Ascend 创始人 David Wurtzbacher:AI Native 会计师事务所需以 AI 为核心重构前后台与运营底层
The 'AI Native' firm
Ascend 创始人兼 CEO David Wurtzbacher 指出,"AI Native"会计师事务所需以 AI 为核心重构前后台全部流程,而非在现有流程上叠加 AI。他介绍,Ascend 已组建 30 人 AI 团队与 CPA 协作,并通过"闭环"模式掌控所有后台工程,捕获工作流、点击、修改及客户偏好等数字足迹供 AI 自我学习;单纯拼凑多种工具并非通往 AI Native 的路径。

Unlocking the capabilities of artificial intelligence may well require rethinking the entire structure of the accounting firm, suggests Ascend founder and CEO David Wurtzbacher.
Transcription (This transcription was created by AI; in the case of discrepancies, refer to the recording):
Dan Hood (00:03):
Welcome to On the Air with Accounting Today. I'm editor-in-chief Dan Hood. You could make the argument that much of the impact of technology and accounting over the past 30 years has been about making firms and accountants more efficient. Hasn't really changed what they do or how they do it. It's really just made them faster. But the rise in the ever-increasing capabilities of artificial intelligence have given rise to a vision of a firm that operates genuinely differently in a whole host of ways because they're infused with AI. And there are practitioners out there right now building what's being called AI native firms. They're experimenting with a model that they really are, to a certain extent, building on the fly. And the results and the implications are, to say the least, really, really interesting. Here to talk about all this and what it means for the profession as a whole is David Wurtzbacher.
(00:44):
He's the founder and CEO of PE-based accounting firm platform Ascend. And I should mention, they have a 30-person AI team that's working alongside their CPAs on a whole host of ways that AI can be applied in accounting firms. So David, you're pretty well positioned to talk about this, certainly a better position than most given that this is a relatively new thing. Thank you for joining us.
David Wurzbacher (01:03):
Thanks, Dan. Great to be with you again.
Dan Hood (01:07):
Yeah. Let's start this. When I say AI native, what do we mean by that? What does that phrase mean for us?
David Wurzbacher (01:16):
Yeah, I think a helpful starting point is just to recognize that we're in this crazy moment where the AI models themselves are becoming so powerful. You're seeing the headlines now about, oh my gosh, they cracked these decades long math problems that no one's been able to solve before. And so they really can do an enormous number of knowledge work tasks, and that trajectory is only becoming increasingly true over time. And so when you have that kind of progress, it allows for some pretty big thinking about what is possible. And I think in the accounting business you can ask questions like if we were starting over today, would we build the firm this way? Would we design it in this shape? Would we have these same processes? Is this what production work and client service would look like? And I think that the folks thinking about AI native are coming up with the answer, no, we wouldn't do all these things the same way.
(02:07):
We would completely turn the tables differently and reimagine everything with AI at the core rather than as something that you add in or put on top of your existing processes. And so that's really what AI native means is that you're reimagining every process in the firm from front office to back office with capability at the core, and it's sort of an innovation that is AI first in its posture rather than AI sometimes in posture.
Dan Hood (02:38):
Gotcha. Well, that's the question is how AI native can you be if you don't have AI everywhere? Is it really required, you talked about back office, front office. Is it possible to be an AI native firm, but only on the back office only for our internal stuff or only for our stuff that's facing clients? Or is it really, do we think it's going to be the whole McGill?
David Wurzbacher (02:58):
I think it's the whole McGill. Mix and match of tools has been the way in public accounting for as long as technology has been around in the profession and it has a lot of challenges. You've got to have time to evaluate the vendors and keep up with everything and then integrate them with your other tools and keep your data straight, retrain your people, and it's a little bit exhausting. But maybe a couple of inspiring dimensions of an AI native vision. One would be, well, what if all of our work was converging into a single pane of glass, no more hopping from app to app and downloading here and uploading there and mastering works of different pieces of technology? And that part of the vision's pretty easy to understand, but here's something that is a bit more of a mind bender to understand the difference between AI native and a regular operation.
(03:49):
And I'm going to introduce a concept here that we refer to as the closed loop, and we didn't invent that term. If you follow the AI world, a closed loop is kind of a hot buzzword. But basically what happens in a closed loop is that you control all of the backend engineering and substrate of the software tools that you're using. And so that allows you to capture every bit of information that is going on in your organization. So every trace of a workflow, every click, every time you correct a field, every communication, every client preference or nuance. And when you're operating in a closed loop, all of that digital exhaust, so to speak, can compound into a data set that you own and that AI can learn from. And when AI starts learning from all of this digital exhaust, it can start to create self-improvement in the product itself, which can really change the nature of the work, it can change the nature of client service, and I think it's going to bring about a lot of different potential for innovation in the profession that's really, really exciting.
(05:01):
But I do think that mix and match is not really a path to AI native, which puts you on the track of needing to build it yourself.
Dan Hood (05:12):
Right, right. Well, Amy, to your point just now, it isn't a single pane of glass. If you're mixing and matching, you've always got that problem of how do I get data from one place to another and there's friction and there's silos and so on. And that has been a pain point for accounting firms for a million years. So even if it just solved that, even if it's nothing else to pick to make running an accounting firm easier or better, having all the data in one silo would be pretty exciting for a lot of people and reducing just an enormous number of pain points.
David Wurzbacher (05:47):
Dan, can I give you a couple examples of what you might be able to do with an AI native closed loop? So right now, if we just think of timekeeping, for example, everyone goes into system and enters their time. When you control the entire software layer, all that timekeeping can happen automatically and you can compare productivity across projects, you can recommend a price for projects based on a certain amount of scope. You can capture who's moving through work quickly and who's getting stuck in the work and then present real time L&D that's personalized to that person using the software tool. You can automatically remember everything that a client has ever told you about their business or their family. And so it really starts to get very powerful when you start pulling on the thread.
Dan Hood (06:35):
Right. And this raises an interesting question that we hadn't really talked about before, but when you talk about that closed loop and having the system have access to all those sort of things, in the end, are we really talking about an AI that's replacing all of your current tech stack? It's replacing your timekeeping, it's replacing your tax prep, it's replacing your GL, it's replacing all the things where you would normally be getting all that information from, or is your AI interacting with all those systems?
David Wurzbacher (07:10):
I think that the purest interpretation of what AI native means and the long-term end state is that you're not going to be very dependent on other software tools and you're doing a lot of this work yourself with your own tools. Realistically, the path from here to there is going to travel through some level of integration with the tools that are already important to workflows today, in particular a tax engine that's running all the actual calculations and filing the returns and the various jurisdictions. But I think in the end, the pure interpretation of AI native is that you're actually not cobbling together a bunch of different tools from software vendors.
Dan Hood (07:51):
Right, right. Well, as you said, this is still pretty early days. People are still building this model, but I say the long-term vision that would make, it would certainly make sense. I mean, we're seeing AI be able to code things, so why couldn't it code in the long term, a tax engine or a GL or whatever? We mentioned there's some firms, you know some firms that are doing this, I have run across a couple of them. They're all relatively small. The one I know of particularly well is literally one person, but it's a guy who's built a California name who's built a million dollar firm based with just him and a bunch of agents. And he suggests that at a certain point you each think, well, I got to have some other people in here and I can't just be an entirely all AI firm.
(08:39):
What is the scalability issue here? Is there a way to make a really large firm AI native?
David Wurzbacher (08:47):
I think there is. First of all, that example is pretty amazing when you think about it, one person doing a million dollars of revenue, that's got to be the highest revenue per employee of any accounting firm in America, if you think about it, you compare it to all the stats out there. AI is so powerful at an individualized level, but I think the trap here is that you build all these agents around a single person and what you have at the end is an AI-powered job rather than an AI-powered business. And ultimately what you want is an AI-powered business and the business has got to have scalable processes that necessarily have to outgrow the individual at some point. So I think what you're seeing where a really innovative person is building a bunch of agents around themselves, that's totally the natural starting point for something like this, but ultimately it has to grow and become more capable and more scalable, and that is going to require a different level of building than the vibe coding way of using the AI tools that exist today.
(09:53):
And in particular, I think you've got to spend a lot of time and money and energy building what is called an AI harness that is code written into a program that's telling the models, "Hey, you can have this information at this time, and then when I get it back, I'm going to test it and I'm going to order all the steps that come next." And when you can get the workflows operating in a harness like that with a software layer on top of it that the human is interacting with, then you start to get into an AI native stack that actually is scalable and that could work in a top 20 firm.
Dan Hood (10:31):
Wow. I think that raises several questions. I mean, it's fascinating and it answers my question, but it also brings up a number of other questions. I'm not really sure which one I want to go with here of the two ones that come to most mind. Maybe we talk about literally getting the software for this. Is this an off-the-shelf kind of thing or is this really firms going to have to build a lot of it, a lot or all of this themselves, or is there some middle ground? As we said, for a while, it will be working in tandem with the software you already have from mainstream vendors. But is the long run, you're really building this yourself or you're training these models yourself?
David Wurzbacher (11:12):
So there is no way to go buy an AI native product suite off the shelf today. And part of the reason for that is because this is more than just technology or we're talking about an operating model that's very different from the way firms operate today. But another reason is that in the world of accounting technology and the venture capital that backs these companies, the venture capital investors, they want to see growth really, really quickly. And the way to do that is to create a narrow point solution that you're only selling to one or two people and to get the sales momentum going that way. If they were to build a huge product suite that crossed all the major service lines and crossed over into management, can you even imagine a CPA firm making a decision like that together? It's not even conceivable really. And so the vendors aren't really structurally set up to go and build something like this for us.
(12:10):
And even if they could, it would require handing over so much of your institutional know-how and your client context that you'd be putting yourself in a pretty vulnerable position. So it's a tricky place for the profession to be in where building is so obviously better than buying, but buying is also the most accessible path into AI at the moment. So we're in this really tricky period of time where folks have to decide which path are they going to go down.
Dan Hood (12:36):
Yeah. And as we say, we're probably going to say this, I'm going to say this four or 500 times over the course of this discussion, it's such early days yet that there's a lot of transition to get through. In a normal time, I would say there's two, three, four years worth of transition because it's AI or because the base is going, I'm going to say there's probably 15 to 20 minutes worth of transition that we'll go through before we settle into a new state. But no, it changes so quickly and it's such really relatively new thing that the move through to whatever the final state may be will go through a lot of stages and take some time. I do want to say, it's interesting because when you think about this and as you talk, I'm imagining, well, what does that look like? And you think of things like when you go from firm to firm, for instance, if they used a different system than if they used a Wolters Kluwer system as opposed to a Thomson Reuters system, you would learn that new system and it would be a new thing and they're not that hard to learn, but you still have to do it.
(13:37):
And I'm wondering, does this mean that every firm ends up having its own sort of system, one, where you have to learn the new system when you go to a new firm and you couldn't possibly know it before, you don't have the choice of being, "Well, I know one of the three or four accounting programs, so I'm pretty good." Or, and this goes to the nature of AI, the interaction with AI is so much more human level. It's sort of a natural language is the word I'm looking for, I think, that it doesn't matter. It's like you're talking to the computer and the computer just sort of knows. I mean, do we think that's where we end up? Is it the interface becomes so much more easier and frictionless for accountants that it doesn't matter that they're using their own individual model?
David Wurzbacher (14:22):
I think your intuition on this is dead on, Dan. I think everybody that builds something like this is going to do it in a slightly different way, but it's going to be so much more intuitive and easy to interact with because it's being built in the AI era that I think it will be easier to go from one firm to another and use whatever the systems are. And I think between now and then, I think a lot of the talent is going to place their career bets, so to speak, at firms that look like they're really on an innovative forward-thinking AI path rather than stay planted at a firm that's moving a little slower.
Dan Hood (14:58):
Right, right. I'm just thinking of you talking about particularly younger staff, they're the generation that... The last time we described something as native, we were talking about new generations coming into the workforce as being digital natives. And so their level of comfort, their level of interest and their level of expectations around all these things are much higher than.
David Wurzbacher (15:20):
Yeah, we have just an interesting story for you to this point. We have a bunch of these tools out there in our firms that we've developed, and one of our firms was telling us that their interns in the spring busy season have leveled themselves up to staff level without extra training because it's never been easier to learn. We've also got chat features right inside with the work product, so you could go in there and ask, "Hey, what changed in this return? Why did the partner make that change? What else did it affect in this project?" And you can actually start teaching yourself a lot more real time than you have ever been able to in the apprentice model where you have to wait for busy season to die down, then maybe the partner will remember about the thing that they needed to teach you or maybe they won't.
(16:06):
So we're entering a very different era.
Dan Hood (16:10):
Well, that's fascinating. I think a lot of people will be very happy to hear that because one of the big questions, and it's not just AI related, but in terms of the way technology and outsourcing and a number of other things are affecting the model of the accounting profession as a whole, is that they are cutting out a lot of that internship sort of level, the first five or six years you work at a firm where you just do a bunch of grunt work and learn that way. That's how you learn. And people are saying, "Well, if we replace all that with technology and outsourcing or replace a lot of it, how are we going to train these people?" So effectively we're saying you can train yourself. I mean, no, we're not saying that, but the capability is there, certainly the possibility of the learning becomes much easier because you don't have to wait for someone to hand it to you.
David Wurzbacher (16:52):
Something that I think is kind of funny is every firm has their rockstar partners who are really smart and they understand complicated concepts across a bunch of different domains, and they didn't go to a training that everybody else missed. They're just curious, hardworking sponges that learn a lot over the course of their career. And I think increasingly we are going to want to make sure we're hiring young people that have those attributes, because if you have those attributes against the surface area of how easy it is to learn and teach yourself these days, particularly if you're compounding all of your client context and your firm know-how into one of these AI native platforms, it really is going to change the learning and development game quite a bit.
Dan Hood (17:39):
Definitely, definitely. The other question that I was thinking about on this front was, there's a degree to which if you're just starting your firm, it's a little bit easier to get into this, or I'm going to say a lot easier, you have no structures in place, you can start with low barriers to entry and no legacy systems that you need to upend. Is this something right now that it's just for startup firms? Or I mean, what does this look like at a firm that's been in operation for a while, has a bunch of people and this, maybe some older partners, has maybe a lot of legacy and ingrained workflows and so on? How much more difficult for them is this the switch to an AI native idea?
David Wurzbacher (18:24):
I think it is far easier to go down this path if you're new and if you're a startup because of all of the advantages that you just described, those are sort of the disruptor advantages. However, the incumbents also have quite a bit of advantages that those startups do not have, like tons of data and tons of clients and proven service offerings and lots of experienced professionals that hold today's intelligence, and a lot of it is in their head, and that stuff has to get out in order for the AI to learn and self-improve. But the incumbents, they're not going to be as fast and they're not going to do the change manager as quick. So I think either can do it. And the trick is in if you're a startup firm, is there a way for you to get some scale so that you can benefit from those advantages of scale?
(19:15):
And if you already have the scale, can you innovate with your own organization outside of the tech stack that everyone currently operates in so that you can move with enough pace? And I think the dream is just to manufacture a scenario where you have the advantages of both the incumbents and the disruptor. And if you can put the two together, it's really powerful.
Dan Hood (19:35):
Right. If you can have a firm with a long history that had never existed before, that would be fantastic.
David Wurzbacher (19:41):
Yeah, yeah.
Dan Hood (19:43):
That way to put it, but you're right. I mean, there are interesting benefits for both sides. All right, there's a lot to think about here. We're going to take a quick break to let some of that settle, but then we'll be back in just a second.
All right, and we're back, and we're talking with David Wurtzbacher. He's the CEO of Ascend, which has been, we mentioned they've been experimenting and looking a lot at AI and what it means for the accounting profession. And we've been talking about the notion of the AI native firm and what it means. I wanted to pivot a little bit to talk about what does this mean for the structure of a firm? I keep going back to, because I only know one firm that's doing, I know there's a couple others doing, but I only know one of them. And he's got a firm with a very different structure, I think, than some others.
(20:34):
I know there are some larger ones that are working in AI native things, but what does it mean for the structure of the firm? What kind of potential changes would we be looking for? And there's a million different aspects of it, but partnership structure, for instance, or compensation or cost structures. You mentioned having that single pane of glass, you could figure out very easily what a thing costs and how much it costs to work it. We'd also get a sense of who's struggling in an area, that sort of thing. It brings up all kinds of potential ramifications in almost every area of the firm. Pick an area and say how and lead us in some thinking about how it's going to impact the structure of a firm.
David Wurzbacher (21:13):
Well, I think there are some critical success factors for building an AI native stack that are, you got to have capital to invest, you have to have an innovative culture, you've got to be able to attract some technology talent that can actually help you do this in a way that is secure and scalable and auditable and all the things that are important to quality management in this profession. You got to have some pace of play and speed in your decision making. And I think that you can have those things inside a legacy partnership model, but I think you're probably more likely to see it in a corporate model. We operate in a corporate model, that one guy out on the West Coast who's building an AI firm, that is effectively a corporate model because you just have one person making the decisions and holding all the agents accountable.
(21:58):
And then I think in terms of the cost structures, what this kind of technology is going to be able to do is it's going to simultaneously increase the value that we can deliver to clients because it becomes more personalized, more proactive, more anticipatory, and also decreases the cost to serve because now so much of the production work is actually being handled by the AI agents rather than staff. You will still need staff just to be clear, but your cost to serve goes down. So when your revenue's going up and your cost to serve is going down, you all of a sudden have a much more profitable business that makes you more competitive in the marketplace that you could maybe afford to pay people what other firms couldn't afford to pay. You might be able to win client work that other people wouldn't be able to win.
(22:46):
You might be able to convince a smaller firm to sell to you because you have a way of creating value that others do not. And so I think that new cost structure is going to be a pretty big game changer. And I do believe, certainly at Ascend, this will be the case, that some of the gains that come from this kind of innovation are going to go right back to the professionals who are delivering the client service. I think it does create potential for probably the most rewarding the career path has ever been.
Dan Hood (23:18):
I want to dive more to the career path because it's a fascinating... I don't know that everyone thinks about how it's a third order, second order ramification of all this. But maybe just a quick though on, are we at all concerned about the tokenization effect or the cost of AI in terms of this? I think a lot of companies in general, not just accounting firms, brought on a bunch of AI tools, started experimenting with them and didn't realize that there were, I think, greater costs with them than they expected. Some of this may be teething pains for AI as a industry sector or as an overall utility in some ways, you want to think about it that way. But any thoughts on what the cost of AI might do to this model or how it might impact it or how the adoption of it might impact the cost of AI?
(24:06):
I
David Wurzbacher (24:07):
Think that the cost of compute will continue to go down, and you also have this optionality of saying, "Hey, do we really need frontier intelligence to do data extraction from a K-1? Or is there some lower level of intelligence that can accomplish that task?" And I think if you're a one-man shop, it would be a lot to try to synthesize all the different model developments that get released on a weekly basis. But if you have some team around it that can really focus on it, I actually think there's tremendous opportunity for cost containment on the token side. And I think riding the innovation curve of those models and open source models and everything, there's this new tool called Jev that's really fast and really cheap. And it's obviously a net new cost in the cost structure, but I do not think that it's going to be very substantial.
Dan Hood (25:00):
Gotcha. This is going to be reference number 297 to the fact that the state of play is in tremendous change because that's an excellent point, that new models will be coming along that may be less capable, less expensive, and can they do the work you need? But then it becomes a question of managing that. Someone in the firm has to be paying attention to presumably a CTO or something like that, is looking at that at all times and saying similarly the way your chief HR officer would be looking at how much are we paying for staff? Are we paying too much, too little, et cetera, et cetera. But it's an angle that people would need to pay attention to. Interesting. All right, let's talk a little bit about career paths. You had mentioned some of the positive upside, I think, for a career path for someone coming into accounting, but it has really the potential in the same way that technology does, and that is already changing the career path in accounting.
(25:56):
It used to be, right? We've talked about it, four or five years of grunt work and you learn from that and then you move on. People are compressing that, among other things, and that's just a start. How do you see this impacting career paths more generally?
David Wurzbacher (26:08):
Well, one of the things that I think is pretty wild to think about is that I really believe that the end state of this is that very little of the production work that gets done today is even going to be necessary in the future. It's still necessary for client work, but it's going to become overtaken by AI in the workflow. And for us at Ascend, I think that end state is less than two years away and maybe only a year and a half away. But what happens when more and more of the production work is getting done by technology, the cool opportunity here is that the career elevates to a new kind of trusted advisor that's not burdened by all of the burnout and busy season chaos that happens in all the service lines that we operate. And so I think the career will become much more relationship driven, taking care of clients, meeting with them, planning things, providing advisory at the right times, applying judgment, winning new business.
(27:08):
And there are people today that are sort of post-production in this way already, just doesn't describe most of the profession. And I think what happens over time is that will become more of the norm rather than the exception. Another thing I think about career paths is that I think that the industry is going to regeneralize. So being a generalist was a little more common decades ago, and then things started to specialize, and I think that actually might unwind a little bit because the AI itself can supplement with so much specialized intelligence. And I think what clients really want is a single point of contact that can help them across domains to be the integrator of their financial life in their business or their personal life, their family, whatnot, the charities they're involved with. And I think becoming a dangerous enough across service lines type of a generalist will become also more of the norm than it is today.
Dan Hood (28:05):
Interesting. Yeah, because there's definitely, even today, there's still a push to say, "Hey, specialization, specialization niches, this is where things are headed." So it'd be interesting to see that, if that should reverse. I think we haven't really talked much about clients, which is probably not good. We should have brought clients in a little earlier, but just don't let them know that we weren't thinking about them first. But I mean, I guess the question is, how much does it impact clients? Does it really make a difference? I mean, hopefully in theory, they don't know what tools you're using in the back office unless you tell them, how does it impact them?
David Wurzbacher (28:42):
Well, there are some people who traffic in this AI native space who really believe that technology is going to move to the forefront of how the client interacts with the firm. I actually don't agree with that. I think that the right interface between the CPA firm and the client is the actual trusted advisor themselves and human form. However, a strong tech stack in this area is extensible into a better digital experience for the client. A lot of clients today actually interface with their CPA firm across four, five, six, seven different applications. I go here to upload my tax documents, I go to some other place, upload my audit documents. I go over here to pay my bill. I go into this other tool to sign my tax return. I actually think that the client experiences on whatever digital interface is there is going to get a whole lot better.
(29:38):
But most importantly, what clients value is memory and responsiveness. What I mean by that is I want an advisor who remembers all of my context so that when I ask them a question, they give me an answer that is informed by all that context that they know about me and nobody else knows about me. And when I ask a question, I want to be responded too quickly. Also, I want someone who's very proactive because they're holding my context against the changing world around me and then giving me ideas at just the right time in my life. And I think technology is actually really powerful for... It's obviously very powerful with memory and respectness and being proactive and doing research. So I think the client is going to experience a way better version of the trusted advisor. I think they're going to really feel like they're getting the advice and judgment and trust that they have always valued.
(30:33):
So that's a pretty exciting picture, I think.
Dan Hood (30:36):
Yeah. Yeah. They may not know where it comes from. They may not know that they're, why is my advisor so much better prepared than in the past?
(30:46):
But then that raises the interesting question of, and there's certainly ways around this, but one of the ways that accountants got to know their clients really, really well was by doing their work, by literally doing the tax return over and over and over again and seeing a bunch of other tax returns. And then so when they saw this client's tax return, they could, "Hey, I saw something similar somewhere else." And obviously there's training and ways that people can help themselves with this, but it does raise the basic question of if you're not doing that work, can you be as well-informed as you were when you had your hands into the muck of the client?
David Wurzbacher (31:25):
I think you will be more well-informed because in the past you can only be as smart as one person is capable of being. But now in this AI native stack, you have technology working for you at all times. So if I'm about to go meet with somebody in a manufacturing context, maybe in the old version of this, I have time to research my other manufacturing clients before I go or maybe not. But in an AI world, all of that research and benchmarking and context can be automatically generated based on seeing the calendar invite in your Outlook. And we can serve up that kind of intelligence to advisors just at the right time before they go in. And so I think there is going to be a lot of potential to leverage technology for that pattern recognition that you're describing.
Dan Hood (32:16):
Gotcha. Makes sense. I'm pretty sure this will be the last reference to things are changing very quickly, so who knows if any of this will be true in 20 minutes. But there's a certain degree, if you're looking at building an AI native firm now or experimenting with this sort of thing, to what degree do we say, "Hey, this is a great opportunity for first mover advantage. You could get there ahead of everybody else. You could be on the cutting edge of this sort of thing." Or do we look and say, "We're not really waiting for a larger software vendor, for instance, to handle this, but are we looking to see which models work best and can we wait a little bit and see, hey, the first 20 firms, 15 of them failed because they did this, but five of them really jumped ahead because they did this." Is it something where the first mover advantage outweighs waiting and seeing what works?
David Wurzbacher (33:08):
I think it does because the gap, when you move from the way firms operate today into this AI native paradigm where you start to get self-improvement in the underlying technology and then therefore in the service, I think the gap between the haves and the have nots is going to widen faster than any sort of innovation gap has ever widened in the past. It's just a direct reflection of how much faster the AI innovation wave is happening than any other prior technology innovation waves. Something that I think is kind of interesting is there's this thing called Y Combinator, which is a startup incubator out in the West Coast.You've probably heard of it. Sam Altman who runs OpenAI used to run Y Combinator. And their number one request for startups right now is AI native service firms that are bypassing the service providers and going straight to the clients.
(34:04):
And this is an actual threat to us in public account.
(34:11):
And these models are going to disrupt the competitive equilibrium that exists today. It's going to change what kinds of services clients expect, where they get those services, how quickly they receive those services, what they pay for those services. It's really going to change a lot. And you want to be on the forefront of a changing standard. I'll tell people in our business at Ascend, 10 years ago, none of us expected next day delivery. We didn't, but Amazon reset the standard and now it's like, man, if I order something and don't get it tomorrow, I'm actually disappointed about how long it's taking me to get my thing. No one really kept up with Amazon in the end and they were the standard setter. So I think it's important I think for any firm leader to think about how the competitive landscape is changing and therefore what that demands of you as a leader to put yourself on the forefront of how those things are changing.
Dan Hood (35:07):
Makes sense. And that holds true for clients, it holds true for markets, but also holds true for staff. The people who are, as you said, the people who are excited about this, the sponges, the forward thinkers are going to want to work for firms that are in that area. We know how difficult it is to find quality talent these days. Excellent. This is a topic we could dive into a lot more and probably will need to dive a lot more into in the future as this model gets maybe next Tuesday as the model matures because that's how quickly it's moving. But for now, any final thoughts, any final takeaways? You mentioned something firms should be looking at sooner rather than later because the pace of change in this area is so fast that you don't want to be left behind. But other thoughts, other key takeaways?
David Wurzbacher (35:55):
I think what I would close with, Dan, is that I'm not a CPA. I'm relatively new to the profession. I've only been around here for four years now, but I always hoped that we could contribute to a revitalization in the profession. It's a wonderful profession, but it's also saddled with a lot of limitations in this particular market moment. And one of the things that has been really encouraging for me as I've dealt with the AI issue is I think that we have great potential to leave the worst of public accounting behind us, the backwards looking work, the administrative work, the burnout, the late nights during busy season, the mastering the quirks of our different technology tools that we use. And instead, I think we're going to usher in an era that is the best of public accounting, which is being with clients and being helpful to them and solving real problems and looking forward and elevating yourself to your highest and best use supported by technology that is intuitive and serves you.
(36:58):
I'm just very excited about this moment of transition for the profession because I think it is going to create a lot of revitalization.
Dan Hood (37:07):
Yeah, it's a very exciting prospect. When you put it that way, it sounds great. Very cool. So something everyone should be paying attention to. We're looking forward to seeing how it plays out. If it's even a 10th is great, as people suggested, it'll be fantastic. David Wurtzbacher of Ascend, thank you so much for joining us.
David Wurzbacher (37:26):
Thanks, that was fun.
Dan Hood (37:28):
Thank you all for listening. This episode of On the Air was produced by Accounting Today with audio production by Adnan Khan. Rate and review us on your favorite podcast platform and see the rest of our content on accountingtoday.com. Thanks again to our guests and thank you for listening.

Editor-in-chief
Daniel Hood has been editor-in-chief of Accounting Today since 2011; he joined Accounting Today as managing editor in 1998. A graduate of Georgetown... Read full bio
来源:Accounting Today(美国会计行业媒体) · accountingtoday.com