# AI Is Coming for the CFO Office | Stacks, Albert Malikov ## Метаданные - **Канал:** Lightspeed Venture Partners - **YouTube:** https://www.youtube.com/watch?v=ndJO9PhZ2Tw ## Содержание ### [0:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw) Segment 1 (00:00 - 05:00) Some of the systems that are in place are not fit for business needs today. What is needed is quick ROI and real uh use cases that you can solve immediately. And it was clear that monthly financial close was that workflow with a company called Niveoda, a trader of diamonds with more than 13 entities, 130 bank accounts. They were spending more than 8 hours for journal entries. So we cut down this time to less than 10 minutes to reduce time for closing the books for 8 days. Just imagine that you can have a view into your financials and your data 8 days faster. Accounting will change as well because there will be tools allowing to do uh a lot of work much easier. And what would really be an important uh skill for a finance person — Hi everyone, I'm Alex Schmidt and this is the investment memo. The show where founders reveal the stories behind their businesses and how they will shape the future of their industries. Today I'm joined by Albert Malikov, founder and CEO of STAX. STAX is building an AI first operating system for the office of the CFO, starting with accounting reconciliations and expanding toward a broader AI native finance stack for the enterprise. Today we'll talk about why the CFO function is ripe for disruption, how STAX is approaching this problem differently than legacy vendors and point solutions, and what it takes to build foundational infrastructure in a highly competitive market. Albert, welcome to the show. — Thanks, Alex. Very excited to be here today. You're announcing your series A today, which is a big day for you considering that you've only raised your seat around less than 12 months ago. Let's start with actually the outset, the market that you're building in, why the CFO office is actually ripe for disruption. H you know this because we've shared this with you. We've actually published a thought piece on the ERP super cycle and replacement layer and there we're talking about the fact that the core ledger of the ERP is in some senses replaced but in other places even more augmented by solutions that really actually do the work versus just support the work. Maybe level set us in this conversation a little bit. What's happening in the office of the CFO today? — There are three u interesting trends colliding together. one there is a transition of the um finance function in being more strategic and driving intelligence and driving the business decisions forward. So this is first one. The second one um so what we're seeing there's a huge shortage of uh CPAs accountants um in the space uh the um number of exams for CPAs in the last 10 years dropped by 40%. and US alone seeing a shortage of accountants uh in the amount of 300,000 uh people at the moment. This is the second trend. And then the third one um so there is a huge uh data fragmentation that is happening um in the space as you mentioned uh with the uh unbounding the ERP itself. Uh we're seeing a huge adoption of the point solution which makes um data being a big problem in the space. All of these trends together with the uh technological shift of the AI making this uh space very interesting for innovation. — So if I'm listening closely to you, what you're highlighting is basically there's massive labor shortage but at the same point data fragmentation and so if we dive a little bit in the jobs to do if you look at accounting reconciliation why is this all still so labor intensive especially when you're in a lack of labor? — Yeah, absolutely. Maybe um just before going into the why I'll explain a little bit uh what are the key workflows uh for the finance teams. One of the core ones is um monthly financial close which is um the core is job number one for um every finance team and um it could be abstracted into um three main steps. First step number one is collecting the data and entering the data into the uh core um system of the record applying uh business logic and also uh accounting standards. Once it's done once the data is in the main system of the record — um so uh the teams need to perform so-called reconciliation process which is in fact it's just verifying the data using the um u external um data sources. Once the data is uh verified accurate, it's all about uh telling the story and telling the so what um on this uh on this data that is in the systems. So usually the company is doing this uh every month. This is um best practice. If we think about this uh workflow conceptually it's the data problem. If ### [5:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw&t=300s) Segment 2 (05:00 - 10:00) you look back uh in time the businesses were different. The complexity of the different of the businesses were different. So um uh they were very much manufacturing uhdriven businesses where the systems of the record ERPs were uh very monolithic uh where um most of the supply chain and other parts of the operational functions of the companies were um in one system. So you go to the dealership by uh BMW then systems transactions trickling down to suppliers eventually just comes back to the ERP everything integrated in one place. Um and then uh while the companies were integrating the ERP they were um using the integrators and putting the um uh automation with the if else uh code logic in place. So what happened in the past a couple of decades businesses changed businesses became uh global — complex — more complex more global and um so it uh it changed uh the core systems uh companies started implementing a lot of point solution best of the breed solution for the different parts for the supply chain um for expenses for payrolls etc and um so what we started seeing is that um there was this fragmentation of uh data and So a lot of this if else logic uh moved into the most flexible and customizable tool that we have on the planet which is Excel. — Excellent. Yeah. — Um so and what we having today is um a lot of this logic lives in Excel spreadsheets with a lot of labor on top of that to manipulate this data and uh eventually close the books. I think what's interesting what you're highlighting there is that these businesses are becoming so much more complex yet kind of the finance function is still very much running on Excel. when we spoke to a couple of you know potential customers in the industry while getting to know you better uh we found it so crazy that some of these customers aren't even able to close month end anymore and uh that obviously gives you no visibility into your processes no planning security you don't even have to talk about a finance function that can be a value partner if you're not even able to deliver on the core promises of what you set out to do — it's pretty insane to think about that um in the world where uh we have s such a powerful uh tools uh so powerful technology a lot of companies cannot even uh have a full visibility uh of the past uh not uh to speak about the um the future — yeah it sounds like a perfect storm why hasn't kind of agentic AI arrived to that function yet in a bigger fashion — it's actually interesting we've been talking with you a little bit about that um as well if we look at the um other industries where we see AI I penetrating much faster like um legal customer support. I think there are a couple of differences there. First is on the data side of the things and then second is on the context. — The data structures of the finance teams are a lot more complex than um in legal. First of all, um legal it's uh very uh tax based — and also very much consolidated in one place as opposed to finance where um data is fragmented and it's also numerical data. That's one. And then second is the context. Um the context allowing um to teach this agent systems so they can learn and become smarter. Again, in the finance world um the context right now is very fragmented. it doesn't live anymore in the ERPs because um teams are doing the work in multiple different systems including uh Excel, Slack, uh emails uh and many other tools. So uh there's a fragmentation of the context not allowing to really sort of bring it together and learn from that. All of this makes the um penetration of the um automation much harder in uh in the finance world. But at the same time uh for those who will solve this fundamental problems in this space there's incredible opportunity uh because the space is massive and the impact that you can make is huge. — I guess this is a great segue to talk a little bit about the role of the earpiece in this whole um uh new world that you're describing earpiece used to be the core tool that CFOs were paying for. It used to be the tool that was the biggest spend that they were actually having. Everybody was talking about this is the single source of truth the core ledger that you're having. What is the role of an ERP in this new world? — As I already mentioned, what we've seen is that um uh a lot of context, a lot of work moved away from the ERPs today, but ERPs are still a very powerful source of the record uh within the companies. So basically um again just building uh a lot of automation um that data that is uh within the ERPs is incredibly important but then the context that is on top of that data is incredibly important as well. So in order to build ### [10:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw&t=600s) Segment 3 (10:00 - 15:00) a system that is uh truly making change and delivering ROI uh for the customers uh one has to bring together context where the people doing the work and the data together. you um mentioned uh this in one of our conversations as well um some surveys with the CIOS that um there's increased trend of uh finance teams thinking about upgrading their systems and I think it's truly signaling that um some of the systems that are in place are not fit for uh business needs today — but at the same time um what we're hearing and seeing in the big uh enterprises changing those systems is very difficult as someone was saying is almost like uh doing a heart surgery on — open heart surgery on the athlet and in many ways uh so it's not needed um what is needed is a quick ROI and real uh use cases that you can solve immediately — I love that analogies you do in an open heart surgery and then immediately after jumping on a treadill to run on the track yeah it's really great I think it's really encouraging for CFO functions that they have to have obviously a great kind of core infrastructure uh on the ERP layer, but they can drive a lot of the efficiency gains and the insights that they really want to do by building kind of around that core with a value adding layer such as stacks, I guess, which is a good segue in talking a little bit about why we're sitting here today. Um, and that is obviously you announcing your series A investment round. And what makes me happier about this announcement today is that this is not the spur of a moment that we're sitting here, but we've actually gotten to know each other over one and a half years uh pretty well. Um I very vividly remember our first day um of having a walk in rainy gray Berlin in a spring day. Um, and then I very vividly also remember the moment that we spent outside of the office where we brought you to our kite trip that we together had and uh there's a great picture of us in here that we can look later on as well. But uh but maybe kind of taking us a little bit on the journey of stacks and I know you know that company is uh still relatively young but you kind of came from a very impressive background. you've built um the core functionalities uh in Uber's uh finance teams. You've helped Plaid to land in Europe and you saw these problems in the broader finance stack today. So Albert, help our viewers understand what was the vision for STAX at that moment when we met and how has that evolved into what STAX is today? — I still remember um our walk in Berlin when we were talking about the um the space at that time. uh what we knew is um since I was a part of the Uber finance team uh working on as a product manager on automation side um of the finance. So it was clear that this is uh a massive space. It was also clear that uh the problem um uh and the manner of mundane work in the space is very acute. And uh finally we were living in this world of new technology emerging that giving a hope that we can completely um rethink uh some of the solutions there uh grounds up. So we had this longerterm vision but we didn't fully have uh a view on how we're going to get there and where when we're going to start. — Yeah. In the case of the Uber, we were building almost 60 businesses at the same time um because every region was a business and decision- making was uh very um decentralized. So we had GMs that had uh had to have um had to do a lot of decisions and they needed um data on their fingertip fingertips with a lot of this data coming from the financial data as well. At Uber it was in a way um uh solved by a very uh data obsessed teams and with access to the data but not every company can do that and for us was really uh a northstar how we can really bring all this finance teams in the world where they have financial intelligence at the fingertips so that everyone in the company can make uh better decisions uh whether they invest in one product another product region or um another region. But to get there, one thing we realized that um it's fundamentally data problem because data is segregate uh data is everywhere and it's difficult to um build intelligence without having a good uh data layer. But you cannot build that data layer from day one and um so uh we decided to go and talk to the customers do research. We uh spoke to hundreds of customers across uh the world to really understand uh how can we get there what would be the best wedge uh through bringing efficiency our why to the companies as quickly as possible and it was clear that uh monthly financial close was that uh workflow uh that uh is very manual where more than 60% of um labor force is concentrated ### [15:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw&t=900s) Segment 4 (15:00 - 20:00) and where AI can make a huge uh difference. So we decided to focus on this area and interestingly this area is also allowing us to get the view of the data more horizontally and really just get us closer to that vision. So um focusing on uh very immediate ROI we decided to start uh with this uh workflows uh related to the journal entries reconciling and then work our way into um just making companies a lot more uh intelligence in their intelligent in their decision- making. What I found so impressive when I saw you embarking on that journey and you mentioned that just very casually right now is like you were obsessed about your customers. You had literally done more than 100 customer interviews at that time and really figuring out what is that best wedge in and at the same time you very much focused on still doing the hard pieces which is building the underlying infrastructure and data platform for stacks. So describe what does stack look today like and how do customers use that product. — Yeah absolutely. Um so uh our product is structured on uh three fundamental layers. Um one layer um that I just mentioned it's a data layer. We call it uh AI ready parallel ledger. Second um layer in our system is deterministic tools. Mhm. — Um it's machine learning tools such as transaction matching uh transaction matching categorization and other different um uh data analysis tools. So finance is a space which uh is um uh very sensitive to the accuracy. So that's why we're using this um deterministic tools where like we're doing the math. And then uh the fin to be right about that actually. Yeah, — absolutely. And then the third layer here is um our workflows. We call them agents. uh the agents orchestrating the work uh the workflows. This is the interesting part because um uh what we've seen at Uber is that a lot of um workflows are not linear and you can automate 95% of um uh some of the um reconciliations. But then uh a lot of time is spent on that last mile and last mile uh is really um uh less deterministic. Uh depending on the exceptions you might actually want to trigger different workflow and that's where agents are very uh powerful because they learn um uh from uh the context uh they learn seeing how accountants uh doing the exception handling and based on that learnings they um can um start different workflows. So those are fundamentally three layers that allowing us to um achieve much better automation uh and also accuracy uh with our product. — I think STAX will stand as a playbook example on how you actually use deterministic workflows together with generative AI tools to really bring value to a function where ultimately precision and reliability matters a lot. Um I'm I'm curious to see and write that case study together one day. — Yeah. It's in this space uh fortunately or unfortunately you cannot just throw uh data at LLM and hope for the best. Um so that um fundamentally foundational layers on the data side and then on deterministic uh machine learning tools are incredibly important and um the way we uh build the um uh this deterministic tools are very much AI first. um they can be used by uh teams uh to complete the workflows uh in semi automated way or they can be used by agents um that can do uh tool calling to execute workflows as well. So um that's uh that system u makes our platform unique because it's learning over the time and over the time we're pushing what's possible from automation uh to the new levels. When I bring down your concepts of how you think about the product into real life examples, can you walk us through how customers are actually using the product and what impact that creates for that finance function? — Yes, absolutely. Uh there are three u most important parts um uh of our workflow of our product. The first one is I call it um a journal entry. So like entering the data and um so be in the world before uh take uh as an example payroll um you um as a company using five different payroll providers and um uh teams were pulling together this data um in Excel spreadsheets and then um manipulating the data and creating a CSV files entering into the ERP system. So we fully automated um this workflow. Uh so now you can do that uh with a few clicks um in our system. So that's one. Second is um uh reconciliations and um one good example there is a bank account uh reconciliations. um especially for ### [20:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw&t=1200s) Segment 5 (20:00 - 25:00) the customers with the um millions of the transactions, this could be a big Excel files with the VLOOKUP uh functions there that takes a lot of time. And then finally like once you have all of the data there um uh telling the story and understanding why the spend is 20% higher this month. — Uh understanding what actually drives that is something that uh requires a lot of work looking at the transactions. So with STAX we automated this process again with a few clicks um you get the uh detailed transaction level explanation uh — for your flux analysis — for your variance and flux analysis. Exactly. I think the good example uh here is um the company called Noda. They are a trader of uh diamonds uh incredibly international business uh with the more than 13 entities 130 uh bank accounts. Wow. — So when we started working with them, some of the processes were uh quite uh manual. What we were able to do together um so they were spending um more than 8 hours uh for uh journal entries. So we cut down this time to um uh less than 10 minutes uh together. They were spending a lot of time on u reconciling this uh 130 bank accounts with the thousands uh even more hundreds of thousands of transactions there. 95% of this work is now fully automated and um so we're also rolling out our flax u uh product with them as well and already seeing a huge reduction of the time uh spent on this. All in all, all of this allowed um the nevada team to reduce time for closing the books for eight days. Just imagine that you can have a view in ter in your uh financials in your data 8 days uh faster. That has real economic impact uh for the business um despite the fact that you finally can spend time on value creating tasks versus these annoying tasks. Um that's a really great example. I want to switch gears a little bit um and talk about the people at STAX and the culture that you're building. Um in fact, I'm going to look at our memo and I'm going to highlight we've written in the memo um that we think you're an incredible productdriven team that strives for excellence. Uh and that also means you ran without an engineering lead for almost all of the time at STAX. Um tell us a little bit about like what are you looking for people that are joining stacks and uh what makes your culture so special that someone like us is describing it as a very high density talent organization. — So I do not recommend um running a team without engineering for as long as we did. But we had uh an incredibly high um uh standard and bar for that person. So we didn't want to compromise. I'm very glad that we have this person already um uh joining our team. Uh I think in many ways um the talent density uh comes from experiencing uh this in the different environments. For me that was again during my times of the Uber and there were like a couple of things that really um amazed me at Uber. First of all it's uh ambition. every time when I thought that I cannot be surprised by the ambition of the company uh someone would come and just uh put it to the next level. Second is um ownership. um everyone was an entrepreneur inside the company and that's why uh what's uh what I'm seeing right now a lot of people I work with uh they running uh they started the companies that are quite successful around the world — and then the final thing is craft — um so having like a high standard of the craft uh for um uh roles such as whether it's design engineering uh was very important and those things that um really sort of after seeing once you don't want to compromise anymore. And for us specifically at stocks uh what we were looking in everyone who is joining um uh three main things. First is the motivation because working in the early stage company requires a certain level of um uh — drive and you know you need to be um in a way like a high achiever — and just prove to the world that you want to do more. — Yeah. So we asking the questions and trying to uh get the signals that um uh the person joining the team been showing this uh before in the previous career. That's one. Second already mentioned uh the craft incredibly important um so especially for the roles uh such as an engineering design. Design is not just shiny interfaces. Um is like how do you think about UX? how you break down the complex workflows uh into experiences that delight customers. And finally, um it's uh analytical uh rigor ### [25:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw&t=1500s) Segment 6 (25:00 - 30:00) and problem solving. In early stage company, uh we would like to have people who love solving the problems. When they solve 10 problems, they're looking for another 100 problems to solve. It's a little bit like mathematicians that cannot live without a new problems coming in their way. — I love the talent density at Stacks. Um, a fun anecdote is also that you uh celebrate uh every new hire on LinkedIn and somehow every new hire has an outdoor sports picture in that celebration. Is there a special story behind that? I' I'd love it if it's you know our kite trip that inspired that but uh tell us what's behind this. — Yeah, so it's uh something that organically started. Well, first of all, what we wanted to do is tell a little bit to the outside world uh about our team uh outside of uh simply the work that we do. It uh just builds um the um uh the trust with our customers as well. just seeing that uh we have uh diverse uh people joining our team with the uh with diverse experiences, backgrounds and uh the interest somehow um many of us uh like the uh outdoor activities and there's a little bit irony there because um most of the time we spend in the office talking to the customers in front of our laptops probably it's a little bit of aspiration to spend more time uh doing outside of activities um at some point in the life. — Okay. So the next when I'm applying to stacks I have to send you a picture on top of a mountain and then maybe I'm going to make it through the recruit process. — We'll really have fantastic picture together uh and uh just guiding together. — I want to switch gears a little bit uh and talk about this new capital stack that you've raised. Congrats again by the way at Lightseed. We think it's an very exciting time because you're fundamentally rethinking how some of the deepest workflows are being done. And so it's time for exponential innovation instead of incremental innovation. Um, and what you're doing here at STAX is pretty much that. Um, and we're very thankful to be part of that journey your broader team in this. You have an incredible set of partners already on the journey. How did you go about choosing your partners for this? For us, it really uh came down to the three things. First, it was important for us to have a partner that is that has similar conviction in this space uh as we are. — It's a long-term decision and um uh through our conversations uh and interactions uh with the firm, — we saw that conviction. That's first. Second, we believe that this wave of innovation is very technologydriven. It's not just business model innovation. Uh a lot of um changes uh in the space happening uh deeply through the technology. Uh as a part of uh getting to know the firm uh we met more than 10 partners across the world and every conversation um made us believe that uh every partner is uh deeply technologist um at the firm and for us it's very important because that's how we think about the space as well and um we never uh got so deep into the conversation with other um firms out there. And then finally it's very much building uh and having the trust with the people that you work closely with and we uh known each other for um more than 2 years from now and had a lot of um uh touch points and conversation about this space that really built this uh trust and um I'm very excited to have uh you and Lighteed uh on the board in this journey of building generational company. excitement is equal and we're thank very thankful to be part of it too. What does this new capital stack allow you to do now? — One thing that I'm uh very excited about is accelerating the product development. So we have um uh a road map that's spanning 12 even more month and uh we cannot wait to bring um some of the products uh a lot faster to the market. We're living in um incredible time when uh technology changing so fast and the speed uh of bringing products to the market is super important. So um u most of the um investments will go into accelerating the speed and many of our products also becoming uh more powerful as uh customers using them more. So we um uh we're investing into distribution to bringing more bringing um this ROI, this innovation to more companies out there. Yeah, I love that you're so ambitious. You wouldn't have needed the money right now. You actually had a lot of the money that you've raised still on your balance sheet, but I think doing this now really enables you to kind of really bring those products to the market that all these customers are actually pulling out of you. And so um I'm excited that we are hopefully transforming this finance function even even quicker. Now tell us what does success mean for you at STAX? How does the finance function in ### [30:00](https://www.youtube.com/watch?v=ndJO9PhZ2Tw&t=1800s) Segment 7 (30:00 - 32:00) in 5 years look like if you've succeeded? One conversation that uh stood out to me uh when we were doing uh research even before starting the company uh controller at a global company told us that uh their team spending 80% of the time uh doing different data manipulations and only 20% of the time looking into the data and looking into the insights. So when we were thinking about the future and success for us is just flipping that equation um really uh just giving time back for the teams to focus on what really drives the business forward. What does that mean um uh in terms of how the teams might look like? If we look at the um enterprise uh company with 10,000 employees, usually you uh have 200 uh around 200 uh finance professionals there with 50 to 60% involved in um accounting workflows uh that are incredibly manual and then the rest of the team is focusing on um multiple other functions including FPNA. I think in the future um the uh the composition of the teams will change um and the uh the teams becoming more strategic and a lot of this manual work will be automated and team members will be spending their time on value added uh work that the work that drives the business forward but also that would mean that the skill u gaps and segregation that we're seeing right now FPNA person is very different to accountant will change as Well, because there'll be tools allowing to um do uh a lot of work much easier and what would really uh be an important uh skill for a finance person is really intuition around uh numbers, intuition around um how to use this numbers to drive the business forward. So, I'm very excited about this future because uh I think we're living uh in this transformational time. — Yeah, I think there couldn't be any better closing words. Um, I'm very thankful that we're going on this journey together. Thank you very much for being here today, Albert. And, uh, excited for the partnership to come and how we're transforming the CFO suite. Um, and by accident building one of the biggest and best businesses out there over time. Thank you for being on the show. It was great having you. Um, I think we should shake hands that you're going to sponsor the next Kite Trip when you build a $10 billion company. Thank you for being here. — Thank you, Alex. Thanks for having me. --- *Источник: https://ekstraktznaniy.ru/video/35168*