549 Meet The Founder Building AI for AUDITORS

Gang, we got a special one today. I recently had the opportunity of chatting with Danny Albertson of Agentive. He's building some really cool AI solutions for auditors, because auditors are people too. Auditors need things to write. I'm sorry, I was never an auditor. I'll be honest, I don't know how you do it, but bless people like Danny who are building cool stuff. This became a really interesting story of Danny, who's an accountant that became a software founder. He's got some interesting takes on, like, how AI can be deployed inside of the audit world, but also on kind of the larger software ecosystem right now and where the legacy players are maybe failing to deploy AI and how the entire, like, accounting AI vertical can start to learn a little more into AI. So without further ado, enjoy this one with Danny Albertson.

Danny, thanks for being here, man.

Pleasure. Thanks for having me, Jason.

So just jumping straight in, who are Agenta's, like, biggest fans right now? Like, who are your power users that love the product?

It's people like many of your viewers who like ChatGPT, who love AI and want to apply it to their work. We had a customer who's like, I tried to build a custom GPT to do this. I'm coming to you. Because it didn't really work. And we showed him our product and he transformed his idea that he had that he wanted a custom GPT to work on into an agent. We hold a work paper, but a workflow automation that was working in 30 minutes to automate 19 different, like, steps of an audit procedure that he does to test whether expenses comply with government regulation. And he was just blown away by how we have built what he wanted that he couldn't get custom GPTs to work for.

So for folks that are, like, familiar with, I know my ChatGPT, I know that interface, I know what an AI assistant looks like. Is it an AI assistant or does it look and feel different? Is it more sort of targeted?

Great question. So we are purpose built to help auditors request evidence from their clients and test evidence at the same time. So at the end of the day, when you're doing an audit, it's what accounts am I testing? What assertions am I testing? What evidence do I need to test those assertions? And then largely they have work papers that are already defined of, like, what procedures they need to do. And so we build those into our platform and based on the evidence you're requesting, what workflow needs to happen to complete your audit testing or get you 80% of the way there. And then we run that as soon as clients upload the data. And so it's purpose built for saying here's what I'm going to test, here's what I need to test and here's how I'm going to test it. And then we run that automatically and point back to citations with answers and documents.

Interesting. So every like every accountant has a version of the PBC list. Right. And auditors, that's especially ugly. But this has even been, I would say the most interesting development in the tax space is like more intelligent PBC gathering information, resolving those items as somebody uploads them. Bookkeepers have their own version of that. So I imagine it's doing, it's helping you build the list, it's identifying, okay, this resolves these items that we need to gather. But then how do you, how do you bridge that to like a specific to do that's then associated with a BBC item.

So an example would be first of all, we don't, we're not an audit methodology, we're not your binder today. So like, like caseware tools like that, you need to do your audit planning and then come once you decide what you're going to do, you come to agentive. Let's say you're auditing a nonprofit and you need to test their revenue. You're going to either roll forward from last year or apply our nonprofit revenue template. And then we're going to apply the asc. Like how you audit nonprofit revenue automatically as soon as they upload their nonprofit evidence. You can upload a sample and say here's the transactions that I'm requesting from the client. Let's say 25 revenue transactions and I'm looking for the grant contract and we can match it back to the, the sample number that, that you've requested. Clients can upload anyways. We have some great user interfaces where clients can actually upload and say this file relates to this selection. Instead of disorganized uploading, which is all of our competitors do. But we, we take that and then we say okay, what are the like restrictions on this? Was this the right year? What like everything an auditor would normally do to test if it was counted for correctly, we have AI do that and point them to the answers within 60 seconds of client upload. And so based on the document types that you're requesting, we have pre built instructions or we pre build it or you pre build it that run on upload by the client.

So is the starting point like the trial balance?

Yes.

And that's ultimately it's then comparing what it sees in the trial balance against the documentation that's being provided.

Well, it's not necessarily trial balance like it might be the general ledger you pull up or a subledger. Auditors will take a sampling of transactions using their preferred like statistical method. We can do our sampling in our tool, monitoring and sampling. For example, based on your risk tolerance and your confidence. You pull your sample, put that in or you have the full population, put that into agentive, pull your sample and then that turns into a request list. But then you decide how you're going to test the documents they provide and put that into agentive.

Interesting. So it builds the request list based on the GL that you've given it.

Yes. Auditors like the process for them today is they know like based on the, the risk levels in the accounts they're testing. They, in their audit methodology software, they determine like what, what the risk levels are. They can put that into our tool. We'll automatically pull the table for them. But they need to assess the risk level independently through professional judgment with their audit methodology. Once they've assessed the risk level, they can come into agentive and then pull with, with like you know, 95% confidence, you know, 5% risk level depending on or anyways they can select their risk levels and there's a standard monetary unit sampling algorithm or, or haphazard sampling that we can just help them do. Instead of them having to go through Excel and pick which ones they're going to test randomly, we can randomly sample for them and give them statistical confidence and explanation of how we did that. But that's pretty, that's technology everyone has. But we've made it to where as soon as you do that, it puts it straight into the PBC list and clients can see it in our product instead of having to download in Excel.

You touched on like where Agentive sits alongside like core audit management platforms. I want to get back to that because I think that's a huge conversation. But in the beginning of Agentive, your background, you are an accountant, right? Like you're not a tech founder in the traditional sense.

I would say I am, I'm both. I'm an accountant.

I guess you can be both, right?

Yeah.

Did you start an audit? Like, I mean that was I think my least favorite class I ever took. And we like to joke on the channel that auditors are people too and they need software also. But like how did you stumble into this?

So I'm going to give you a little bit of Backstory. I, growing up had many mentors who were accountants. My Scout leader growing up was a senior manager at PwC. When I was doing my Eagle Scout project, he had me in his office downtown. And I grew up pretty poor. And he's like, you can have a life like this. His secretary was helping me write up my like Eagle Scout final paperwork and like, she was putting white out on, on things. We were anyways, we were collaborating. I'm like, I can have life like, this is amazing. My bishop in my church was the managing partner at Grant Thornton. And I was like, I, I want to be an accountant. This is what I want. Like, I want this life. I want to be able to serve others, have a great family. Like, I, I want stability. Like, forget technology, I love technology. As a kid, built websites and things, like, for fun. But I was like, I want to be an accountant. Like, this is the life I want. And so I studied accounting at Brigham Young University. And then I actually worked doing like, worked at a startup, but then left that to take a job as a. Like, I spent my first summer after studying accounting, like for my junior year doing like bookkeeping, CFO type work, taxes, financial planning, and then came back and then decided to do audit at Ernst and Young and started my career as an auditor at Ernst and Young. But throughout my first year, as I got more understanding that like, the majority of my time was going to be spent doing substantive testing or test of details, I was like, this is really it. And I kept trying to figure out how I can improve those processes. And they're like, you can't just do it the way we tell you to and get it done fast at a high quality and just keep going. And you can do planning and strategy later in your career, but right now you just have to get this work done. And I was like, I can't live like this. I need to fix these things. And I. Long story short, I was able to transfer at UI to their digital transformation like, strategy practice where I worked with engineers and designers to build technology. And so I went from doing test of details type work to building the virtual line product at Universal Studios in Orlando at Volcano Bay, working with engineers and designers. And I was still at Emergency United States at EY for seven years. Left as a manager, but had a great time. Loved my time at the firm, but then joined Facebook at the time to work on augmented reality glasses. So I think I can check the box of like, I have been in Silicon Valley, been in traditional tech roles, care about design and quality of a Product, but also spent my time as an auditor with an accounting degree. But my two passions now are one in agentive, which I'm really happy to be working on in both these spaces.

What is your most. At least in my experience, most folks that come the traditional venture backed path, they are not accountants first and foremost. They are tech founders is how they would identify. And then they come into accounting and say, let's find a problem to solve. What's your take on like, those two different approaches? Because there's you saying, I couldn't handle how inefficient that was. Like, there's a lot of people that heard that and they're like, yes. And they may even hear that and be like, does that mean I'm supposed to be doing tech? Like, like Danny's doing? What would you say to those people, like, versus the traditional path of someone that's not coming from the space?

So here's what Silicon Valley would tell you. There are no rules. You can do whatever you want, but at the end of the day, the market's gonna hire or fire you. And so you should figure out what you're good at and what you enjoy, and you should go do that. There are some people who like improving processes. There are some people who just like using very well designed processes. And you should do what your heart draws you to. There are no rules. You can do whatever you want, but the market's gonna hire or fire you. So you should find something the market's gonna hire you at. Um, but you should experiment, take risk. If you're young, like, or, or not. If you're young, like, you should figure out what works. Like, like you starting the, the podcast and what you've done. Like the market's hired you doing this. It's. Not everyone can do it. And that's okay. But did I answer your question?

Yeah.

Okay, great. Yeah.

And it's like so many things we think about where it's like the worst case scenario. You go back to being an accountant too, and you're like, you're still very, very in demand.

Yeah. Or I've learned since then that a lot of firms have entire groups of teams that are just in charge of like, managing the tech at audit firms and deciding whether to buy or deciding who to collaborate with and who to partner with. And there's, if you're interested, there's a realm of like, being, I want to be an engineer or a designer or product manager, or I want to be just an accountant. But there's, there's all sorts of opportunities in between both of those.

Okay, back to the audit software ecosystem. Audit is, at least from the outside, it seems a little like tax, where you've got these, like, monolith, humongous systems that everybody's running on that are all legacy systems and they don't talk with anything else. And so the biggest limitation of AI tools thus far, one of the big headaches is so many of these solutions require you to take a bunch of things run over here to the AI tool, put them into that thing, it does that specific job, and then you gotta take it back to whatever my core tool is, where I keep everything, and it's a big old headache. And then buyers, accountants end up stuck between, oh, yeah, I see that cool thing that you're doing and I want it. But I'm going to wait until my platform that I use today rolls out the same thing. Because they told me they're working on it. Right, of course.

How do we solve for that?

Like, are you struggling with that right now? The fact that it can't be like the everything today? And are you like, do you have to fight that with folks to get them to come over here and use us for part of the process?

If I had unlimited resources, which no one does, because you can only do a few things really well at a time. And so at the end of the day, like, there's a long list of things we want to build. We're, we're prioritizing what's going to help customers the most. And the way we're, we're prioritizing is what takes the bulk majority of billable hours, and what's the bulk majority of pain for an auditor? Like, what's causing auditors to quit their jobs? Why did I quit my, like, why did I leave audit? Why are your people leaving? Why are your clients unhappy? We're prioritizing based on, like, how can we make the experience enjoyable? And honestly, it's like, if you take a look at the first couple years of an auditor's job, they're spending 50, 60, 80% of their time doing, like, test details, which is largely like, hey, based on this, this is how these transactions were recorded in the general ledger. Do the documents agree with how they were accounted for? And this is the part most people don't like. And so we're focusing and it's an unsolved problem. It's a problem that hasn't been solved yet. And you've had tools that say, hey, here's this exact match ocr, like word, or like, Text that you recognize in this document. Here's this text. Let's find this text in this document. But never before have we had a technology where you can say, what are the restrictions in this contract for this grant revenue? Please tell me. And it will answer back with sources and citations. And actually, like, do work that you would have done manually for you on your behalf, but at scale, 50 selections at a time. So, like, you can do work on one and have it replicated over. Like you can 50x your work with using Agentive, which is really fun, or using AI, but then you still have to review it. You need to exercise professional judgment. But we've just seen an incredible. Sorry, what was the exact question you asked me?

I think it was like, how you manage. Is it like this side process? You got the core system still?

Oh, yes, yes. So sorry, great question. We're solving problems that haven't solved yet. So the alternative to using AI to help you actually do your audit test is pulling up the document yourself and manually doing it or using a tool that just doesn't do everything. We can do the. In the market today, you have some tools that can use AI to ask questions. You have some tools that can do, like, document matching. But one of the big differentiators we have is you can ask a question and get an answer, reference it in your next question, and then further down the line in other documents, reference answers that you got up the stream across the evidence that you vouched or traced, vouch or traced back to evidence in the gl. And so we've built like a tool that can help you do it end to end. And we're solving a problem that hasn't been solved before in unique ways. And so it really isn't a problem competing with incumbents. But people do need to come and say, here's the evidence, like, and they come to Agentive. They say, here's the account I'm testing, here's the assertions I'm testing, here's the evidence I need, and here's like the. Essentially, it's the same workbook or the same work paper and procedures that they have laid out in an Excel template. We take that, put that into Agentive, and tie that to a client request and make it work from there. And so it is a little bit of like, let's take this from Excel, let's take this from Excel, the definitions that are in Excel. Let's take our audit planning, which was done inside of your audit methodology, and taking those two contexts together. But the time you save. Putting it into agentive and running it is drastically more than you would save by not putting the context in if that makes sense, because you can multiply your efforts by 50 and then you can save that as a template and reuse it next year. Or if you have a colleague that's not super tech friendly, if you set up the request list, put the automation there, it's going to run for them automatically and they can just get the benefit without having to set it up on their own. And so you can really 10x your your work by giving it the context that is split up across Excel work paper procedures or across the the audit methodology where you come up with your plan and risk levels.

What is your level of confidence in incumbent software platforms ever figuring this out?

So I respect my competitors, I absolutely do. But I think they'll figure it out by looking at what we did and following behind us by a couple of years. Yeah, and that's a bet that I've heavily invested in and I'm one I'm doubling.

If your firm still runs on spreadsheets, sticky notes or half a dozen different apps to manage work, there is a better way Enter this Episode Sponsor Financial Sense it is the all in one platform for accounting firms to manage everything workflows, client requests, capacity, billing and team collabs all in one place. Automate recurring work, track deadlines and get visibility into what your team's working on. Built specifically for firm owners who want less chaos and more control. That's actually an important detail if you're the type of firm owner that wants more chaos, or if you want less control.

This one? This one's actually not for you.

Now, this December 12th to 13th, Financial Sense is hosting WorkflowCon, a free virtual conference built around the theme Clear the chaos, Find your calm and me. I'll be joining for a live Q and A on the subject of smarter client comms. But comms of all kind using AI alongside a bunch of other firm leaders, it's pretty like Rockstar. List of speakers they got Register for Workflow Con via the link below. Start your free 14 day trial of financial cents using the link below with coupon code Jason10 Grab all that stuff down in the show. Notes Tax peeps Listen up. Do you have a tax workflow solution yet? I'm not talking about Grandpappy's little auto Scan, the awful intake OCR thing that comes with our dumpy old tax software. That's not what I'm talking about. There's a new crop of AI tax workflow tools that are doing some really, really cool things. One of the leaders here is this episode sponsor Sorban. They're one of the first to this. They do a really, really good version of automated 1040 intake. A tremendous amount of the time spent during tax season is just chasing people for docs and like building out the request lists of all the things that we need. When the client emails and they're like what do you need from me? And you go back to the prior year file and you look at what you already have and you just got to do all that stuff. Sorban man, it wipes out all of that complete no brainer tax workflow. Tools like this is not an exaggeration. Are honestly like a must buy right now for tax firms. I'd say they have been for the last like two years and they just keep getting better. Sorben also they're working on pushing the data to your tax software. Really hard problem to solve. They're working on it. They're also working on automated delivery. So ingest the tax return and then gives the client like a super smooth.

Signing and sort of to do list.

Experience once the return is actually done. Really good product. Know a lot of happy users. That's Sorban. Check out the link down in the show notes. That's Sorban.

If you do more than a couple.

Hundred ten forties a year, definitely check out Sorban link down in the show notes. Hey, real quick attention. If you have clients that do project work, do ugly stuff like manufacturing, like inventory, construction stuff, anything where you got hairy itemized bills that have to go to specific destinations, when it comes to bill pay, there is a new sheriff in town. It is this episode Sponsor makers up. Even if you don't do bill pay, this is something that is worth showing your clients. Actually we've done a number of videos on my channel. You can start sharing my videos with your clients. What fun Gang, we are light years beyond what the old guard like bill ever did with automated bill extraction and get like getting these things through the appropriate approvals process which is oftentimes much more complex than just is this bill.

Over $10,000 or under $10,000 man.

I worked with a church that had this like multi level like just this crazy hairy combination of approval approvals that was dependent on like a mountain department all these different things and they had this physical thing that had to get passed around for every single total nightmare. You probably got a bunch of clients with citrus situations like this and you can absolutely be the hero with maker's hub. It is seriously good. Worth checking out. Learn more about makers up at the link down in the show.

Notes.

Okay, let's go dark here for a minute. Yeah, there's a really bummer timeline. Let's just, let's call out an example. I'm sure you're familiar with Materia. Materia got acquired by Thompson Reuters.

Yep. There was another Facebook employee that started Materia. I haven't met them, but great respect for them. Yeah.

How.

And, and like you could be excited about that. You could be bummed by that. There's a lot of examples. Thompson Roy is a great example. A lot of companies that have been acquired by Thompson Roadies and that's, that's kind of the end of this. The party ends at that point. Right. Because you get a bunch of big ambitious people, they want to build a thing and then you plug them into a big legacy machine and they're like, yeah, I'll work my own out, but after that I'm going to go build another thing. Totally understandable.

Are we stuck in a place right.

Now where the only people, like the only path forward for these legacy platforms is acquiring the companies that are doing amazing things and maybe they plug that into their ecosystem somehow but ultimately like that kills what made it special long term. Right. So like it bums me out to think about this, but unless somebody is going down the path of like we're going to build a full suite ourselves, right. Like we don't need any of that. Meanwhile they've all got the war chest, the x100 million of dollars that we set aside just to tick the box to tell investors, yeah, we're investing in AI and so they're looking at acquisitions to do this because they don't how to build it in house. Is that just a sucky timeline like it like how do we not end up in a place where you just have the legacy providers doing all these roll ups and, and I don't know what's your take on that?

So first and foremost, like there are no rules. Anyone can do anything. You just have to bend the world to your will. And so I respect legacy competitors like or legacy companies. It is really hard. Like you generally have to almost throw away everything you've built and start from scratch in order to really design a system to use AI well. And so I that it's very hard to do. So you should choose when you're doing startup you should choose your competitors like strategically and like betting against like competing against PE companies or companies that existed for a long time and betting that like they have to keep and maintain hundreds of products that they've already built. It's very hard to compete. And so that's often why they buy startups that can think fresh, think originally. And there are founders that want to create enough wealth for themselves and then they want to just build a product, sell it and have someone else scale it because those are two different skills. And there are some founders who want to build and scale a product and make it a billion dollar company or hundreds of millions of dol. And at the end of the day the market hires and fires you. And so you've got to figure out like can you keep growing? Can you keep solving real problems? And those are hard decisions to make. I don't know when what went through other people's heads when they've decided to sell, but I am just focused on solving this problem. I want my, I have a sister in law that just started an audit profession and I'm like, I want her to enjoy her job. I have a cousin who just started in the audit space and I want him to enjoy his job. I have friends who are clients that are just like, oh, I have to deal with an auditor again. And like I want them to enjoy interacting with their auditors. And so I've had the opportunity to sell agentive. I thought this through and I have to believe that like I just want to solve these problems and kind of push away like the monetary part for now. But that is kind of how corporate America works though is buying startups and that's where most liquidation comes from or you either go public or you sell to someone else. It's, that's how venture gets paid back, is by selling eventually. But I like, I think there will be some startups that will make it through and I, I plan on it being us and I think there will be some that will make it through to, to compete with the case wares and the Thomson Reuters and the Walters Kluivers of the world.

So you've had the opportunity to sell agentive.

Yeah. Had an unsolicited offer from people we really respect. It didn't make sense at the time because we believe in what we're building and we think we're, we're making about it ourselves.

Is that a question of what the number is or is that a question of the mission of agentive not being fulfilled yet?

The mission of agentive not being fulfilled yet. And are you familiar with Palmer Luckey who started Oculus and Palmer Luckey sold Oculus to Mark Zuckerberg, because Mark Zuckerberg told him that he would invest a billion dollars a year in making augmented reality and virtual reality the next computing platform. Microsoft and Apple were both pitching at the time to sell it. But he was like, Mark Zuckerberg told Palmer Lecke that if you sell to them, the Oculus is going to be an accessory to the MacBook or an accessory to the iPhone or an accessory to the Xbox. And he's like, we don't have an existing computing platform. We want to turn it into the next computing platform, and we're going to invest a billion dollars a year. And he's like, come build your vision with us with a billion dollars a year. That's why Palmer Leckie sold. I really respect him because the question was, how can he accelerate his vision towards building the next computing platform? And Meta just released glasses with displays built in that have AI Anyways. I used to work on that when I was at Meta. I'm really bullish on that vision coming to reality. But if, like, we care about accomplishing our mission, and I'm not planning on selling, but if someone were to come to me and say, we're spending a billion dollars a year to make auditors enjoy auditing more and audit clients to enjoy being audit clients more, and it was a mutually aligned fit, like, I would definitely consider it, but I'd want to figure out, like, what is the best pass. What is the best path towards accomplishing our vision and our mission. But I would not immediately shut down that conversation. It'd be pretty cool to have a billion dollars a year to fix auditing.

It would.

Yeah.

But he didn't finish the story.

Oh, and so of what?

Palmer Luckey, what ultimately happened? Palmer Leckie there.

Yeah. So he sold to Facebook and worked there for years. Now he did leave. Now they're back on friendly terms and they're collaborating on augmented reality glasses for the military together. But they're. Yeah. Mark and Palmer Lucky are friends. But that story has always resonated with me about Palmer being mission first and vision first.

Yeah, he kind of. He kind of got thrown out. He did get thrown out very unceremoniously. He did follow the vision, and then it was like, oh, sorry.

Yeah, yeah. And he's said some things about that in the past, like, year or so, and just like, look, Mark had to do what it was to, like, protect the company anyways. They put that behind them. But, yeah, that was not pretty. And the reality is, most founders that do sell don't stay forever. At the company that they sell to.

This is such a big question in the accounting tax base right now. And there's so many versions of this that we're talking about where if you're like running a venture backed company and the goal is ultimately an exit, who are the people that you can even exit to right now? Where accountants who like that, that's who I work for as accountants and like that is who I advocate for. What's the outcome that account that is good for accountants. Like I have a heart. That's what I struggle to visualize right now. Where it ultimately raises the bar for the tools that we have at our disposal and like the cool innovative stuff that we can do. Because we have so many examples of really cool innovative companies where it's gone down kind of a bummer path post acquisition.

So that's an interesting way to look at it. So once again I don't, I've never used materia. I don't really know like anyways never used it, never met the founders. Nothing but respect. But like do you consider a bummer path after they've sold to Thompson Reuters?

I would say there's definitely a history of cool products.

Yeah.

That have in Thompson Reuters is a great example. I go down the list of cool things from, from taxcaddy to now. Latest is Safe Send. But there's definitely a track record of cool products going that way.

They just kind of stall out at.

That point and the team ultimately leaves and that's sort of the end of that innovation.

It is a fundamentally different skill set to build a product from scratch and innovate and then scale the product. What big companies are great at is scaling and getting distribution. Absolutely respect like that they have that skill set. And so the question is like was the vision done of those founders and what did they have left plan to build when they, when they sold? And I don't, I guess you don't know that answer. But the companies, yeah, you're right. Like in the, in the accounting and auditing space there are few like acquisitions. There are some interesting private equity funds or other types of investors that are looking to like bundle startups together and like bring them to market as a new solution or as a new, as a new market. But sorry as a new solution to the market. But they're like. There are relatively few acquirers and that does make it weird. But I just don't even think about who my potential acquirers are. I'm not planning an exit. I'm just planning on solving the problems for our Customers maybe a little irrationally, but we're, we're heart first. Like, we're, we're building a product with the heart first, not a, like, exit plan first.

All right, let's get nerdy. We've geeked a little bit on. We've, we've geeked a little bit in social media interactions before. What are, what are some examples of where AI has gotten good enough? Like just in the last six months. Like, what is some new sort of emergent capabilities that you've been able to apply, apply to the product and have you super excited.

So Claude Sonet 4, 5 just dropped this week. What it's great at is having a. So most of the people listening to this podcast today, what they're familiar with AI is you ask a question, you get back a response, you're done. Yeah, maybe you've clicked the reasoning model and turned it on and it thinks and you see it thinking and then spitting out its response. But in programming, I've written like 120,000 lines of code in the past year. And AIs helped me go from someone who cares about design, knows a little bit about how, like, how something should look, to someone who can for the most, my, my co founders clean up my code. But, but anyways, as someone who can go in and actually make core contributions to the product, it's been a phenomenal helper. So like the same way I can now write 120,000 lines of code while being a CEO and doing sales onboarding, taking care of customers. But I can anyways, I can do 10 times more now because of it. But what I can do really well is in the past six months is getting better and better at what we call what's called function calling. But like, AI can look at what tools you've made available to it, it can decide which ones to do. Well, first of all, it can have a goal, come up with a checklist or like a plan of what to do and call different functions that are capabilities in our product that it's able to decide how to use. It can evaluate the output, decide if it's done, and decide if it keeps needing to take more turns. So it can independently go and work on a problem for 30 minutes and then come back to you when it's done. And the ability for it to come back with the right answer has drastically gone up in the past six months. And I suspect over the next six months it's going to keep getting better. And that's what I'm most excited about. And so fundamentally Today's people's how they interact with AI Today is one message. Get back one response But AI is going to be able to take more work off your plate. I came up with the term agentive or I didn't come up with it. I went to grad school at Northwestern for product sign and management. I dropped out after a semester because it wasn't a good fit for me. But my entrance essay was on agent of design and how technology should do work on your behalf without you present instead of you having to be there. And so that's the general premise and I, I wrote that that was going to be the next trend over the next 10 years which is, which is very much coming true. I just didn't know that AI would be the thing to do it. But that's what's changed the most in the past six months is the ability for AI to do work on people's half without constant human in the loop interaction. It just understanding the goal and eventually getting to the point where it accomplishes it at a much higher quality level. In the past six months.

Yeah, I think this was even a big talking point with, with rpa the notion that like you can have all this stuff working in the background. And it's really interesting from a product design standpoint, what does it look like to have AI do longer time horizon tasks for you? Because right now someone will use GPT5 thinking and chatgpt and be like this is slow man. I don't want to wait 60 seconds for an answer. Like through another lens. It's like that's the power of it is there isn't really a limitation besides cost to how long it could work on something.

That's why we do all this work at client upload. Instead of having the auditor have to click run, we do it at client upload. So it can like we, we do it. Most things are finished in the 60 seconds. But we see a world where and we're actually building this where AI can like right now we have it execute each step and like think logically but we're building it now that it can say was this the right out? Like it can check and review its own work and decide if it needs to change course but like still follow the actual like audit path. But it can synthesize and evaluate its own work along the way. Because one of the biggest problems with AI today is if you make like it's the classic. Like if you're one degree off when you're sailing, you end up hundreds of miles away right Depending on how far you're going. So the like, it's having AI review and evaluate its own work along the way and decide if something does look suspicious. It can now, instead of saying it looks suspicious, it can say, why does it look suspicious? It can go and decide, you know what? I'm checking three different types of evidence like an invoice, a bill of lading, and a bank statement. I'm going to go back and look at the invoice because this looks weird on the bank statement. It can now decide, I want to go back and look at this or I want to now look at this document. It can jump around. Instead of following one predetermined path, it can jump around on its own and try to figure out what went wrong along the way, if that makes sense. And that's the next level of, of AI and auditing is when something doesn't quite go right, it has the tools and resources to figure it out. Or it can go search the web and see why it's not coming back as expected or what next steps are. Search your internal documentation for guides on next steps as opposed to just stopping and saying these numbers disagree or these terms disagree.

Yeah, you talked about tools, you know, function calling being the technical framing, like the. These sort of abilities that you give it access to to do more specialized jobs. If we think about the place of a generalist tool like ChatGPT, which is kind of a Swiss army knife, to do a bunch of versus a point solution, like agent of. A lot of people right now I think are stuck on where do I start. Because like a point solution, like, it may not help me fix my dishwasher, but it has like a big difference is the tools that it has access to and the specialized tools for folks.

Right now grappling with, like, how do.

I get my team up to speed.

With AI and make sure they're leaning.

Into it as much as possible. Is there like a Maslow's hierarchy of where we make tech investments? Like, do you start with the Swiss army knife to build the foundational skills and then go to the point solution.

When you need it?

Is it the other way around?

So, great question. I think at the end of the day, the people who would be good with like points or the people who'd be good at general solutions probably already using them. Like they're the early adopters. They're people who can. They have to figure out how to apply it on their own. Like, they have to figure out how to apply it on their own. They have to innovate they have to like, be creative. But the people who, if you're, I mean, at the end of the day, if you're doing a workflow, like, you should make your decisions based on return on investment. In my opinion. You can have some like, innovation, budget and like, that's where I think generalist tools should go. But when it comes to like the auditing workflow, like auditors spend a significant amount of their time like testing audit evidence. And we, one of the ways that we sell our product right now is we believe the product should like win their trust. Not like you see a demo and then signing your contract after trying the product. So we have a two and a half hour, like, so we do a discovery call, a demo. Then we invite people, like, we'll say, give us five people, come try the product for two and a half hours and then we take them through different use cases. Because there's one thing seeing a demo, there's another thing them actually typing it out and saying, like, you know, asking these questions, doing this audit procedure. And their eyes just open up when they see the results themselves. Like, and they'll go off script pretty quickly. They'll go, I'll try this on my own, I'll add this question or I'll do this. And they're just amazed when they see the output. And so to me, like, we see some people, and I actually regularly have to go into those and say, is everyone on the same spot? Has everyone at this point in the test? Or is everyone at this, at this point in the training? And we'll have someone say, I got lost, I'm not sure where I am, I'll go into a breakout room with them. And then we kind of have them present and go at, at their pace. But we take people who really struggle at the beginning and by the end they're running everything on their own independently and having a ton of fun. And so I think the best way to teach AI skills to people is in the context of what work they do most of the time. And so if you can put it in their workflow, like our adoption rates I think are better than our competitors that are like Excel plugins or other tools that are elsewhere. Because if we link it to the request, it's like it's naturally to say, okay, I'm requesting this audit evidence. It's like, okay, how do these two audit evidences relate to each other? What questions do you need AI to answer about each one of these different types of documents? If we build it into the workflow we can make it really simple to use and drive adoption because it's a part of the process they see every day. And so that's one of the benefits to building it into the request and test workflow. And it removes a lot of the friction of having to say, is it worth it using this tool? I've got to go somewhere else, download this from here and upload it over there and then try this. Am I going to get the right answer? And people have doubts that. And we don't have as much of that because we put it into the request to test workflow.

Attention users of Thomson Reuters Virtual office. Is that you? Listen up, pay attention. This episode sponsor Barretto. They'll host your Thompson Reuters stuff. You tired of, of slowness, of stuff breaking. Tired of sharing your cloud with other people?

Gross.

I don't want anybody else on my cloud. I don't know what that guy's doing. Get faster hosting from Verito. But check this out. It's actually so much cheaper to do this through Verito that for the cost of virtual offices, you can not only host all of your TR apps with Barretto, but also use their fully managed IT service for less than the cost of your virtual office. That's right. Verito does hosting, but they also do full blown managed it. And the cost of both of those things for a better hosting service and managed it is less than the cost of Thompson Reuters hosting service. I don't, and I, I don't, I don't know what to tell you, man. I first got on Verito when somebody asked me like, what's a better hosting service I should go use? I went to my online community, realized almost 700 firms in there now. I literally typed in Thomson Reuters hosting. Somebody had asked the same question and like eight people in a row had commented that they absolutely loved Verdo and.

Had a great experience.

That's honestly, that's why we started talking about Verito and now they're sponsoring our stuff.

Woo.

Okay, if you're on VirtualOffice, you know what to do. Check out the link to Verito down in the show notes. You know what's the worst, man? When you hire somebody and they start working for you and you're like, I thought this person knew stuff. The number of times I've been hoodwinked by somebody that interviews well and they have umpteen years of experience and you pull them in and you're like, what?

You did?

You don't.

What?

You don't even know your stuff, man. This isn't going to work. Let me tell you what one of this episode's sponsors, Cloud Accountant Staffing, is doing about that. So Cloud Accountant Staffing, they help accounting firms find talent or around the world, be it the Philippines, South Africa, Latin America. And they have an online portal where you can go out and see folks that they have recruited, schedule interviews with those people directly on their platform. And one of the things now that every single one of those people have is a collection of assessments. I'm looking at one right here.

Assessments for business tax, for individual tax.

For critical thinking, for understanding instructions, their personality type.

And so many of these things like.

Individual tax, this person I'm looking at, they're in the 93rd percentile. And so cloud account staffing, they have a huge volume of folks that they have go through these assessments and they hand pick the highest performers and they serve them up for you to make your wildest dreams come true.

Cloud Account Staffing's backstory there actually ran.

An accounting firm, hired some offshore folks.

Had an amazing experience, and then just.

Started helping other accounting firms do the same. Interested?

I thought you might be. Check out the link down to Cloud.

Accountant Staffing in the show notes. Hey, quick word from one of today's sponsors. Me. Me. Yes. I run a community realize that is for accounting firm owners.

And realize is kind of the secret.

Behind how I've done all of my content and been able to like financially.

Turn thought leadership into a business.

About 100 firm owners in there realize is a systematic way to build connections with other peer firm owners. It's not a framework for how to run a firm. It's a way to introduce you to other super cool firm owners. Right now we're putting together small groups so you can have a group of like 5, 6, 7 other firm owners.

Just like you that you meet with.

Every single month run very similar firms. Having this for me, totally transformed firm running my North Star like my goal for you coming into this group is to make a new best friend, somebody that runs a firm because that was what I needed in my journey. The whole community is built around that. It's called realize. Link down in the show notes if.

You want to learn more, there's people that will argue right now that basic comfort with AI assistance is like Microsoft Office, you know, capabilities, right? Like on your, on your job qualifications you used to put 40 words per minute typing speed or proficient with Microsoft Excel. There's a lot of folks right now that would tell you like AI is that. And like every job listing you see like proficiency with AI is Very, very important. So do you think there is an imperative to say you employ 10 accountants, like is there an imperative to give them general AI expertise right now via general tools? Or do you think point solutions, like AI enabled point solutions, does that bring those people along? Because it's ultimately the most direct impact AI has on the work that they do every day.

I think they're going to get like, if someone doesn't have. So this is all opinion based, forgive me, but like if someone doesn't have the general interest to open up ChatGPT today and say, please help me rewrite this email, please help me understand what I should do, I'm lost in this situation. Here's, you know, X, Y, Z context. If someone isn't grabbing for that today and trying to do it on their own today out of their own personal interest, they're not likely to ever. But like if you bring them into this workflow where they're expected, or they're given examples of it already working, like hey, here's a template that was run on your nonprofit revenue or, or this contractor or these invoices, et cetera. And they see it there and they can just edit the question and then click run and then see the answers. I think they're going to be exposed to and actually adopt generalist tools faster because they'll get exposure and be given use case examples and have the ability to edit. So it makes it more accessible. Like me doing engineering, probably not the right audience, but like how do you set a backend infrastructure to this? Like, I don't know how to do that. Like the barrier to entry to learn engineering has, has never been lower. But it's like now I can go to code that's already working and I can say I need to do this. And it's made me actually able to apply all the engineering concepts I've learned in classes or general interest. Now I can go in and actually make contributions because AI has helped me get past the roadblocks I've had before and by putting AI into the workflow that auditors follow. Like in agentive, they don't have to open up another program. They don't, they're not looking at blank boxes. They can start with things that are already working and they can edit them. And I think that'll give them the confidence to say if I can do that here, now let me go to the generals tool and do this here. But the cold start problem of like everything's a blank box. How do I do this? Like that's almost insurmountable for some people unless you put it in a workflow and say you need to do X, Y and Z to get them the initial reps in practice, if you will, so they know how to do it on their own.

You said something earlier that's kind of like the, the biggest question about AI and accounting. You said you still have to review the output, right? So like what does it look like as we are like heading so fast into this future where we have super powerful AI that could do a whole bunch of the work, but then us.

As humans haven't figured out the right.

Way to wrap ourselves and our processes around that thing. That one moment will be superhuman and the next moment will fail in ways that surprise you. Like right now the structure is you have a prepare, you have a reviewer, maybe you have a second reviewer and then something goes out the door. Does anybody get cut out of that process or does it just change the responsibilities of everyone involved?

I think it changes the responsibilities of everyone involved. So the reality is like there's less people wanting to study accounting, there's less people wanting to go into audit. And we're going to fix that. Make no mistake. We're going to fix that. It's going to be really fun, but younger.

To be clear, you're going to make accounting cool again?

Absolutely.

They're going to look at people like Danny and be like, yeah, this is, that 12 year old guy is going to be like accounting. That's what I want to do. No, you're going to fix it by automating stuff, right?

I mean law is just as boring to be honest. Like I don't know why they're, they're considered sexy and odd. It isn't. But, but we'll, we'll figure it out. But no, like auditors are strategic advisors, they have tons of great insight. But like they do spend a disproportionate amount of their time doing things that are unenjoyable. But we can cut that out and it can be about strategy and relationships. So I have a sister in law who's studying right now and she's afraid of AI and for good reason, because she's like, I'm afraid I'm not going to learn, I'm afraid I'm not going be able to do these things on my own or I'm going to spend time, less time thinking on my own. And like that's very valid. You should learn how to do everything on your own. But that doesn't necessarily apply to me because as like a business Owner or manager, like I need to get more done with fewer resources and increase, like deliver the best product. And so if I have things that need to get done and I can give it to AI and get a good result that I would have used to have either done myself or give to another human, if I can get it done with AI, then I can get, I can 10x my output. And so I think what you're going to start seeing is younger people managing AI agents to get work done and their throughput is going to be much higher. And so you'll be able to get audits done in less time. But there does need to be an auditor at the end of the day looking at this and saying this was accounted for correctly. They follow generally accepted accounting principles. Their investors are not being lied to. Like the public is not being lied to. And so that's like in society an incredibly important role. We're not taking that away, we're just helping them maybe spend less time on lower risk areas, focus on higher risk areas, do more interviews, do more walkthroughs, like do the things that are going to uncover more like where issues really are, less time on. Do these amounts agree across revenue to cash. Do these. Yeah. Do these details agree or was this accounted for correctly? Help them find where judgment needs to be applied. And just so more people will shift to the reviewer part as opposed to the preparation part. And like engagement budgets are going to shift more towards review and strategy as opposed to 50% of like small to medium sized firms probably spend their time on test details for like an engagement budget that will go down the hours spent towards the initial testing and more will be put in towards planning and strategy and review.

Is there a world where the capabilities of that all come together really fast, but then the ability of humans to evolve and adapt around that is just vastly slower.

So great question. I think that if you look at something called the technology adoption S curve and it starts with early adopters, innovators, like early adopters, like the late adopters. And like there's early, I forget the exact quotes. Then there's laggards at the end. And so like there's always people who adopt technology last. And there's the common like statement online where it's like AI is not going to replace auditors or AI is not going to replace X auditors with AI are going to replace auditors. Like having the ability to understand how to do like systemized thinking and to be able to manage three or four like AI agents doing different work at the Same time. Just like you can manage different humans doing different part of the audit at the same time, if you can manage different parts of AI or different AI agents doing different TAs, you can elevate the impact you can make as an individual. And the people who are able to make this transition to managing AI are going to be able to run more profitable audits, are going to be able to deliver better client experiences, better employee experiences. And so it's going to be a real competitive advantage. I don't see it as like there may be some people who didn't like didn't want to make the transition from paper to, you know, paper based work to Excel spreadsheet type work or whatever technology adoption shifts we've had in the past. There's people who have not wanted to make the transitions and the world is not going to be as kind to them as the people who are willing to make the transition and figure out how to increase their leverage with technology. It's the same pattern as existed in the past. Yeah.

When somebody didn't want to use a computer to do their job, somebody else.

Was and they probably won the work in the long run.

If you could snap your fingers and get a new model, a new capability, something like that. You talked about spending 30 minutes to work on something, some new AI capability that you could have right now that would unlock, you know, the next really cool version of what agentive could be. What would that be right now?

Oh man. If, if. So I've Talked to like OpenAI and different teams about our wish list. Yeah. And it's like AI models are trained on all this data from the past, but sometimes they're trained on old tax law where they're. We're 100% trained on old tax law, old audit like rules or old. A simple example ran into yesterday talking with some of my customers yesterday. It's like is this hotel rate allowable per this government. Yeah. For this government regulation. And the number they were referencing was old. And it's like the training data cutoff was just like six months before. And so it doesn't have the most recent one but it confidently gave an out of date like answer. And I would love it to just always like, say is like. But you need to ask the user like what year do you care about like anyways, the training, a lot of the training is old. If we could selectively say here's the most recent version, take out any of the old training data and only reference the most recent version out of all the training data that would Be fantastic. But that data labeling job is just simply not going to happen. And so somehow the models need to be able to ask themselves, am I using the most recent version? And was this something that's updated every year? And if so, if I don't have the most, I don't have the current year, I should probably tell them I don't have the current year. Instead of replying with an out of date version, I should say, this looks like this was something that's updated every year. I don't have the most recent year. I'm not going to give you an answer instead of giving you an old answer.

Even in a meeting transcript, somebody can say something at the top of a meeting and then say something that conflicts with it by the end of the meeting. And like it has to be able to know, like, no, this is the answer that we want to get to. But doesn't this make it so that we can't factually rely on any model training data and have to instead rely on like retrievals to get to? Well, here's the truth that I actually want to get to because that could have been an IRS rule, It could have been an Australian tax rule, could have been from 2022, could have been from 2015. Like do we then, does it mean that we ultimately then can't really trust any of what just the model itself produces?

Factually, yeah, great question. So the way we saw that yesterday with our customer Brenton at Stambon S, we decided to just import the most recent version, you know, like tie out to zip code and say what should that government rate be for this lodging expense in this zip code? And we did have to reference the most up to date. That's how we had to do it. And so right now we do need to point to factual sources. It would be just super convenient if it just had all the factual sources offhand. But that's not a reality right now. So we need to supply it with the factual sources. And that's the conversation we had. And Brent's had that up and running in five minutes after our conversation. But yeah, that's what we have to do. And if we could solve that, that would be fantastic.

I know this is incredibly hard to do, but what do you think audit looks like in five years?

So I think we touched on it already. Where junior level people are expected to be spending. This is what I wanted to spend my time on was audit strategy, like audit planning, like discussing the risk levels, like I wanted to like effectively think about like CFO level topics and discussions and I was put towards do these details agree? And I'm like, I, this is not what I want. Like this is not why I studied accounting. I want to like be at higher level discussions. I think it looks like people earlier in their career understanding the big picture, focusing on higher risk accounts as opposed to just auditing cash for your first year at the Big four, I think it's going to be a much more interesting job. And so I think those auditors, they still might have to manage auditing cash, but they're managing AI agents doing the preparation they're reviewing and it's a lot simpler. And so I think they'll spend less time on less risky areas and more time on more risky areas like revenue for example. That's what every big four auditor wants to spend their time on, but none of them get to touch it because that's what seniors and managers spend their time on. And so I think you're going to see more fulfilling, more fulfilled accountants in their, in their careers.

Okay, I want to hear your biggest, juiciest hot take on the tech ecosystem before we wrap here. What is, what is a non obvious take about the state of the ecosystem? Where the gaps are, where the opportunities are. From what you've seen, what would you say?

I think people should be able to try software before they buy it. I think that yeah, there are some companies that know that they're buying early products and are investing in the founders and want to help them. And first customers are everything and you should just be transparent about where you are, show people the current limitations. But I would say there's too much fraud in selling capabilities that don't exist. Like someone I know just bought some software and he got locked into a three year contract and I think three or I think three year contracts should be legal but, but it hurts startups. Honestly I told that told someone, one of our competitors, I was like, your biggest strategic like advantage over us is your three year contracts. I'm like make no mistake, we will win. But like that's your biggest advantage right now. But, but anyway so, but they're like look, I thought we were buying all the AI features you just showed me. He's like, I thought we were buying that. We don't have any of those. He's like I want those but like I also can't pay for both at the same time. And I'm like, oh. I'm like, we, I'm like one year is something but like three years. I mean, I'm sorry dude, I anyways, Like I, I hate three year contracts because especially if you haven't tried the product before because they thought they were getting something. Now they're locked into a three year contract and they didn't get what they paid for or what they thought they were paying for. They had a salesman who didn't like who oversold something. You know the engine anyways, there's misalignment between sales and what they're actually buying and I just felt bad for them. And we encounter quite a few people that are locked in through your contracts and it's what I hate the most. And it's probably the biggest strategic advantage to like the legacy players out there that can keep their people from trying new tech. Yeah. If that makes sense.

And to selling into more mid market and large firms. Like there's a ton of companies doing this now where it's a 12 month contract you're buying based on the landing page and the features they say they will have. But it isn't real yet.

Yeah. And so I think if, yeah, I think people should be able to try software before they buy it. And I think that you shouldn't sign a three year contract even if they offer you a discount. It's. Unless you've tried it like don't sign it through your contract. Okay.

Look at me, look. You heard it from Danny. Stop buying into three year contracts. Kay. Right. What does agentive do?

We do one year contracts but we let you try it first in a two and a half hour training session where you can kick the boxes and make sure it's a good fit for you before you buy. Cool.

Danny, this is amazing man. Thanks for doing this.

Pleasure. Thanks for having me. Jason.

A heads up for my podcast friends. We put together the biggest, scariest project my team has ever worked on. A big old run of live events around the country headed up until year end to to ensure that your 2026 is just the best year ever. Let me run you through the cities real quick. We're coming to Austin, Oklahoma, Oklahoma City, Phoenix, Seattle, Portland, San Francisco, Los Angeles, Atlanta, Nashville, Chicago, DC, Philadelphia, NY City and Boston with some super cool half day workshops that roll into dinners. You get CPE for it. It is an event that is built for your entire team to get the whole squad feeling inspired so that if you are a firm leader, you're not feeling like you got to put everything on your back. Headed into 2026 gonna be a super fun experience. There's going to be a surprise guest at every city. We're going to record an installment of the podcast live at each of these events, you're going to be part of the podcast itself. What? And, man, I would love to see you there. To learn more, head over to jasononfirms.com events and I hope to see you there.