What if you could answer any IRS question—instantly, accurately, and with complete confidence?

In this episode of the Tax Rep Network Podcast, Eric Green is joined by AI strategist Denis Mezheritskiy to unveil a game-changing breakthrough: a private, purpose-built AI “Core Brain” trained exclusively on real tax resolution expertise—not unreliable internet data.

They break down how this new technology allows tax professionals to:

• Instantly access answers from over 700 vetted Q&As and decades of experience
• Eliminate delays, guesswork, and risky AI hallucinations
• Scale their practice without adding staff
• Increase profitability while working fewer hours
• Build enterprise value with systems that reduce key-person risk

This isn’t generic AI—it’s a fully trained digital expert that thinks like a seasoned tax professional and delivers step-by-step, actionable guidance in seconds.

If you’ve ever felt stuck waiting for answers, overwhelmed by IRS complexity, or limited by time and staffing—this episode will change how you think about the future of your practice.

The question isn’t whether AI will transform tax resolution… It’s whether you’ll be ahead of it or chasing it.

Join Eric and Denis for the free webinar on May 7th here: https://taxrepllc.com/20260507-answer/

—

Watch the episode here

 

Listen to the podcast here

 

How To Answer Any IRS Question In 60 Seconds With Denis Mezheritskiy

Building The “Tax Rep Vault”: AI Without Hallucinations

Thanks for being back here, everyone on the show. I am joined by my friend Denis Mezheritskiy. Denis, aside from doing a lot of M&A work, buying and selling businesses, coaching people, and consulting with people to get their business in line, is part of a development group that is on the cutting edge of the agentic AI stuff that is happening, and has been working with us. We have developed the Eric, AI, Terminator, Core, Brain, whatever you want to call it.

Those of you who are not TaxRep members, you do not know what you are missing. We have TaxRep Vault. Vault is a place where we have accumulated everything. All the books are there. It is over 700 Q&A questions, all fully answered. All of the critical cases in the IRS rep. All the important information we think you need to know about contacting the IRS. Everything is in this one place.

We called it Vault because this way, you do not have to scroll through all the training and look for stuff. It is all there. You could go, and you could search for it. You could scroll through and find stuff. Not anymore because of Denis and his team. We now have our own AI brain, for lack of a better term. We know it will not be hallucinating or doing anything other than what you see in the headlines now with the crazy stuff, because there is no contradictory information. Nothing.

It is finding on the internet that confuses it. It is literally just going through our Vault. Now, our members, I think, in about a week, we will go live. We have been playing with it ourselves. We will just be able to query. Denis, first of all, thanks for taking the time. I know you are a busy guy. We literally have a game going. Where is Denis? Is he here? Florida, Chicago, Europe, Bahrain, you are all over the place.

I appreciate you having me here. I love the way that you pronounce my last name as well. That is a big thumbs up. Just for everybody who does not know me, I do a lot of M&A, I do a lot of growth, I do a lot of acquisitions. I train a lot of clients on how to grow their companies through acquisitions or other means. AI is the way to grow at this point. There is no better way for me to help my client increase the value of the company than by utilizing AI.

 

 

There is no other way to increase a company’s profitability, no matter what industry it is, unless you implement AI, especially in the accounting industry, where the biggest problem is that you cannot find people to work in this industry. There is absolutely no other way. I dumped all of my time at this point to partner up with the largest implementation company in the world at this point and create AI implementation for companies that literally will grow the companies in no other way better than AI. I am excited, and we have been working on this thing for some time. Is this a monster that we created?

Look, I have to tell you that if I were to leave my firm right now, Kevin and I were talking about this. I said, “Look, if I were to leave my firm right now, I am not outfitting an office in New York or Boston, I am going to go to Regus. How did $25 a month?” I am now a member, I can book as I need it, $35 a day. I am not going to have a receptionist because I am going to be virtual.

I am going to have an AI receptionist who, by the way, can actually go through a complete decision tree. “You have an IRS issue, Mr. Smith. Sorry to hear that, but yes, I am sure we can help you. Is it IRS only or state?” There is a whole decision tree we can create. They will take all the information and email the entire team. Not only when you get that, but you get the summary. You can get a transcript of the entire call, and you can listen to the recording.

Here is the problem with this. Everybody who is trying to implement this AI infrastructure in their company, and they are using the databases of ChatGPT or Google or whoever, though publicly traded, publicly available information, cannot provide the same value that you would provide. You hold the key to your knowledge base, and that is the whole differentiation.

We are going to get to that, by the way, because if you are tuning in to this, going, “Eric, that sounds cool, I am not going to have a development.” You can develop your own brain. We are going to get there. This is where I think the world is going. I am going to tell everyone a number of reasons why you want to do this. In fact, let us just do it now. The reasons you want to do this, aside from the fact that you can, in a way, duplicate yourself. For me, here is the thing for TaxRep.

What does it do? Our members can now get questions answered instantly instead of waiting for Eric or the trainers to get on the help desk and respond. In fact, I got an email from Nicole saying there are still three questions. You have got to get on. Why? It’s because I check twice a day. I probably checked at 4:00, and at 5:00 they came in, and I am not going to check again until this morning.

Scaling Beyond The “Key Man”: AI Employees Vs. Physical Staff

You get your question answered instantly. Here is the other thing it does. Now Eric would like to sell TaxRep. I do not. Eric wants to sell TaxRep. Any buyer is going to look at this and say, “No, key man risk. What if Eric leaves, drops dead, or we get into a fight? What am I buying?” With the AI brain, if Eric, God forbid, vanishes, TaxRep just keeps rolling on.

In the most relevant industry for this particular reason, all of your business is on this one individual who is operating the business, and if that individual steps away, there is literally no business to sell.

Anyone who buys an accounting tax practice has got to be thinking about this. What if the owner leaves? I saw this. The owner would hang around for two years, and all the clients would stay, and the buyer would be happy. The owner now retires. You know what the clients say? “I stayed because of Bob. Bob is not there anymore. My cousin Charlie does taxes. I am going to go with him.” All of a sudden, that practice starts to vanish.

They only hung around to make sure he got his buyout. Versus, let us say I said, “Look, Denis, I am selling my practice. Everything is automated. The engagement letters are automated. Everything is done. Electronic upload through SmartVault. We have an AI brain.” Everything is automated in terms of preparation. Someone can walk in, sit there, and it is not dependent on Eric. In fact, I may have clients who have never met me. The transition, the value the seller is going to get, goes up.

Even if you are not selling, when you bring a new person in the door, it is a lot easier to say, “Here is the infrastructure.” You will find all your answers. You find all the SOPs. You have the guided star. It amplifies every single hire by a thousand. One person hired is a thousand people at this point, but you have to have that infrastructure in place, already programmed into your business, not trying to get cookie-cutter.

Where I was going before with the receptors, but with the AI employees, they are on 24/7. Work is going 24/7. Imagine if you could double your practice and work fewer hours. It all falls right to the bottom line. I know because I have a friend who did this for his what is called Inside Sales. He has clients who will call and say, “We need another machine, right? We need another this, another that.” Those are referred to as inside sales. They already have a relationship.

He had someone full-time, 85,000 a year plus benefits, who did that. They put one of the AI employees in place. Not only does the client get a faster response, like instantaneous, “No problem. We have that machine. I can have it out in two days.” In fifteen minutes, the transaction is done. They have an electronic agreement to sign. Now you do not have to worry about vacation, sick time, or all the benefits that go with that. Having a desk, having space, the whole world is going to change.

It used to be you scale. I am going to focus on the accountants because that is who I think most of our audience is. You scale because I got a bigger building. I got more staff, and I had to have seniors, had managers, and then I had partners. The whole thing about shit slides down, money slides up. You had this army of people. Now you need HR, you needed IT, you need the overhead, the space, the physical. Not anymore.

That does not mean you have to fire people. It just means that you just magnify the people that you already have, and everybody has the same problem. Everybody I talk to, especially in this space, says, “I cannot hire the right people. I cannot grow because I cannot hire the right people.” That has absolutely changed completely.

AI, if you are an entrepreneur or business owner, this is like a panacea with AI. If you are somebody who just wanted to go to work for a company, punch the clock, and get benefits, you are a dodo bird. Your days are numbered.

We are not talking about helping you write emails using ChatGPT. That is not utilizing AI. ChatGPT does not know the accounting practice. It does not know how to process it. This is why we took all of that knowledge base in order to create the most advanced accounting system in the world. We took Eric’s knowledge base, decades of experience all throughout different disciplines, and trained our language model based on the exact knowledge base that we know is correct. Not that ChatGPT makes up 40% of the time. We use the knowledge base, and that is the differentiation factor. It is not for email writing. It is for actual implementation.

The Danger Of Generalists: Why ChatGPT Fakes Citations

I am going to tell you a story. My wife we have the premium ChatGPT because it is worth playing with. My wife, who runs a PhD program at a major university, and they are having this whole struggle with the students using it, and how do they use it? They are going to use it. They are not going to sit in the dorm room and say, “No, we cannot use AI.” They are going to use it. How do you use it properly? She actually has summarized the state of the science in her area, and it cites an article as the definitive article on this topic, one of her articles.

She said, “Yeah, the article blah, blah, by Dr. Dina.” I said, “That is cool.” She said, “Look, I have been published about 150 times. I am pretty sure I would remember if I wrote that article. I did not write an article by that name.” She actually said, “Are you sure about this article?” It came back, “Absolutely, blah, blah, blah,” gave her the citation again in the summary. She said, “I do not remember writing that article. Did you make this up?” ChatGPT said, “Yes, I made it up.” She was all excited.

She knows what to ask for, right? If you have somebody who does not know and is looking for these answers, they rely on this data to make decisions for their clients.

She said it is a little frightening. She is waiting for the day she gets a call saying, “Can we talk to you about this article because we cannot find it?” She will be like, “I do not know what you are talking about.” That is the part that she was all excited to take these screenshots because, especially where it said, “Yes, I made it up,” to show it to the PhD students and the master’s students, saying, “Look, it is good to use as a tool to refine your ideas and other things.” You lawyers are finding this out the hard way, too. The IRS Chief Counsel got caught doing this.

The secretary went out on ChatGPT, and it had it write a motion. This is a motion they file all the time. Why would you even go on to ChatGPT to draft it? Just pull one and just change the name. They filed it. The attorney on the other side is saying, “We do not know what these cases are.” They had a whole hearing, and it was in the news. The court was like, “First of all, you lose because you are citing nonsense, but what the hell is going on in the Chief Counsel’s office?” The government is not supposed to be doing it because it is a public domain. You are not supposed to be doing it anyway.

That is right. You can be subpoenaed for all those records, and you cannot do anything about it. If you are talking to your clients and you are chatting about the clients, you are helping ChatGPT draft those letters, you are in big trouble. If something happens, you are screwed. It is happening over and over again. We will acknowledge this.

One of the things, aside from the fact that what I liked about this is that it is a very potent AI, and we can build it and feed the brain. Just like a child, we are going to raise this the way we want it raised. What I also found is this. It is private. Our stuff is ours. That is a big deal, especially if you are in the tax world, if you are doing planning for clients, certainly if you are a lawyer. ChatGPT and all this stuff, that is not private. It is in the public domain. Even if you have a premium account, it is a public domain. That is, for me, legally, that is problematic because I want to make sure we own our data. The stuff that is between my ears, whatever it is worth, we want to make sure we own that.

For everybody as well, when you are working with your own data, the whole idea of AI is just a knowledge base. If you are working with the wrong knowledge base and there are lots of hallucinations, because that is how this program works. If it does not know the answer, it hallucinates. It will never tell you, “I do not know the answer to this.” It will hallucinate. You are dealing with completely wrong sets of datasets.

The whole idea of AI is just a knowledge base. Share on X

Cracking The System: Dissecting Data Into Segments

Denis, if you could walk people through what is involved. Even I know I gave you guys content, and I remember you telling me you do not just dump it in. You are feeding it in. Can you walk people through just what the process looks like from your end?

The way that it worked, we cracked the system because everyone else thinks that you just take all your SOPs, all your articles, all your books, all your materials, all your courses, all of it, and you just dump it into one bucket. Somehow, AI is going to figure this whole thing out. That is the biggest problem. When you do that, when you just upload all of your stuff as a whole, that is when the AI hallucinates. AI only hallucinates when there are two contradictory answers to a question.

That is the key to breaking down the data. The way that we did it, we broke every single part of the process into multiple segments, and we separated it into how AI should interpret it. When we went back and tested, when we ask questions, specific IRS questions, tax resolution questions, or accounting questions, any questions, we have thirteen different disciplines. We have the CEO brain, and we have the CFO brain market.

We have a lot of different disciplines as part of the structure. The way that we structured it is that for every single test, we know exactly where the answer comes from. When I am asking, like, “I have received this IRS notice in the mail, what the hell do I do with my client?” It knows exactly where to go and look for that particular data. That is why it does not hallucinate. It gives you an understanding of where that data is coming from. That is the key differentiation.

The interesting thing when we were doing this is I had to review everything in case, not that I would have contradictory information, but I could have answered a question five years ago that had since changed. The book is updated. Now, what would happen is the brain does not know the Q&A from Eric, saying for an offer, you do this. Let us say it had changed in those five years. When we did this, those 700 questions, first of all, we did it when we built Vault last year.

We had already gone through and culled all the old stuff. The last thing I need is members relying on something from years ago that is no longer relevant or accurate. We did that again as we uploaded stuff for the very reason that I did not want the brain running into contradictory information, and now making stuff up because we have members who are relying on this. We have firms.

Especially in such a rapidly changing environment, there are so many new updates coming from all different directions. Everybody individually cannot take the time to figure out what is right. We need to be sure that we are working with data at all times. That is why separating into segments is the only way to do it. There is a special way that the team designed a specific process to dissect all of the data into different segments and underline the details of each sector. Now we are only working on the details, the way that AI thinks, not the way that we get information.

For the folks tuning in, I think the way I would urge you to think about this is, what could you do if you had another you? How much more could you do? For instance, we are working at the firm and using the same brain in terms of drafting our documents. Again, using just our document management software. There are some other players out there, Harvey, and these other ones. That is what they do. They are very expensive. The issue is that we can get rid of all our paralegals. First of all, packages still need to get mailed.

 

 

Things need to be signed. Our staff is not going anywhere. But if we could free them up to do more, we could have the AI draft the initial document. At least the partner comes in, says, “I need a motion for whatever. I have a brief that is due.” If you could have that draft brief within 30 minutes, we do not need a paralegal now spending seven hours drafting. First of all, we are not dropping our price. It makes us more profitable. Two, we can now do way more with the same. We can punch way above our weight because, effectively, we now have these AI employees that are on 24/7. It would almost be like staffing up more people and having them go.

That is a big differentiation between an agent and everybody is talking about AI agents and AI employees. An AI agent is just a sequence of automated tasks. An AI employee is an employee that you have to train. You have to nurture that employee. You have to give all the data to the employee. The employee has to see how you operate. Just work on your onboarding process currently with those employees. It is the same process with training the AI as they come along. The only thing is that they work 24 hours, they never ask for a raise, and they do not bitch and moan.

Jerry Seinfeld, years ago, had a stand-up routine where he talked about what kids are. Kids are where people are like, “You know what? I am sick and tired of this. I am going to make my own people.” This time, you do not have to go through the whole childbirth process and spend all the money on it and everything else. You create an AI person, and you have to train them properly. Once that is up and running, it will be life-changing to you and your business.

That is the same thing we did for all the other different disciplines. We took the most knowledge base of a CEO executive and created a CEO persona. Now you can bounce your ideas from a CEO perspective.

Actually, Denis, that is good. Let us get there. I built the IRS brain, but I am actually only one of over a hundred of these on this platform. Describe the platform itself.

The Virtual Conference Table: Accessing C-Suite Expertise

The way that we are looking at AI is that I do not want to use a generalist. Using ChatGPT to me is like Googling medical symptoms. You are getting all kinds of things, like you could be good, you could have cancer, you could be dying tomorrow. You do not want to be Googling medical advice. That is how I think about ChatGPT. What we have designed, which helped my own company tremendously, is to have an expert for every single team.

Using ChatGPT is like Googling medical symptoms; you do not want to be Googling medical advice. Share on X

The way I am looking at this infrastructure is I am at the conference room table, and I have all the bunch of seats around the conference room table of the top-level executives around me, helping me, the small business owner, grow the company to the next level. What can you change if you had Elon Musk as your CEO? What could you do if you were Gary Vaynerchuk as your marketing executive? That is the thought process. We created the highest possible level of COO executive. We created the most interesting marketing brain.

A marketing brain that was trained the same way by the most experienced people in that trade in the industry. That is every single anybody around the conference room table at your company. All these seats that are empty, and you are wearing all of these hats, you have the ability now to actually go from one question to another. That is the beauty of this whole thing. You are asking the professionals who give you the professional answer, not some bullshit emoji ideas, but actual, concrete advice.

You go ask ChatGPT how to open an accounting company or accounting practice. What kind of answer are you going to get? “Do a business plan. Create a name.” They are very fluffy. You ask the CEO executive or an intelligent brain in our database, and it will go, bam, bam. “How much money do you have? How much time do you have?” It will give you the step-by-step process. “This is what you need to do. This is how many calls you need to make. This is how many clients you need.” That is the difference. Asking an expert or asking a generalist. This is like asking Google for medical advice.

JT and Denis basically went to a CEO and then had what I did. CFO, marketing experts. Now, with this platform, you, as a business owner, can get all your IRS questions answered. Easy peasy. Everyone keeps talking about “I need to market more.” How do I put together a marketing plan for a digital campaign for an enrolled agent in Mississippi? You ask the marketing brain, you are going to get a very detailed response, and then it will even say, “Do you want more detail?” It will walk you through step by step. Our team is loving the platform because we can now ask not just IRS stuff, but marketing, CFO, and CEO.

There is no fluff in it, either. That is one thing I really hate. I hate this gaslighting that all of these AIs are giving you. I need honest, brutal honest responses. I do not need to be gaslighted. I want honesty.

New Revenue Streams: Monetizing Instant IRS Analysis

I want to know not just this broad brush. Tell me what to do. I can tell you right now, our members, our people on our webinars, do not want me to talk about what an offer in compromise is. They know what that is. Tell me what to do. Tell me, “I have the client now sitting staring at me. What do I do?” It is a step-by-step approach, and that is the part that I really love about this. For those folks tuning in, Denis and I are going to be talking about this and doing a demo.

It is a free one-hour webinar on May 7th, Thursday, May 7th, 1:00 PM Eastern time. 10:00 AM for those of you who are in Pacific time. It is free. I am going to put the link in the description below. It is TaxRepLLC.com/20260507-answer. It is called How to Answer Any IRS Question in 60 Seconds. Because actually, that is probably a little bit long. It will probably be more like 15 to 20 seconds. You put it in, and it will give you not just the answer, but a complete analysis of the answer.

If you are like, “Why is that the answer?” it will explain it to you. You are going to get a chance to take a look at this in real time. For the TaxRep members, we are going to roll this out. We are looking at probably April 30th. It will probably be before then. By April 30th, this will roll out in the members’ area. You can just query and get all your questions answered. For non-members, you will have a chance at a deal to get access to the entire platform.

 

 

If you can join us on May 7th, 2026, I am going to put the link in the description below. Denis and I are going to walk you through how it works. We are going to actually show you the output that you get from this. For those of you who are interested in building a practice or have enough clients that you understand you could be charging, we charge $750. If you come in and want to do a consult with a query, it is $750, and we get it all day long. I have almost had no one balk at that. If you are like $750, $350, if you could answer questions and get $350 a pop, this is an entire revenue stream. You sleep better at night because now you can answer the question very quickly.

This is not the only thing. Other questions as well. If you have a business owner coming in and you want to give them some advice on growth and strategies on marketing, you have the platform. You can ask any C-level executive, “What would you do?” You can upload financials from a client you are working with and ask, “Give me some detailed structures on how to grow this company? This is my cost. How do I reduce operating expenses?” It will give you the whole analysis. You are going to become a genius in all aspects of the business, and you will get that advice not from the system that you would need to question, but from somebody who actually walked that path before and has that experience to walk you through that step.

As you hit on it, it is not the fluff stuff because actual people were recruited to do this. That is the difference. It is real-world, very detailed information. It did take me a while. I already had all the stuff I had to pull together. There was a lot of effort into this. I am actually really excited for people to play with this. They are going to be extremely pleased with the fact that they can get this output this quickly on any IRS matter.

To go on board on the platform on almost any matter, any C-suite matter, there is a brain. By the way, there is one for insurance. There is one for financial advisors. People who are at the top of their game were recruited by Denis and JT to sit down and do this. The best part is that as you play with this, there is a place where you can connect to build your own brain, talk to the team, let them know what you are thinking, and they can do this for you for your practice, if you have a specialty area.

Magnifying Productivity: The Future Of Specialized AI Brains

If you have team members, let’s say you have somebody who is doing your marketing, why do you not magnify that team member with a whole infrastructure on the marketing side? You can magnify your staff. You do not have to replace them, but you can certainly help increase their productivity, and money will flow.

It is a great opportunity. Guys, it is free. For the May 7th webinar, I will put the link below. How to Answer any IRS Question in 60 Seconds. Denis, thanks for taking the time. I know you have got to run, and I have got to run, but on May 7th, 2026, everyone, and for the members, look for our announcement when this goes live. It is going to be game-changing. I can guarantee you.

I am excited. Looking forward to it.

Everyone, and thanks for tuning in on this episode. I will see you next time. Bye-bye.

 

 

Important Links

• Denis Mezheritskiy on LinkedIn
• Denis Mezheritskiy on Instagram
• Denis Mezheritskiy on Facebook
• How to Answer any IRS Question in 60 Seconds

 

About Denis Mezheritskiy

Denis Mezheritskiy – President, ROI Advisors, Inc. | Founding Partner, AI Agency Group

Denis Mezheritskiy is President of ROI Advisors, Inc., where he advises business owners on buying, selling, valuing, and preparing companies for exit. With over 20 years in the industry, Denis has supported transactions and engagements across service, distribution, wholesale, and medical businesses, with deal sizes ranging from $5M to $100M. He is known for working in the “middle zone” – companies that are often too small for investment bankers, yet too complex for typical Main Street brokerage. In addition to M&A execution, Denis provides value-building consulting for owners who are not ready to sell today but want to systematically increase enterprise value over time. He tracks performance, identifies operational and financial levers, and builds a clear roadmap that aligns growth strategy, risk reduction, and valuation positioning.

Denis expanded his focus into AI implementation as a core value-creation lever for lower middle-market companies. As a founding partner at AI Agency Group, he helps businesses integrate AI into their infrastructure to optimize workflows, strengthen knowledge management, improve employee effectiveness, and create scalable operating capacity without proportional headcount growth. His work focuses on practical AI adoption that produces measurable operational ROI, strengthens exit narratives, and positions companies as more attractive acquisition targets. For sellers, AI integration can increase valuation multiples by showcasing modernized operations and reduced execution risk. When executed correctly, AI integration can support higher valuation multiples by demonstrating modernized operations, stronger reporting and reduced key-person risk. For buyers, Denis identifies AI upside opportunities during diligence and integration planning. This allows the buyer to build an integration plan earlier, quantify the improvement potential, and avoid emotional decision-making by grounding the deal in facts, comparables, and execution reality.