This Is The Next Industry AI Will Disrupt

1. DVDominic Vitucci0:00

   From my perspective, I understand why it's hard to imagine a post Big Four, a post accounting firm world. It is a foreign concept. Conversely, I don't actually think they've earned the right to maintain the reverence that they've, they've been granted for all of these decades, centuries sometimes. And so if this keeps going the way that I think it's gonna keep going, the way that I'm kind of betting the house that it's gonna keep going, is that's exactly what we're looking at, right? A, a functional and tectonic shift in the way that these outcomes are ultimately provided to customers. Accountants have kind of just wormed their way into the middle and artificially sel- set themselves up to be, you know, "I'm the middleman. I'm the expert." But like what if we just didn't have to have that?

2. TBTom Blomfield0:39

   [upbeat music] Welcome back to another episode of The Breakdown. Today we are joined by Dominic Vitucci, the CEO and founder of Onshore, uh, which uses AI to automate corporate accounting and tax. Dominic, thanks for joining us.

3. DVDominic Vitucci0:57

   Thanks so much.

4. TBTom Blomfield0:58

   What is the job of an accountant and an auditor at a, at a level of, you know, how a c- c- computer scientist might-

5. DVDominic Vitucci1:04

   Right

6. TBTom Blomfield1:04

   ... like describe it?

7. DVDominic Vitucci1:05

   You know, you start off as a junior guy. You're, you're 22 years old. You come out, uh, you, you... Hopefully you passed your CPA exam. If not, you can take it while you work. There's a bunch of professional stuff. But you show up and, and really blocking and tackling nuts and bolts is open a spreadsheet and copy some data from one type of document to a different type of document, and like do some formulas to do some arithmetic, right? So a good example of this. I was doing a R&D, research and development tax credits, right? And it was, I started on Tuesday and on Friday, the partner said, "Hey, let's go out to this client site." It was a client that made boxes. We went out there and it was me, this manager, and this partner. And I brought a yellow legal pad and a pen. And the partner said, "Hey, listen. Don't say anything, just take notes." Okay, easy enough. So we show up. Partner, we have a meeting scheduled all day Friday, and then Monday, Tuesday the next week. Eight hours, 30 minutes with as many employees as we can get at that company. Just like you, Tom, you come sit down and say, "Hey, listen. Here's a piece of paper. Read the paper." And you took, you know, 90 seconds to read this paper. It was all about the R&D tax credit rules. And so you read the paper, and then you look up, and the partner looks you in the eye and says, "Well, how much of your time do you spend doing an R&D tax credit?" And now you're an engineer, you're a sharp guy. You say, "I don't know, 30%. Can I leave now?" Partner says, "Oh, 30%. Uh, don't you mean like 80%?" And the guy goes, "Yeah, 80%." Partner says, "Oh, perfect. Dominic, write that down." And so eight zero, and I did that for 20 something hours. Then the worst part is I go back to the office, open my legal pad, and then I just, you know, Tom Blomfield eight zero equals this times it, and that, that's it.

8. DLDavid Lieb2:37

   Okay, so the idea of the tax credit is like there's certain kinds of work-

9. DVDominic Vitucci2:40

   Right

10. DLDavid Lieb2:41

    ... where the, if you're doing that work, the federal government wants to incentivize it, so we'll-

11. DVDominic Vitucci2:45

    Right

12. DLDavid Lieb2:45

    ... kind of give you money back-

13. DVDominic Vitucci2:47

    Exactly

14. DLDavid Lieb2:47

    ... if, if you have engineers doing r- research basically.

15. DVDominic Vitucci2:49

    Anybody... I mean, in, in this particular domain, and this is one of our like flagship products, right? So I, I do wanna take just a beat to talk about it. Really, it's an incentive at the federal and the state level in the United States, amongst other countries in the world, that encourages research to be done onshore domestically at some of these, uh, organizations, right? The idea being, hey, if you're conducting research and that's the creation of a new product, a new process, or a new technique, or the improvement of a product, process, or technique, you should get an incentive, especially if you're keeping technical jobs in the United States, for example.

16. DLDavid Lieb3:18

    Yeah. What you're basically saying is, like, this, this is a pretty straightforward task of, like, identifying how much of each employee's work is in a sort of area that's, um, eligible for this tax credit.

17. DVDominic Vitucci3:31

    Right. So straightforward arithmetically, yes. It's, it's numbers and ma- Like I could... If you gave me a pen, I could just write down how to do the math right now. It'd be very simple.

18. DLDavid Lieb3:38

    Yeah.

19. DVDominic Vitucci3:39

    The trick is for this particular workflow, right, is, is substantiating it. As we zoom out, this is the case for, like, why do accountants and, and advisors and consultants exist, period, right? Because it's rarely about what math you can do. It's about what you can prove happened. And so for the R&D tax credit, for example, it's all about contemporaneous documentation to say, "Hey, how do you know Tom spent this time doing these things?" "Oh, hey, well, we pulled, you know, Git issues. We pulled Jira tickets. We pulled..." And then, "Hey, we can prove he spent hours doing this." Easy for a software company. Somewhat more nuanced perhaps for a manufacturing company or an architecture company or an engineering company even that might have different levels of tracking of these things.

20. TBTom Blomfield4:18

    Is this the reason this field is now able to be automated, whereas five years ago even it wasn't? Like, is it that the computers can understand words and read documents in a way that unlocks those types of customers?

21. DVDominic Vitucci4:31

    Yeah. I think a large part of it is that, right? So when I, when I was at Grant Thornton for, for all those years, I brought up to them and said, "Hey, you know, there's, there's technology that can achieve these outcomes." Obviously, I, I was not, uh, thinking of what we have today, right? I, I was a young guy. I said, "But hey, at its absolute worst, we can automate certain aspects of this workflow, right? Certain like repetitive tasks." This was in like the kind of like RPA quasi res- renaissance, right? This was, you know, some of the companies I knew spent like millions of dollars on like Alteryx licenses. I'm really just dumping on a lot of companies today. Point being is that they bought all these RPA soft- I remember like last year at, at Grant Thornton, I'm laughing, they had this thing called Automation Anywhere, and they wanted everyone to become a certified Automation Anywhere expert.

22. DLDavid Lieb5:11

    [laughs]

23. DVDominic Vitucci5:12

    Which is just like the funniest thing in the world. And, and so you do this Automation Anywhere certificate, and like you just like play the video in the background, you get a certificate. And like the firm's like, "Guys, we did it." And it's like, "Yeah. Did we?"

24. DLDavid Lieb5:24

    Yeah.

25. DVDominic Vitucci5:24

    Right? And so in the past, you know, Dave, you, you, you could have automated a lot of tasks. I think now what we've seen in the last, you know, three... I mean, our batch time, you remember our batch started in January '23, which was like six weeks after ChatGPT came out.

26. TBTom Blomfield5:38

    Mm.

27. DVDominic Vitucci5:38

    Right? So at, at our company, at Onshore, I was, uh, just like fortuitously enrolled in like the GPT-3 beta in like early '21, and it was like, it was like pretty bad, right?

28. DLDavid Lieb5:50

    Yeah.

29. DVDominic Vitucci5:50

    It couldn't achieve the outcomes today, but I could see at the time like, hey, if this gets just like a little bit better, this is just as good, if arguably not more capable than like your junior and even your mid-level accounting staff. And so now, you know, having arrived at where we are today-With these incredibly capable frontier models. I mean, the fact of the matter is that today is the worst they'll ever be. And so we're already at a place where I, I personally believe that they're outperforming people at the junior, the mid-level, and, and certainly kind of touching, if not exceeding, the capability of some of, like, your most senior technical experts. I mean, and, and, and that's kind of the reality of where we're at.

30. DLDavid Lieb6:26

    Tell us how you got to, to Y Combinator in 2023. Like, what was the backstory?

Episode duration: 33:55