Why AI Adoption Is a People Problem for Accounting Firms
Download MP3Attention: This is a machine-generated transcript. As such, there may be spelling, grammar, and accuracy errors throughout. Thank you for your understanding!
Nikita Komarov: [00:00:00] I think in the next five years, if you learn how to orchestrate the genetic systems, you can work less but have a better quality of living and income than you have now.
Blake Oliver, CPA: [00:00:16] Are you an accountant with a continuing education requirement? You can earn free Nasba approved CPE for listening to this episode, just visit earmarked.app in your web browser. Take a short quiz and get your certificate. Hey everyone, and welcome back to the Earmark Podcast. I'm Blake Oliver. Every firm I talk to has AI tools by now, but what they don't have is a way to get 100 busy professionals to actually change how they work. My guest today says that's a real problem, and he's watched it from both sides of the table. Nikita Komarov started as an auditor at KPMG, went to UI and then spent years at McKinsey, where he learned that the hard part of AI isn't the technology, it's change management. Now he's the founder and CEO of Dobs AI, which builds AI for client services and public accounting firms, as well as corporate accounting teams. Nikita, welcome to the show.
Nikita Komarov: [00:01:12] Blake, thank you for having me here.
Blake Oliver, CPA: [00:01:15] Great to have you, Nikita. Eager to talk to you about everything that you've seen with AI over the last few years. I know that you have spent basically a decade helping clients transform. You were at KPMG, ey, McKinsey. When did you realize that the firm's doing that work had not transformed themselves yet?
Nikita Komarov: [00:01:35] It's an interesting point. You have all my career, I've been in professional services. And, uh, what I realized that we are helping our clients to be better. It can be if you talk about the county services, we are trying to put the financials in the like in a in a good way. If you talk about like transformations for like large enterprises or like B portfolio companies where they've done a lot. You are trying to improve your client's business. You help them to implement the technology. But right now we see that with the I professional services will be dramatically different than the next like 3 to 5 years. And what I realized we as a professional services, we are working the same way for the last like 20, 30, 40 years and nothing changes. So, um, I realized that there is nobody who is actually helping us to be better. And, um, I decided to be the one who can combine in the I, but also, um, know how painful it is to adopt like any dramatic changes in the business. Uh, we become the company who is helping to transform, uh, accounting firms and accounting teams in the large enterprise.
Blake Oliver, CPA: [00:02:53] So what do you think needs to change with this new technology? We've had tech shifts before, right? We got, uh, computers, first of all, back in the 60s and 70s, we got personal computers in the 80s and 90s, uh, then we got, you know, electronic spreadsheets. Uh, we got cloud technology, but like you said, professional services, it adopted that tech, but it pretty much stayed the same. The way we work didn't really change all that much. It got faster, but it was still the same thing. So what is changing now as a result of AI?
Nikita Komarov: [00:03:26] I would say we are going through not just the tech adoption, but the whole transformation of our business model historically. Uh, we have human experts who've done the work. You have a typical kind of pyramid of, uh, juniors managers like C seniors, directors, partners who has a different roles to come to accountability. And they drive this work right now with the, with the I it's going to be dramatically different. All the repetitive work, like in Excel, all analytics will be done by AI, and the role of accountants will be to become a trusted advisor for their client who can navigate the army of agent that works like on their behind. Um, and that's a dramatic shift. You need to think how you, uh, hire people like how you check the quality of the output, what are their new responsibilities, how to train them, how to price your services now because like time and materials doesn't work anymore with the eye. So how to adopt this across your organization? Um, there are more questions than answers. And when you are like leader in executive, like in accounting firm, you just don't know where to start because you have a lot of like ongoing activities you need to serve your current clients. You need to find new clients, and you don't have a mental capacity to think about like the huge adoption transformation of your core model.
Blake Oliver, CPA: [00:05:07] So I want to dig in to that with you. There's a lot there, right? Hiring, training time and materials. We've talked about the, the death of the the billable hour for a long time in accounting. Seems like it might actually be be happening. Maybe we can start there. Right? In terms of time and materials, that's how we've always priced our services in the past. Why doesn't that work anymore?
Nikita Komarov: [00:05:30] Because, um, so far you had a human expert who spent a certain amount of hours to provide the work. Right now, um, significant portion of this work will be done by AI that reduces, uh, if you charge like time and materials, it reduces your fees, but it's not fair, right? Because you are providing like Awaiting the better outcome, the better service to your client. That's why there is a shift to the fixed fee like value based approach. I would say the goal of the professional services of the accountants to leverage any available tools, but to give enough trust to their clients and enough value. One of the verticals where we're working is a financial due diligence. And if accountants like if your advisors can find the specific EBITDA adjustments or transactions that are not obvious, it can drive millions of changes in enterprise value in the deal. So it's much more important than thinking about like how many hours they charged. Um, and I think that's where the industry is going. That is still a kind of complicated conversation sometime, uh, for the accounting accounting executive to have with their clients. But That's inevitably what you should do.
Blake Oliver, CPA: [00:06:55] Financial due diligence is a really interesting area, and you've chosen to start there. And that's, you know, mergers and acquisitions, right? Type of work. Why is that a great area for AI? Why did you start there?
Nikita Komarov: [00:07:14] In all type of accounting advisory, I would say that financial due diligence is the most complicated space. Um mostly because of every transaction is unique. And um, if you compare it like with a different like tax services with the tax services, you probably have like the certain flows every year that you can describe in the hard logic and in the hard code. The challenge of the financial due diligence, because every deal is different in terms of the set of the documents that you receive. Type of the business. You should have something very flexible to help you automate it. Um, that's the reason why you want to leverage large language models. But the problem of large language models, they are non-deterministic. That's why, um, when you need to have something with 100% accuracy, sometimes it's like hard to rely on LLMs. So that's a big challenge there. The second thing is most of the firms, um they have a much higher revenue in their tax. So like insurance services. And um what I see from the conversations probably like 70 out of top 100 CPA firms, transaction advisory, financial due diligence fee firms, IT teams, they are underserved and they don't have the solution because most of the resources in focus was provided to the larger verticals in these businesses, and we decided this is a great challenge to take. If we can figure out the solution for the financial diligence, we can tackle any other financial services. And yeah, now we are here.
Blake Oliver, CPA: [00:09:10] That's one of the big challenges with due diligence and M&A is that every deal is different. I mean, even among deals that are similar, they can be very, very different. The companies are different. The transactions are different. And that's made it impossible to automate. So far, at least with the technology we had in the past. But you've figured out how to do this with LMS, with this probabilistic AI. Can you walk me through like, what that looks like? How does it actually work? How do you, how do you automate something that is like where every deal is so unique?
Nikita Komarov: [00:09:51] So we leverage large language models as a reasoning orchestrator brain, but to make it actually work to avoid hallucinations, we build a lot of like the hardness and like foundational platform around. We take this reasoning layer, but then we provide multiple other components. Um, I typically describe this as we are building a human body. Uh, we are taking, uh, one part of the brain from anthropic or OpenAI, like who invested billions, like hundreds of billions of dollars in building these models. But the rest of the body, we are building ourselves. So the second part of the brain is really important. It's about the different controls, um, evaluations. Um, I would say traceability, how you can Um, build a certain guardrails and the multiple kind of feedback loops to make sure that the output is correct. We are building eyes, nose and ears how to extract information from the source files with high accuracy to make sure that every number that we extracted is sortable. We reconcile every transactions with like subtotals with totals. So there is a lot of practices how to make sure that like extraction in the beginning is correct. Um, we are building like arms, uh, that knows how to build Excel models with the formulas, everything traceable.
Nikita Komarov: [00:11:29] Um, you have like a heartbeat of like our project management agents that understand, okay, like where is we are now like in the deal what's happening? So that's where you can have an illustrative example. But if you will go specifically how, um, We help to how we figured out to minimize any potential kind of hallucinations or provide enough accuracy for these specific services. That's a lot about orchestrating the agents. We are structuring them like as a teams, you have an agent, director, manager, analyst, and they can run like 50, 7000 cycles until they got to the outcome that you're comfortable to share with your client. Or we bake a lot of controls and checks that allows you to do the same type of verification as you would do with your analyst. There are also lots of the kind of calculation tools. So every analysis we do like, it's an Excel based, formula based traceable. That's why it's natively integrated in what you do now. And that's why we see excitement from the client and the high level of adoption.
Blake Oliver, CPA: [00:12:47] So what I'm hearing is, and I love this analogy that you're using, like of the human body. You've got agents that do specific parts of the due diligence. Like you said, you've got arms and, and, and eyes and a heart and you've got an agent that's orchestrating the others just like a human team, right? Would have a manager. And then you've got these analysts doing different tasks. So so that's how you get the deterministic output. The, the, the, that's how you know that what you're getting is correct in the outputs. Like if all this work is you've got basically, it sounds like agents that are checking each other's work. Is that the right way to think about it?
Nikita Komarov: [00:13:32] You can't avoid the hallucinations of one part of the brain of like large language models. They give you a lot of reasoning and flexibility, but they still hallucinate. That's why you need to build the. Like all other parts of the body that will kind of balance it. And when we run like 50, 7000 cycles and we like multiple agents checked each other, we are burning a lot of compute. But this allows you to provide 100% accuracy output that you are comfortable to share with your clients so that we are optimizing for. We are not. We are burning a lot of calories in tokens. But to make sure that the output we provide to the clients is accurate because in our business, if you have even one example where something is incorrect in the accounting world, like it's a no go.
Blake Oliver, CPA: [00:14:25] So that's how you get to that certainty. I and that's so important because we've got a lot of teams out there experimenting with AI and they're, you know, getting clawed or they're getting ChatGPT and they're trying to run these complicated workflows through a single chat. A single agent, if you will. And then they're they're running into issues where there's a hallucination. Maybe a number is wrong in the initial analysis, and then that just cascades through and causes everything else afterward to be unreliable. But you're fixing that with the way you've set this up.
Nikita Komarov: [00:15:08] Exactly. And, um, I think there is a huge difference between in, in efficiency tool and production ready solution that you can deploy for the client services and accounting. Um, what does that mean? I love like cloud code codex. I use it like at least ten hours a day. I have a separate kind of machine in my office that's working 247 and doing different set of tasks and setting up like agents, but it's very different from actually building a production ready solution. And I think there is a misunderstanding right now in the industry. Um, it sounds like now you can build any type of the technical product. Like much easier. Um, you have models that can write the code, but the actual challenge is an ability to have a kind of product mindset, understanding the problem, understanding the architecture, um, having the ability to maintain this technology every couple of weeks, you like have a new, um, kind of unique approach, new white papers you need to incorporate. So if you want to stay on top and really have a production ready system, you should have professionals who are doing this. And that's, um, I'm a big proponent of like experimenting with the eye and learning it. But if you think about actually client services, you need to be kind of very careful, like what we are putting like in front of them. Um, because I hear a lot of stories. Oh, I tried copilot. Like. Yeah, it's not working. Um, it's a analogy here. Um, you can buy like a very expensive, like great groceries in the shop, but if you don't know how to cook like you, you don't like your meal, right? Mhm. Um, that's the same here. Um, and even if we see that AI can do a lot of like code generation, um, if you don't know how to do this in a proper way, you will have more like kind of like tech debt than the actual value out of using these tools.
Blake Oliver, CPA: [00:17:24] So can you walk me through like the impact of this, uh, when a firm is using Dobbs AI to do, let's just, let's just use, you know, the due diligence as an example, right? Like what changes like, uh, How long does it take? What do they have to do? What's the difference between the old workflow and the new one?
Nikita Komarov: [00:17:48] There are several angles of it and we can touch some of them later. So I would say the big part is around working with the leadership team, understanding like their priorities and helping them, like with the orchestrating, like the broader transformation. There is a second piece around adoption and change management within the organization. And the third piece around the actual technical capabilities, as we were talking about the technical part. I'm happy to dive here now, and we can cover the rest of the pieces later. So if you talk about technology, um, you have a financial due diligence. Um, it historically took probably like 4 to 6 weeks. Uh, you receive some data rooms or access to the accounting systems and you need to produce the output. So we deploy the system that tailored specifically for organization that back test it on your previous deals. It incorporates your templates, your frameworks, uh, different, uh, kind of control checks that you have in your organization. So it's natively integrated, uh, in your organization with the principles, how you do the work, um, how you provide reporting, how you look at the adjustments. Um, and we allow to automate this, like right now, probably to extent of 60%, it means, um, we first get access to the data and it can be like accounting systems, data rooms, or you can just upload a bunch of the different files.
Nikita Komarov: [00:19:26] Then, um, we do a lot of like indexing and back and forth management with the clients because this takes a lot of time. Have we received all the files? Is anything missing? Can we actually run the analysis so our agents understand what was provided, and then help you to manage this back and forth with the client to make sure that you receive all the data. But your client is not kind of annoyed by your juniors who are asking the same thing for multiple times. Um, then we can do a data extraction and this can be like real world, real world, messy data that you receive like thousands of pages of PDF, like bank statements like complicated payroll structures like GL. So whatever you receive, we can digest it with a high level of accuracy and provide you traceability to every source. And after that, we are building the whole set of the analysis, the whole data book with a proof of cash expenses, net working capital a PER, aging vendor, customer concentration. It can be something very complicated as healthcare revenue waterfalls percent of completions for the construction.
Nikita Komarov: [00:20:34] Um, so the actual work that's currently done by like CPE team. Like we do fully by agents. And you don't need to touch the analysis at all. You just upload the documents and you receive the output. Um, it suggests we have a very, um, developed analytical framework that can run analysis and provide you this is a suggested EBITDA adjustments that's a different deviations. So you receive all these insights and the list of managing questions. You go to the call with management. You ask the questions. You can take the transcript of this conversation, put it back in the platform, and it will post you all the EBITDA adjustments, finalize the client ready data book and build the PowerPoint report. So all the flows, the only you need to do is be a client facing expert and provide some additional judgment where it's needed, but the rest will be done by A. So that's, that's how we work right now. And that's where like the industry is going. Um, accountants should be the trusted advisor who do the client hand-holding, but all the work on their back should be done by AI.
Blake Oliver, CPA: [00:21:45] So basically it's like, like your solution is becoming that team of analysts that take in all the data, analyze it, you know, propose all these adjustments, like you said, the EBITDA adjustments and, and then give me that, that, that presentation that I can then review.
Nikita Komarov: [00:22:03] Exactly. And the biggest difference, you don't need to kind of chat back and forth, like with a cloud code or ChatGPT, it just takes the input and get back to you in a couple hours and provide the full output. What typically takes you like 7 to 10 days? We're doing in two, three hours. Uh, most of our clients, they had a management meeting on the day ten. Right now they have this on the day three. This has accelerated the cycle. This is a much stronger value proposition to the private equity firm who they are serving and, uh. Yeah, uh, I'm very excited about it.
Blake Oliver, CPA: [00:22:43] So, I mean, theoretically, right? I could go do that myself with, like, clockwork. I mean, I can, I can create, uh, agents. I can have it spawn subagents, that sort of thing. Right? But like, like, what's the problem with that? Like, why wouldn't I just do this myself?
Nikita Komarov: [00:23:00] So this is a thing when I'm describing, okay, I can wipe code something for myself. And it's, uh, it takes like one weeks of effort to build something I can use in probably once it takes me like five weeks time to develop something that I can use myself on a regular basis to share something with the team. It will take me like 15 next time to share something like in open source, like FTX time, but actually have a production ready system that I can use an ongoing basis for the clients. It's like 500 x time. And this is the biggest challenge because you can get to a certain level of automation or help you with this, like some small tasks, but you will never get to the level of actually like the art of possible in the industry. Um, the problem of it, you don't have enough, um, kind of product expertise and understanding, like how to orchestrate AI agents. It's very kind of scarce, um, capability right now. And yeah, you have guys in OpenAI and anthropic who knows this like, like Stanford PhDs, but there are very few people in the market understand how to, uh, orchestrate complicated genetic systems. That's the one. The second you just don't have enough, um, mental capacity for this. I've been working in KPMG and McKinsey. They invested billions of dollars in different tech automations, but all of the products, products that I saw, you don't want to use them. Honestly, this is mostly because they're not doing their job perfectly. You have a certain partner, um, who've done a client study, uh, and then they wanted to codify this expertise and try to sell it somebody else or try to use this for internal efficiency.
Nikita Komarov: [00:25:03] You still have your kind of personnel and the goals of bringing additional revenue to the firm. Uh, you have certain metrics of like adoption, you can't invest like millions, tens of millions of dollars in just building one good product solution with dedicated capacity. You don't have a perspective on how other firms are doing this. That's what differentiates the product companies because they Because we have multiple clients who they help, like with the same problem again and again. They have a dedicated people who are thinking about this problem day and night, and they just experts. And when you try to do this yourself or like an internal function, either like you're amazing and you hired like 30 engineers from like OpenAI and you have like billions of budget for that, or you will stack it like 20%, like efficiency level. Um, because again, every couple of weeks something is improving and you need to be on track, uh, of it and continuously improve your system. If you don't do this, you just have a risk that you have like an AI in your organization who is like responsible for delivering like AI solution, but they have limited resources and they can't say that they will not deliver on like certain outputs because that's their job. Right? Um, the thing is that you will have like false confidence that you are using AI to the extent of possible because you have like somebody doing this internally, but you will not be looking what's actually available on the market and how others are solving this problem.
Blake Oliver, CPA: [00:26:48] I guess it's a lot like, uh, when we went through the cloud computing, uh, revolution and firms started deciding, what am I going to do? Am I going to move from my hosted server to cloud? And then if I'm going to do that, am I going to build my own cloud or am I going to use someone else's? And it seems like, uh, you know, like, what are the lessons from that change? I mean, you were, you were around when that was going on, right? Like just like me. Like, is there, are there any parallels here?
Nikita Komarov: [00:27:23] I would say so. Um, I think what we learned you need to focus on your core business. You can't do everything because then, uh, all the work will be like, kind of half baked. The focus of the accounting firms is to find clients, make sure that you deliver like a really good outcome for them, and they are happy that you are continuing building the kind of trusted relationships with them. And you also you learn about the different ways how to do different, like accounting or financial analysis, um, in the best way. So that's your knowledge, um, that's your IP. And then you also need to think how you train your people to make sure that they can progress and be the next, uh, the next wave of folks who actually owning the relationships with the clients. I would say, um, yeah, the same analogy with the kind of cloud compute somebody like tried it, but now they are like kind of three, four major cloud providers. Everybody is using it because it doesn't make sense to do this business yourself.
Blake Oliver, CPA: [00:28:41] Let's talk about training people. What needs to change with how we are training our people in public accounting, in corporate accounting, in accounting in general?
Nikita Komarov: [00:28:54] It's a good question. And I think it's relevant not only for accounting, not only for professional services, but for all professions that we have in the world. For example, software development, you don't need junior roles anymore because I can write the code, but you need junior people because you want them to progress and become like the next seniors. The same situation in accounting firms. You need junior people who will become the next partners. But how to teach them in this new environment. Um, you don't need to make people like crunch one excel like ten at night. Uh, but right now with the I, they should get an exposure to much more cases and understand the judgment. You can have an AI system that where like you see, um, the trace of the thinking, you can understand why is it happening in this way? You can challenge it. So I would try to hire the curious people who is open to adopt new technology, but also ready to have an ownership, um, on the output that they deliver. So, um, what we are doing, for example, when we are deploying solution, um, we are trying to provide a certain, um, kind of milestones where like junior people need to approve the output.
Nikita Komarov: [00:30:23] It will be good enough without it, but we are working with the leadership to input this kind of small checks that allow to train your people how to work with output. So for example, we start with a trial balance and reconciliation. And um, you should have like an owner from your team, probably like somebody like junior who should go to the output of the system, review it all and provide a check mark that they review it, they can work with additional AI agents and talk about the output and challenge it in some way, or try to understand why it happened in that way. But that's an interaction we are also trying to design. And I would say this is a more complicated, um, piece than just making like technology provide a good outcome because you also need to engineer this kind of social interactions and the way how the people collaborate and how they learn new things. Mhm.
Blake Oliver, CPA: [00:31:25] If those junior associates aren't doing the work in Excel. Assembling these workbooks anymore. How are they going to learn to review them? Because that's how we learned it. Yeah, right. We would do a tax return. We would manually create work papers or roll them forward in audit or in a, in a, in an M&A deal where, you know, like you said, grabbing all that information from the, the deal room and we're, we're assembling it and verifying it and all that. Like if we don't do that ourselves anymore, how do we learn to, to do it or to, to, to review it?
Nikita Komarov: [00:32:06] Uh, when you say, do you describe the certain way, how, how you've done this before? Uh, for the last like 30 years, you used in Excel to put like some data sources together and run analysis. Before that, you just like use paper, right? And do this everything like by hand. And this was a different way how you've done this. When you transition to the Excel. Have you lost your skill of doing like accounting and reconciliations that was previously done manually, like on paper? No, you just started to do this work differently. I think the unique skill that everybody need to learn is how to think systematically in a structured way about different components. And you don't need to touch Excel, but you can orchestrate agents and review the output by just connecting the different logic box. Okay, we reviewed, we received the data. What actually we have, what gaps do we have here? Why? Like I kind of structured, um, and indexed like these files in that way. What's still missing after that? Like we see the output of the analysis and we know what checks and controls we need to implement.
Nikita Komarov: [00:33:27] We probably need to check, like the certain judgment or like the potential judgment that the system like suggested to you. So now you can review not a one deal and spend it like kind of four weeks on only one deal. You can, uh, at the same time look at like 5 or 10 of them, but also have, uh, an eye for reasons you can review it and just, uh, work with it in a different way. Now you're describing like in plain English, what you want to change. You're asking the questions and you don't need to try to work on the, uh, kind of formulas in Excel because what, what are the formulas? That's a, I would say like finance, accounting, uh, code language, right? If you use like Python or like C++ in, in the engineering and computer science, uh, like in finance, you use in Excel to describe what you want in the certain parameters. Right now, you don't need this complexity. You need just to focus on the logic. But you can describe this in plain English.
Blake Oliver, CPA: [00:34:36] Do you think? Let me put it this way. I'm leading technology at a at a firm. Yeah. Let's let's do a let's do a hypothetical situation. And I have to make a choice. I have to choose whether I'm going to focus on automating the work that we're doing today, or I can use AI to offer new services. What should I do?
Nikita Komarov: [00:35:05] So, um, there is no just one answer for that. Um, that's why I described you that you are working with the firms in three ways. One, we are working with the leadership to understand their goals. The second is change management. And the third is actual technology. So now we are touching like. This first level, our goal not to be like just a tech vendor profile, like provider. Um, but at the same as I was consultant at McKinsey helping organizations to grow. So I start with a, having a workshop for the leadership team where we understand. What are the goals for your organization or your practice? Do you want to increase the throughput? Because you can't digest the current volumes. Do you want to launch a new service offering? Uh, to the same clients because you have access to their body? Or do you have actually. The problem with the, uh, like retention of the employees right now. We know that there is a shortage of CPAs. Um, and like the average age of these CPAs also grow. Um, so that's another challenge. So we're trying to understand how to like get additional revenue, how to improve your margins for the next like 18 to 24 months. And we're working with these goals. Um, my idea is not to bring you something like off the shelf that you can, um, just sanitize like solution, but actually understand what is the core priorities for your organization and then help you to drive this transformation to make sure that we kind of hit numbers. That's why, I mean, in short, answering your question, we need to understand, um, what is the biggest challenge? Um, and if you have like enough capacity right now to provide current services, yeah, probably we can start launching something new because this will be additional revenue generation channel, but potentially you need to have like a more capacity of your people so you can support these new services to.
Blake Oliver, CPA: [00:37:15] The thing though about, um, increasing throughput, right? Speeding up the work that we're doing now is that when we've done that in the past, it doesn't actually make us any less busy, right? We adopt this new technology, we adopt cloud, and we did that starting 20 years ago, 30 years ago. We adopt it and we're just as busy as we were before. So why would anything change? Right? When we when we when we do AI.
Nikita Komarov: [00:37:42] So I would say they were going to be only more work going forward. The GDP will grow because you will need to have more services, the higher quality of the services and the humans who want, uh, kind of to stay relevant and do work, they going to be like as busy, but they will, they just much, much more work, um, and, uh, be able to have better outcomes. I think there are like several waves of the AI adoption probably in the next 15 to 20 years, every everybody will have like a kind of base income and they don't need to work anymore. Uh, but before that, uh, if you want to have like the high quality of life than others, or you just like passionate about the work you do, um, you continue to work in the same way, but you can provide much, much more value to your clients. I think, uh, where accounting accountants where like this evolution is coming to, we are just becoming the trusted advisors to our clients because there is an interesting story that accountants are much more trusted than lawyers and, uh, management consultants. And I think that opens a huge opportunity for accountants. They can become a trusted advisor, a one stop shop, um, window into different type of the services. So they manage relations with clients, but then they pull in a different set of the agents to provide them certain services. Oh, this is a financial diligence. This is a valuation. This is like value creation plan. We can do. This is a, I don't know, designing KPI systems for the new portico. That's like an exit readiness. So you have an expert who understand the client's business, but all the work like on the behind done by different set of genetic tools that that's where we're getting up at the same time, I think in the next five years, if you learn how to orchestrate the genetic systems, you can work less, but have like a better quality of living and, um, income than you have now.
Blake Oliver, CPA: [00:40:05] But only if we, uh, can price it properly, right? Because if a portion of like a, a, a deal that took ten days now takes three. Are we billing for the ten days of work? Probably not if we're sticking with the time of materials approach. Right. So how are you advising firms that you work with to, to change? This is one of the biggest challenges when it comes to AI. Actually, I was at a roundtable discussion of technology leaders at top 100 firms earlier this year, and they were talking about how they're getting calls from partners, audit and tax partners who are loving the new tech, but are now seeing that their charge hours are dropping for their teams and are really worried about that. So what do we do? How do we how do we avoid having to pick up, you know, three times, four times, five times as much work in order to make the same amount of money?
Nikita Komarov: [00:41:07] That's what you need to focus on in this transformation, to think how you can charge clients a new way because, um, what clients are looking for from you is not to provide like the cheapest possible service, but actually they hire the finance professionals, accountants to get enough trust, enough confidence in their business that they are running right now will be in good shape or the business they are buying right now. It's like all the numbers. That's probably like the seller is like, provided they are supported and you know, uh, what you're buying like in the end. So they are buying confidence and they're buying support. Um, that's why I think, um, it's a wrong approach of trying to sell our, our time. Um, because then, uh, we will put our like ourselves on the treadmill and like, we will never stop. Um, for let's get back, for example, to, uh, Financial due diligence. Private equity hiring you not to, um, optimize like $15,000 on the quality of earnings, but, uh, their value is, uh, reducing couple million dollars, uh, from, uh, the enterprise value because you help them to find something or have like enough confidence that you like all the other deal is structured in the right way because they have much higher risks of like, um, kind of having certain like mistakes in the financials and ruining the whole business case. Then thinking about like how, how many hours like you charged. So I think that's the perspective. It's kind of sometimes it's scary and it's hard to communicate to your clients, but you need to get there, uh, if you want to stay relevant.
Blake Oliver, CPA: [00:43:11] So I'm sure we have listeners who are. Following along the whole spectrum of adoption here. We've got people who haven't started maybe just tinkering with it. We've got folks who are fully in on agentic AI solutions like yours. For those who haven't started or are just beginning the journey, what is one thing that you would recommend that they do before the end of the year?
Nikita Komarov: [00:43:41] I would say there are. Yeah, I can frame probably like several bullets and we can decide like on one that's that's most helpful. Um, first, uh, AI, it's not hype anymore. That's, that's a revolution that already started. You can't avoid it. And you need to start doing this today. The second piece is you need to have somebody who is responsible for the whole AI adoption. That's not easy. And that's your biggest differentiator going forward. So you need to make sure that there is somebody who is actually responsible for end to end adoption for your organization. Um, the third one, you need to invest in people because that's your biggest asset, the people who know how to use it, how to, um, work in a new way. Um, so just starting from the simple like automations or analysis or just trying to burn as many tokens as you can. It's a good start. So I know these are three bullets, not a one, but, uh, start today. Have somebody credible who is, uh, kind of helping you or like having an ownership with your organization and, uh, invest in change management.
Blake Oliver, CPA: [00:45:05] I think those are great points. Nikita, thank you so much for joining me today and sharing your insights about this new technology. Is there anything you'd like to add before we go? Anything that we haven't discussed? And, and also, could you let our listeners know where they can learn more about Dobbs?
Nikita Komarov: [00:45:25] I think one thing to add is, uh, we are living, um, an exciting time and the world is going to be dramatically different. Um, I think the biggest recommendation is just be open minded and, um, iterate and try new things. Be curious. I think that's the best approach to, to stay relevant and also, uh, to scale your organization in terms of us. Um, I'm happy to be a third partner for you for your organization. Uh, we have a lot of like the anecdotes and successful deployments, uh, in top 100 CPA firms. We're working with large enterprises, so just happy to share the stories, what's working, what's not. And our website is, uh, dobbs.ai. And you can reach out to me at, uh, by email at Nick and I are happy to be, uh, yeah.
Blake Oliver, CPA: [00:46:30] Nik@dobbs.ai, that's DOBS. And it's not just about, uh, financial due diligence. It's a whole lot more than that because like you've built this agentic AI system and it can do, it can do a lot, right? You've got AI auditors now that you can plug in, um, case studies about not just public accounting firms, but also corporate accounting teams.
Nikita Komarov: [00:46:57] They're working with large enterprises, like on Sarbanes-Oxley compliance, internal audit vendors, band audits. So like largest companies, we're working like $80 billion in revenue for accounting firms. We are helping with the different client services from the financial diligence tax provisions, audit, uh, valuations. So again, I would say the foundation of it is similar. And I'm happy just to share what's working, what's not working and, uh, be a third partner in this journey.
Blake Oliver, CPA: [00:47:28] And that's what I love about what you're doing is that it's not just a platform. It's not just something that I buy and I log into and I do the work. You're actually helping these enterprise teams, these firms, uh, figure out how to approach it and where to focus. Because I think that's one of the biggest challenges is there's so much you can do with this technology deciding where to start and where you're going to get the best return on your investment is really important.
Nikita Komarov: [00:48:00] Exactly. Can't agree more. I think that's a lesson from any tech adoption, from any transformation. And I saw a lot of, uh, kind of failing cases when I was at McKinsey. Just the lack of change management adoption. Uh, that's the core reason why most of the projects failed.
Blake Oliver, CPA: [00:48:19] And it's, it's interesting because I focus a lot on talking to small firms. And it's the same problem. We just describe it differently where we say, you know, if you automate a bad process, you're still going to get a bad result. And it feels like this is the same across organizations of all sizes. Yeah. So getting the human involved, the human expert is super important. And so that's why I'm so glad to have had you on the show. Nikita, thank you so much for sharing your knowledge and your expertise, and I hope to see you around here soon.
Nikita Komarov: [00:48:53] Thank you for watching. I really enjoyed our conversation. Yeah. Looking forward to reconnect.
Creators and Guests
