Hagit Oren, CEO of TaxRay: How AI is Transforming Tax Compliance and Risk Management.

speaker-0: Click the switch.

speaker-1: bitch hit connect turn it up

speaker-2: Hi, I'm Yatsey Tannenbaum.

speaker-1: And I'm David Leishner and this is Heart in Te-

speaker-2: Conversations with people shaping what's next. Here we go. Welcome to Hot in Tech. We have a really good episode for you today. Today we have Hageet Oren, who is the co-founder and CEO of TaxRay, an AI powered platform focused on identifying and preventing tax risk in complex, high volume transaction environments. Through TaxRay, Hageet is applying advanced technology to modernize how organizations manage tax compliance, approvals, and exposure, bringing greater accuracy, transparency, and control to a traditionally manual and risk-prone domain. Prior to founding, Taxray, Hageed held senior tax and finance leadership roles across high-growth technology companies and global firms. She served as a head of tax at Monday.com, where she built and led the global tax function and played a key role in the company's IPO. Earlier in her career, Chagit was a tax partner at Herzog and a tax director at Deloitte, advising multinational companies on cross-border tax planning, ⁓ &A, and IPOs. She began her professional career as a commercial lawyer, bringing a deep legal and regulatory perspective to her work at the intersection of finance, technology, and compliance. Chagit, welcome to the show.

speaker-0: Thank you. Thank you. It's a pleasure to be here. Thank you for both of you.

speaker-1: ⁓ This is going to be a really interesting show because for me, tax was something that I learned in university. I was actually thinking about going into the field at one point and it was a little bit, you could say dry for me maybe, but then it comes together with AI and we're going to be talking tonight about some very interesting things about tax and AI. And I think that's going to be really interesting to hear exactly what you've been doing. And you you've spent much of your career dealing with complex tax risk in fast growing global companies, what led you to believe that this was the time, this was the right moment to build an AI driven solution in this space?

speaker-0: Okay. This is a great question. The tipping point was the combination of two critical factors. First, we realized that global taxes hit an expertise wall. There's an explosion in tax complexity and massive surge in cross-border transactions, especially in fast scaling companies. Yet the processors meant to manage those taxes still remain slow. manual and reactive. But the deeper issue, the heart of why we build tax fray is that the tax intelligence is not reaching the face of workflow where it actually matters. Procurement, sales workflows, the expertise exists, but it's trapped either with external advisor disconnected from operations. or within internal tax teams that cannot scale to every transaction with the speed and volume required. This makes real-time transaction-level tax risk management nearly impossible. You end up discovering problems too late after contracts are signed or payments made based on incorrect tax classifications. When damage like trapped cash, overtax withholding or compliance gaps is already made. The second reason was the technology. The technology is finally mature enough. With AI, NLP and graph-based reasoning, we can remove from static post-factor reviews to real-time pre-signature risk elimination. TaxRai embeds elite tax consulting knowledge directly into enterprise workflows through AI tax specialist agents. It's like having a senior tax expert inside every transaction, providing real-time tax mitigation and optimization steps at the exact point where it really matters, where the risks actually begin. I saw this need, by the way, at Monday.com, where I build the global tax function. Even in highly tax environment, tax workflows were disconnected from core procurement and sales workflows. I knew there had to be a smarter way to shift tax left. embedding proactive controls into day-to-day. By embedding our AI specialist engine directly into business operations, we enable our organizations to prevent exposures, to execute optimizations, and ensure compliance before they become liabilities, instead of discovering them after damage is done.

speaker-2: we're seeing AI being used across the board, right? Every industry has AI in it. I think most companies are most afraid of bringing AI into finance because you know, AI can make stuff up and could hallucinate. And I think there's a fear, like that's the last.

speaker-1: hallucinate I like that ⁓

speaker-2: No, AI hallucinates. It makes things up. It's a known thing. Now, it's

speaker-1: That's why at the bottom it always says, you know, don't trust me. ⁓ Make sure you double check everything.

speaker-2: And David, if we're writing a landing page, then maybe a word is wrong. But if Hageed is, is submitting taxes for a company and it gets something wrong, then that's where things start getting dicey. So I'm wondering Hageed, what are the biggest misconceptions you see about applying AI in a tax and financial compliance?

speaker-0: Okay, I'll tell you. I think that the biggest misconception is that AI in Tux means black box automation. Something that is fast, but unexplainable and risky. In reality, explainability is the core requirement for any AI applied in regulated spaces. At Tuxray, by the way, we didn't just fine tune a generic LLM. We built a domain-specific AI engine that understands treaty logic, sourcing rules, beneficial ownership issues, classification standard. It doesn't just present a result, it shows you why a transaction was flagged, what the tax consequences are, what actions to take. So the misconception is not just technical, it's a cultural thing. Many think tax cannot be, by the way, digitized because it's too nuanced. The nuance is exactly where AI, if trade correctly, adds the most value.

speaker-2: So it kind of reminds me of in university when it wasn't enough to write the answer, if it was five or 12, you had to show the professor how you got to the answer. So it sounds like what you're suggesting is exactly that. I'm not going to say the answer is 12 or 12,000 or 12 million, but you're going to break down exactly how it got to it. And then the tax expert can come and use that as a tool to make the right decision.

speaker-0: Absolutely. And the tool is very important and served him for audit readiness in front of the tax authorities or any other regulator. They need the reasons. They want to understand what was the basis for the determination of the AI and not just a result without any explainability. So the explainability is crucial. It's not just a feature. It's crucial. It's important. It is a requirement for the AI. Yes.

speaker-1: So we're going to talk a little bit about your background, but before we get into, you know, like some of the companies you work for and how that affects, you know, what you're doing today, I'm curious about Oxford. How did you get to the University of Oxford? ⁓ you know, learning law in Oxford is a really big thing. And how did you make that switch from Oxford into, into, you know, tax at, ⁓ I wanted to be a little.

speaker-0: Yes, so I've started in Oxford University. I did my law degree in Oxford University. It was many years ago. It was amazing experience. ⁓ My husband started there. We got married and then I joined him to Oxford University. I applied and I have been accepted. It was a great experience.

speaker-1: It's a beautiful place. I've been there a few times and wow, I mean that must have been amazing.

speaker-0: Absolutely, you know, it's like a legend. The university looked different from what everyone think or saw. So I really enjoyed it. It was a great experience, very tough, very tough, excellent standard. ⁓ But it was a great experience.

speaker-1: around the corner.

speaker-0: Yes. You're dancing very young. Yes, yes. So it was an amazing experience, you know, once in a lifetime. Yes, absolutely. And it helped me. It helped me, you know, to be excellent in what I'm doing, to understand in excellent standards. So Oxford was absolutely something that defined, defined myself, defined what I am. Absolutely. Yes.

speaker-1: Interesting. Okay. That's great. So now, now we'll talk a little bit about more about you. Cause I always, when somebody is like, ⁓ you know, going to Oxford or, or Cambridge, my, my, my wife is British. ⁓ personal disclosure time. So I've been seeing Lynn many times, actually lived in London for a while. And so, ⁓ yeah, I have a very soft spot, ⁓ for the British people and, ⁓ in the far England. So I really like it there. And in fact, I'm dying to get back to the West End to see a good show. I haven't been there in a while. It's time. So we will, we will get there sometime this year again. So, you know, you've worked across law, accounting, finance, and now technology and AI, which is, know, like the, the, the top of tech today, or you could say what's hot in tech today. So how does that multi disciplinary background influence how you design your products and make decisions as a founder.

speaker-0: It's everything. I'm a tax person who become a product builder and that shift informs how we operate basically. I think in risk patterns, edge cases and real world compliance consequences, because I've sat on both sides, advising global companies as a big for tax director and running tax in-house through an IPO, I understand not only the theory, but the operational pain points. disjointed workflows, messy documentations, late reviews, and the constant pressure to balance compliance with speed, especially given the fact that tax management is siloed from the day-to-day business workflows. That experience directly shapes how we build tax rate. It means that our product is built with empathy for the enterprise reality. We don't just offer insights, we deliver actions, items, audit-ready documentations, real-time risk classification, and integrations with systems like CUPA, ZIP, Ironclad, Oracle, because that's where the pain lives and where smart decisions need to happen. And today, it's not enough to say you are compliant. BAC's authority expects you to prove control. to demonstrate that tax risk is monitored, assessed, and governed in real time. That's exactly what we enable. We embed enterprise tax in the enterprise's workflows so companies don't just react, they can own their tax positions strategically. And everything is done based on my experience. It was a long experience throughout more than 20 years, and I bring all this to TaxRay.

speaker-1: Wow, that's amazing. I have a side question for you. My background over the last few companies has been cyber. I'm with C2A Security now. And Yitzy just finished a stint in Palo Alto. And how much are you seeing cyber compliance inside of these companies that you're working with?

speaker-0: In terms of taxes, you mean?

speaker-1: Yeah, yeah. Or the more on the compliance side, I guess. Maybe it's a parallel universe.

speaker-0: ⁓ You know, tax compliance is a culture. You can see the difference in countries. You can see, for example, in the US, the US companies, US-based companies, are more compliant, I can say, sometimes from other places in the world. They have the means suggested, delivered by the IRS to bring them and enable them to be compliant. ⁓ It's not like a recommendation. It's something that they are doing in practice. In other places, you can see, it depends sometimes on the audit and the enforcement come from the tax authorities. You can see less compliant culture companies in other places. So first, think it's a very cultural thing, how compliant they are. ⁓ By the time you can see more compliant tendency around the world. because the tax regulations are dynamic and increased very dynamically. So they don't have any other choice. And the tax authorities, by the way, are trying now to see how they can embed automation also as part of their process to make audit and to enable them to make it more increasing, more quickly. So there is like, know, like... ⁓ something like an arms, I don't remember the name, like something like a weapon arms, like a race arms race between the tax authorities and the side, sorry, that both sides wants to embed a tax automation to be compliant because they know that if they will not do it, the other side.

speaker-1: You

speaker-0: will make it. So this is something that you can see in terms of compliance.

speaker-2: I wonder, Chakit, what that means. know, there are a lot of conversations when we talk about AI, about the future of the domain of accounting, ⁓ obviously tax being a part of it. I know it's probably, you know, as a tax professional, you're probably extremely practical, but I do want to take you to the more philosophical place of where do you envision the world of tax being? in the world of AI, give it three years and five years time.

speaker-0: ⁓ I think that what's truly hot now, by the way, is what I call domain aware autonomous systems. We are moving beyond generic co-pilots into a generation of specialist agents that deeply understand specific fields like law, like tax, like finance, and can actually take professional actions, not just make suggestions. And this is exactly where I see the tax. In tax, that means shifting from, tell me what's wrong to fix it before it happens. It's about proactive, explainable automation system that don't just react, but, or extract tax from, I don't know, from contract or make a tax calculation, but instead they shape workflows in real time with embedded business logic. and regulatory context to the work, to the day-to-day workflows. But then what is matter is the accuracy. Because in tax, 80 % accuracy simply is not enough. You need AI that performs deep tax analysis, understand tricky, treaty logic, how to interpret source of income rules, and to navigate classification nuance with the judgment of a seasoned expert. Without the level of intelligence and precision, you cannot trust the output. And without trust, the system cannot deliver real value. So this is how I can see it. It should be very accurate, okay, and act like an expert, like an AI specialist tax expert.

speaker-2: So then follow up, Hageet, what would be the role of the AI and what would be the role of the human? Is there a role for a human?

speaker-0: Yes, there is a role for the human still. It's going to be ⁓ an expert that will operate the AI. ⁓ It should still take decisions because sometimes there are gray areas that the AI will provide action items or recommendations. So the experts still in a very edge places or very high value transaction would still... require or want to take a decision. But all the research work and all the hours of make the research of this should be made with the AI will be made by the AI and the tax expert will take the final decision, but he will have all the previous preparation work that will enable him to take the decision. Yes.

speaker-2: I'm seeing that a lot with AI that AI will do a lot of the tactical work, a lot of the kind of busy, like very structured work and the human's place will be strategy. And I think that's exactly ⁓ what you're proposing here. There's a lot of strategy in preparing tasks, but then there's also a lot of tactical work. So if we could allow the human to be strategic and make strategic decisions and leave the tactical work for AI, that would probably be the best combination.

speaker-0: Absolutely. But even the tactical work, by the way, should be very intelligent. This is why, know, tax, the AI should be very intelligent because it also provided, you know, some conclusions or recommendations or make research based on the, on the law, the case law. So we understand or should understand the AI, should understand the, to make the analysis. So in these fields like tax, legal, healthcare, whatever, you need the AI. will be a specialist AI that he knows the rules, you know, how to make the analysis, deep analysis in order to bring it, to deliver it to the tax expert to take the strategic decision.

speaker-1: I think, I think when you talk about law and even tax law, I think one of the things that, AI will not be able to do anytime soon, especially the way the systems are set up today is that when you get into a courtroom and especially if you're talking, ⁓ whether you're talking in front of a judge or you're talking in front of, ⁓ maybe, ⁓ a jury, depending on, on what the, what the court cases, there's a certain human instinct that trial lawyers have and it's irrelevant whether it's a criminal trial lawyer or a tax lawyer that I think, at least the way I see it, that they have these instincts that you're not going to see an AI or AI taking over anytime soon. And I think this is also in, like you mentioned, Yutze and Hagi, the strategy of how am I going to approach You know, the way I'm setting up structures, the way I'm handling offshore accounts, the way I'm setting up where I need to set up my business because potentially there are geopolitical issues involved. what, where do my customers want me to set up business? What, where do they want me to keep my code in SRO? Where do they want me to, to, to know that my, my, business is actually based, ⁓ you know, depending on where the, the flavor of the day problems are occurring in the world. ⁓ so I think, I think a lot of this is, ⁓ you know, you could say, okay, well AI can react, but there, are certain things that, ⁓ and, I'm a believer that, you know, we should use AI for as much as possible. But I think, I think that there definitely will be plenty of room left for us humans, you know, to do what we need to do.

speaker-0: Absolutely. You know, one of our challenge here, or you know, that we are trying is to bring the common standard behavior of companies in terms of complying to taxes, because there's a lot of, you know, verbal common standards that are not written anywhere that we are trying to bring into our AI. ⁓ that when he think and make the analysis, it is not just on the dry, you know, ⁓ rules, tax law or case law, but also based on common standards that are in the environment that nobody knows it. It's not on the website, it's nowhere. And it's a result of behavior of companies in respect with compliance. So absolutely, I think that the AI will, it is something which is enabled. to the companies to take strategic decisions, but there is a great room for them at the end of the day to consider the decision and to take it.

speaker-2: It's the nuance that you mentioned earlier. So, there's a question that we ask all our guests on hot in tech. I spoke to you about five years in the future, 2026 in the world of tax in the world of AI, any, any kind of predictions or gut feelings of what's going to be hot in the upcoming year.

speaker-0: the AI specialist agents, that's what I emphasize. It's going to be the agents that are specialized. ⁓ they know, they know how to make deep tax analysis or other deep analysis in other fields, very specialized fields. We want to bring the specialized agent to the day to day. The, the expertise is trapped. is trapped with the external advisors or with the internal tax teams. They cannot have the ability to bring it to the day-to-day operation. So by several agents that you just embedded to the day-to-day systems, you can bring special expertise. And this is what I think the expertise and the specialized agents are going to be very hot. ⁓ This is my opinion.

speaker-1: Makes sense. So I have a question. I have a question for you before we get into our last and final question, which is quite often the most interesting. you you came from a very solid background, as we heard. What was your biggest challenge, you know, pulling together the team, the, you know, working with AI, making, getting the product. you know, off the ground. Did you have like one major issue or two? I don't know. What was like the... What kept you up at night, you know, early on when you were first starting out?

speaker-0: Okay. The team is the most important thing of the company's success. Absolutely. And we have another challenge here because there is two domains, different domains, the technology and the tax. Different people, different cultures that are coming from different, different domains. The challenge to integrate, to combine them together because we need, you know, to make integration between the tax and to take the tax, the service world. to technology, this was the main challenge. But when we succeeded to do it and they speak the same language, this is our success. Because basically we transform the tax services to a product. We productize the services world. And this is by people that coming from different domains. So this was one of our great challenges, but. it's going very well, the tax persons are part of the product. They're not providing us any ⁓ consultancy services or advice. They are part of the co-product. At the same time, the technology guys should understand the customer. The buyer is different. They used to customers to buyers like ⁓ the CISO or other engineers that using AI ⁓ tools. Now it's another buyer. It's a buyer that understands my tax. They should fit and match the product to his needs. So the combination, the mix between them was one of our challenge. And if you succeed there, this is one of the success of the companies. So it's very important. ⁓

speaker-1: Very interesting. keep it, keep bringing the domain expertise, bring in the domain expertise early and make sure that they are in sync with the tech, with the technology people. That's great. So for founders and operators building companies in highly regulated domains like you're in with AI becoming increasingly embedded in core workflows, what advice would you offer to help them innovate responsibly while avoiding unnecessary risks?

speaker-0: I think it's about three things. The first is build domain intelligence first and AI second. Don't start with the model. Start with the problem, the regulation, the decision logic. Only when you can build a system that's compliant by design, you can make it. You need to understand what the system is supposed to decide, how it's supposed to justify that decision and what outcomes are acceptable under regulation. This is the first thing. The second one is make your explainable always. This is the explainability. In regulated spaces, you don't just need results. You need reasoning, documentation, and defensibility. If a regulator or audit ask why, your product should answer before your tax or legal team do. Regulators and internal stakeholders may ask. Why was this classification chosen? Where is the supporting documentation? Explainability, as I said before, is not a feature. It's a requirement. The third one is design it for integration, not isolation. In highly regulated domains, innovation lives or dies on adoption. If your tool cannot fit into existing systems, like, I don't know, in our in our case like ERP, CRM, procurement, it won't matter how good it is. It should be part of the systems of the enterprises, part of the workflow. So in short, in regulated domains, your AI must think and explain like a professional and fit like infrastructure. That's the mindest we bring at tax rate. It's what let us bring real innovation to a space that have been overlooked for too long. That's amazing.

speaker-1: That's amazing. And I think what you said, the number one is the domain, the domain, I think, ⁓ just trying to use AI in order to make something run a little faster or maybe with less resources is not the true benefit of AI. think, I think AI from ground up with the domain expertise to be able to come out with something so much better, so much, you know, that provides so much more for industry, for each industry and each solution in its own right. I think that's something that people have to realize and you can't just come out and say, okay, I know AI, so now I'm going to go into this industry and I'm going to conquer it. Well, first things first, bring somebody who has years of experience like Hagi Doran to the table and then talk about creating a company.

speaker-0: Absolutely. You know, it's also because you understand when you are the subject matter and you lead this enterprise, this company, you understand the needs of your buyer. You've been there. You've done that. You understand what feature will be interesting, what their pains, real pain, so you don't spend time on developing things that are unnecessary. You understand everything in advance. Of course, you listen to your customers, you understand what they need, but you In most cases, you understand what is going to be their response and what they need. So absolutely the subject matter should be part of the founders.

speaker-2: And it's also that deep, it's more than just understanding. It's knowing what that night is before you have to like fill it out. That endless night, you're ordering pizza, your eyes are like killing you. You haven't seen your kids in a week. It's understanding that and building something ⁓ to meet that feeling. So, Chaguit, I kind of wondering, and this has been bugging me and I'm sure people will hate me for asking this, but I assume that governments will also turn to AI as the flip side of this, No, scan every single ⁓ tax report that comes in for a company for an individual, flag those that seem problematic and audit them, right? Are we going to see, like, do you anticipate a B2G move or governments adopting AI to kind of, because that's how AI works, right? In cybersecurity, we see ⁓ The good guys using AI to protect and the bad guys using it ⁓ to attack. Are we going to see the same thing in government?

speaker-1: By the way, and then you have the compliance regulators who are telling you that you have to ⁓ regulate against AI and you're using AI in order to provide those compliance regulators with the reports.

speaker-2: You

speaker-0: I'll answer you because this is a very good question, by the way. I think for sure, okay, that the tax authorities are going to have more and more automation, as I told you before, to help them, to enable them to make audits. Okay? This is their incentives. But you can build a platform that can serve both sides. I know that it sounds crazy, but you can make it for both sides and make collaboration, good mutual collaboration between the governance, the tax authorities and the business market. Because at the end of the day, the market want to be compliant and the governance or the tax authorities want them to pay taxes. Okay. Very, you know, legit. And you can build a platform that can serve both sides. in a good collaboration, I can tell you that Taxray delivered to the customers all the positions of the tax authorities, okay, of the Israeli tax authorities, the US tax authorities. We deliver the position and the policy and the guidelines of the tax authorities because enterprises want to be compliant and they want to know what are the tax authorities' guidelines. So basically, know, we are something like provide them, okay, in a platform instead of getting to the end, being audit and some disputes will be, you know, started. We basically eliminate, we can eliminate by platform any future disputes with the tax authorities. So this can be a vehicle. to deliver the tax authorities position to the market and still to enable the market to take legitimate positions. This is not contradicts one of the other, but for sure both sides want to enable themselves ⁓ to buy and to embed automation in this era.

speaker-1: Amazing. You've made tax very interesting. And it's not your eye. And I think it's pretty incredible. And I think what you've done with the company, it's really something special. And we wish you all of the success going forward. And thanks for being on the show.

speaker-0: Thank you very much. I really enjoyed it. Thank you very much. Thank you.

speaker-2: If you enjoyed this episode, make sure to subscribe so you don't miss future conversations with the people building, breaking, and reshaping technology.

speaker-1: This podcast is proudly sponsored by C2A Security, the only context-based product security orchestration platform. Stay tuned for lots more Hot in

speaker-0: the switch. connect.

speaker-1: turn it up. Roll the feed, too much noise, Get the

speaker-0: We dive straight in future in your ears this time. Hot in tech. Turn it up. What's coming next? From the labs to your desk. Yes, we're all the spans. Hot in tech. Big ideas, bold plans. Plug in, you're set. Hot in tech. ⁓ drops another wild start a change how we move how we pay

speaker-2: You

speaker-1: Tiny chip in your hand on your wrist Tomorrow starts here you

speaker-0: you

speaker-1: What's new? What's now? What's next? We're live. Big dreams. Hit play, lean back, let your mind connect. It's one track. Hot in tech.

speaker-0: on the edge of the line Three words.