2024 Town hall series with Barry Malonson joining us as well as Marcy Russell. Marcy is an economist and that'll be an interesting discussion how she's seeing 2025.
- [58:20]
C
It's a competition on who speaks faster, Marcy or Pascal.
- [58:26]
B
Okay. So with that, I'd like to thank the town hall community for being with us and I also want to thank everybody here at Digital CPA for being with us for this session. We've got two more great concurrent sessions occurring this morning and then I encourage you all to be back here, not miss the final keynote with Hannah Pryor. She's going to talk about some techniques to drive influence and change. All the keynotes have been fantastic here. You do not want to miss this final one. So have a great morning. Have a great day. To all of you online, thank you very much and we'll see you back here shortly. And we'll see the Town hall community on the 19th. Thank you.
- [59:11]
A
Thank you for your participation. You can now subscribe to the AICPA Town Hall Series on your favorite podcast.
- [59:18]
C
Platform, as well as watch archives on.
- [59:21]
A
YouTube and AICPA TV. Tune in for live broadcasts Thursdays at 3pm Eastern Time.
- [59:29]
B
This podcast is designed to provide illustrative information with respect to the subject matter.
- [59:34]
C
Covered and does not represent an official.
- [59:36]
B
Opinion or position of the AICPA or AICPA.org it is provided with the understanding.
- [59:42]
C
That The AICPA and AICPA.org are not engaged in offering legal, accounting or other professional service.
- [59:49]
B
If such advice or expert assistance is.
- [59:51]
C
Required, the services of a competent professional.
- [59:54]
B
Person should be sought.
- [59:55]
C
The AICPA and AICPA.org make no representations.
- [59:59]
B
Warranties, or guarantees as to, and assume.
- [60:02]
C
No responsibility for the content or application of the material contained herein and especially.
- [60:08]
B
Disclaim all liability for any damages arising out of the use of, reference to, or reliance on such material.