So you break through or are you at the US Auto right now?

I'm at US Auto Force right now. Yeah. So I currently lead finance for US Auto Force on the VP of finance here. So I oversee FP&A, I oversee collections, I oversee the traditional accounting function. We have a lot more like financial operations, customer onboarding, customer set out, and then AP with vendors and that kind of thing. So I've been in this role, you know, maybe a year and a half now.

Prior to this, obviously, you know, I was at VP finance at Breakthrough, which is a logistics technology company based out of Green Bay. I did that role for about four years. And then prior to that, variety of roles in treasury, credit, spent some time in commercial banking as well. So I didn't follow the traditional accounting path. I didn't go through public accounting. I took a little bit of a deviation from that, but maybe ended up in the same point anyway.

That's right. Which function do you like the most?

So I like, you know, what's great about this role is the ability to help shape and influence the strategic direction of the business. How we go to market, how we serve our customers, how we work with our vendors. And so I really appreciate that. And then I'm able to bring all the valuation aspects of the things that I know into the business so that people that I work with on a regular basis, they understand how do we translate this into enterprise value.

And so like that's the business partnering is probably the most fun part of my role.

I mean, I run Chatfin and my most favorite part is usually the problem solving. If it's marketing or product or technology, if I get involved, I like to sit on the problem, think about it and solving. And that's, I think, is what the most fun I get. The rest of it is all, yeah, I just go through that stuff.

You know, I worked in various functions, but one thing recently after we started working on Chatfin, and very recently I just came to know how big collections is a pain for various companies. And it's never a unified process and never a standardized process across industries and companies. So people think, we can build it. I can get a product. I can buy a product and streamline my collections efforts, or my receivables, but it's not. It's just, there's a lot of manual work that is involved.

One of the conversations that we are having, someone saying, it's not about sending follow-up emails to customers for payments, but it's when they actually respond and they ask more questions. Who's gonna get all the data? Who's gonna respond to it? That's the biggest pain point. I think AI is gonna solve some of these problems, but again, there is lot of non-standard procedures that we follow in companies, and there's no shortcut for it. We just have to go through it.

Yeah, I think you're right. You know, when I think about what happens on the collection side, and by the way, I hate the name collections. I feel like we need to come up with something different than that because it sounds so negative. A lot of what we do, I don't feel like it's collections. It's problem solving, kind like what we talked about earlier. It's how am I applying this invoice? How am I applying my payment? You know, where is my invoice? I didn't get this or I didn't get that. And that's the area where you do so much problem solving in terms of trying to run down, well, what is the issue?

And I think us, like much a lot of other organizations too, have information across disparate systems, right? So you got to go hunt and search to try to find the solution to the problem that the customer is asking about. I think where AI starts to become transformational is, can it help you with that process? Can you be the human in the loop where AI is listening to what's going on, saying, I know where to get that, I know where to get this, and it brings it to you.

So it's a much more efficient and productive conversation between you and the customer. And I think that's an area maybe where we don't necessarily think too much about how do we use it. But I think there's a tremendous opportunity there for us to leverage that among a lot of other areas too. I think a lot of times we just don't think about it and it's changing so exponentially fast. We're not really thinking about it and it's happening all around us.

Yeah, that's right. Again, I think models are already there. AI is ready for it. The biggest roadblock is the data sources. Our conversations are in email, in Slack. Maybe you're already sending messages to their phones for, let's say, an example of account receivables and collections. How do we get all this data together? How do we give access to AI, all these disparate data sources? That has been a big blocker.

I think it's the same for everybody, right? So we have to have a good partnership with the IT department. I don't think finance can do it on our own. So we need to have conversations with IT and come up with the right governance and data infrastructure to make sure that we set up the data in a way to be able to leverage a technology and leverage a technology that we know today and maybe even the stuff that we don't know that's going to come. And so we spend a lot of time talking with our IT folks around how do we manage this, right?

So I think a lot of times what we've built in the past isn't necessarily the best foundation to capitalize on AI for the future.

Right. Yeah. Collaborating with IT is very important. Like any other team, right? IT can be a blocker or IT can be an enabler as well. Again, it's for various reasons. I worked on both sides of the business and I know if you're in the IT and if you're looking at the finance needs, things change faster. There's prioritization and there's things like if you have to upgrade a system, right? You have to upgrade a system. There is no ROI of an upgrading existing system. Of course, a little bit here and there, but there's no support.

So things like that is kind of a conflict that happens with business teams and the IT teams. But I think a great combination. If they can work together, aligning with the goals, that would be a great outcome. And I've seen this already happening with AI. In few companies, the IT teams are, they are actually in the forefront of AI revolution. They are using various tools. They're looking at various AI tools, they're bringing it to their finance teams and showing, hey, maybe you can use this, maybe this will help us doing it. But that's totally up to how the IT team is structured. Do you want to be a blocker? Do you want to be an enabler in this whole AI transformation that is going to happen?

Yeah, I think that's a great point. You know, I just had a conversation with somebody on our IT team yesterday and they said, I can bring the business problems. So I think it's really a circular function in respect to IT brings the technology, right? What can the technology do? What are the capabilities? What are the constraints? How do I best leverage it? Me being in my role, I can bring the business problem, right? I know where the pain points are. I know where the inefficiencies are.

Not just within finance, and finance is a great use case, but even throughout the organization. So we need this continuous loop between IT and finance collaborating together with IT bringing forward the, here's what the technology can do. Here's how it's evolving, here's how fast it's evolving. And then we're bringing the business use cases. Where is the biggest value we can drive for the organization? Knowing what the technology is and the business use case, right? You can marry those two things together in that partnership and really create the transformation of value, I think that has yet to be done in a lot of areas.

Right. And Adam, I think one thing which you and me discussed before as well, there very few people in the company, and as the company size grows and they're very fewer, who actually understands end-to-end process and also knows all the nuances of it. And I don't think you can expect one person to know all the 50 different workflows that happen, at least in the finance team, right? And including the details of it. Do you think there is a way to document it or to know or to democratize this process data to everyone in the company?

Well, I used to think it was Knowledge Graph. So I worked with a gentleman at Breakthrough, our VP of IT there, and he presented this idea of Knowledge Graph. Some people are familiar with that. It's basically gathering all this data from everywhere. And I thought that was a solution, right? We tried to do a little bit of exploratory work around that here. So that could be an area, but it could just be AI, right? When I look at, you know, my Microsoft co-pilot, I jumped in there and it knows what's in my email. It knows the correspondence that I have. It knows the things that I've done on SharePoint.

And it can go out to the web and find the public information that's out there as well. And so that's been a big help of trying to bring together like, what are my specific processes? And then what are best practices that other companies are doing that's available publicly for me to take a look at?

That's right. Knowledge graph is a great idea. The only problem with knowledge graph is building them is time consuming and expensive. Because how do we get all this tribal knowledge, all the information, all the experience that you have in your mind and your team into a knowledge graph is actually very complicated. We can take process documents. That will help a little bit. But you know, not every company is documented.

Some companies, they don't even document any of their process. Some companies, they will do it. So that itself is a big problem in terms of how do we get the source of this knowledge graph and then build across it? That's why I think it's kind of a distant dream. Don't quote me on it. At least in the next few years, to think that we can build a knowledge graph of all the things that we are doing, even if you're talking just in per view of finance, bringing all the things that we do in accounting and finance into a knowledge graph would be a very huge task.

Yeah, I tend to agree with you, right? And much like we talked about the ROI earlier, that's one of those areas where it's, I think people struggle to see the ROI. Downloading my information into an SOP doesn't feel like it has a strong ROI, right? And there's plenty of other things for people to work on. And so I think they'd rather focus on that. I personally would rather focus on the other things than documenting what I do every day until we get to the point when I just download myself into the matrix. You know, I don't know that we're going to have an easy ability to capture all that tribal knowledge to be able to build the structure that we need for the knowledge graphs.

Right. But that also goes to the question, how do we ensure the institutional knowledge is not lost when a senior team member leaves? What do we do in those cases?

I think as managers, we have to take a look at where those critical risks are at, right? And make sure that we have redundancy and documented processes to make sure we don't have that. To the extent that we can automate a process or find a technology solution to eliminate the need for tribal knowledge, we should focus on that. We've had a lot of success around here with trying to deploy new technology. We've used some of the co-pilots. We've also done a lot of work with Python in terms of not necessarily just data science, but a little bit of what I'm calling desktop automation.

And that's actually been really transformational for us, for people that, number one, learn what we're doing. And number two, automate the things that are a time suck for them, that they maybe are not necessarily adding a tremendous amount of value, but have to be done anyway. And throughout that process, as you're building the automations, that's when you're documenting, here's how the process is working, and then you can kind of capture it that way.

It's, I think, at least a few years ago, we all thought Python, learning Python in finance would be a great value addition because you have the ability to automate a lot of your processes. But now it became much easier. I wouldn't say you shouldn't learn Python, but now you can use just text, which generates the code and which you can do it. And that's one of the features of Chatfin as well. I think you have probably seen it, is whatever, if you have a process and if you can document the process in very simple steps, and if AI can understand the simple steps, think about all the exceptions around it, and actually creates you a workflow for you.

Now, if you don't need probably a Python or even any expensive automation tool to manage it, you have already automated just using text. I think that's going to be a big thing as we move forward in this AI world. All the business logic is going to be in text. And as people in finance, it's our responsibility to manage the text.

I think that's important, right? When you think about the AI tools, right? Python, so a number of years ago, I started learning Python, right? Because I'm like, I think this is something I need to know. I started to see the benefits in finance of using this tool. Most finance people were stuck in Excel all day, right? Like that's the tool that we use for everything. It's the hammer that we use for every nail. I started to see these benefits of using Python, but then I quickly learned it was kind of at the same time that a lot of the AI was converging, that I didn't need to learn how to write code.

Having some basic understanding of it is great, but then having AI tools to help me write the code is the way to go, for me anyway. That way I don't have to really know how to do it. And the text piece, the conversational aspect of telling the AI, I want this or I want that, or help me solve this is really huge.

Right? See, converting our complex business process directly into Python or text is actually difficult. Because if you're thinking, let's say you're making a reconciliation template, building that just in text, all these steps is actually very complicated. The easiest way is to build an Excel template. Whatever you have in mind, just build a mockup template and AI probably can understand that mockup template and all the formulas that we built and how you did it, and also add some of your company context and can build an automation over it.

So that's actually easier way to take our process and automate it, rather than thinking directly a finance user will talk to a technical person or even by themselves converting that into a text. I tried it and it's very complicated just directly converting it into text. So the best way is to convert that into Excel first and let AI read the Excel and come up with its own process.

Which for finance people, that should be pretty easy. I feel like Excel is the easiest thing to convert in. As you think about the future of your product and your development of technology, how much is it that the technology is just gonna do the work for you versus you need to tell it, I wanna build this, it's gonna complete the rest of the steps for you. Is that in the future at some point or is that still farther away?

I think the honest answer is I don't know. But one thing that we have seen is every company, and if you take any process, and I'm talking about mid to large companies, there are nuances that you need to consider before, because you can directly ask an AI agent, can you do this reconciliation for me, this validation for me? It'll do it and it'll give it to you, but that output is not useful for you. Just in terms of because you want it to consider a few of these points as well. What it means is that you have to give that context in terms of what you need and how you want to do it.

And that becomes much more robust process than letting the agent run by itself and figure out a process. And so far with all the companies we have worked with, there is no process that can come out directly from an AI agent and it can do it and people will like it. It'll do it. It'll figure it out. It'll do it. But you won't like the output. You will say, this is good, but this is like an entry level finance person who comes in and does that work. But you're like, yeah, you're right. But you have to consider these things. You have to do this work as well. This is how you group them so that you can think about this data.

All those things are the nuances you have to add into the workflow. That's why when people say AI agents are going to take over, they are going to take over, but only when you give lot of context to it. And now to give context, you need a person, you need a subject matter expert in that process to do that. So is AI going to take jobs away? Probably yes, but you still need a lot of expertise in that specific process to run that agent, to give that nuances, to ensure that output is correct. So in the future, this is how it's going to look, right? There will be less number of people, but these people will have AI assistance to execute more work. And that's what we're seeing right now, at least for the next 12 to 24 months.

Yeah, I think that makes sense. We've done some work with AI agents around our sales data set, which has been interesting because the way it's coded in the system is a site name. But then I go and I interact with the agent and say, hey, which warehouse should I look at? Agent comes back and tells me, I don't know what you're talking about. What's a warehouse? I don't have a warehouse. And so it didn't have that reasoning ability. So then I had somebody on my team go and tell it that a warehouse equals the site name. And so that's how we solve that.

But that's me, right? And then how do we move forward from there? And I can't come up with a database of aliases for everything, right? That's super inefficient. And so how do I build the reasoning or the logic or the interpretation, right? The context that you and I have around, hey, site name means warehouse. How do I teach it that, you know, in mass so that it just knows that in the future? And then on the flip side, how do I manage hallucinations? Because that's the other thing where it spits out data so confidently, like this is correct. And you go back and double check and you're like, hey, are you sure? And the agent's like, sorry, I made a mistake and that was wrong.

But the level of confidence in the first response was very high, so much that some people might just believe it. So how do we manage the hallucinations?

I think as part of AI product building, that's what we realized, particularly with finance teams is you need to have a way for people to review and ensure that it is correct. Even if you're scanning just an invoice and you're getting all the values by scanning an invoice, how do you know if that number is correct? Sometimes four can be read as six. How do you do it? So there should be a checkpoint where a human, where AI can alert him, hey, I'm not sure about this number. Can you check? That's the loop we need to build in.

And that's what we do it in product development. Again, this goes to the same thread as, can AI, can a large language model by itself run the whole finance? The answer is no. You still need all these guardrails, checks, human in the loop process. And again, a lot of people think, what is human in the loop? Human in the loop is not just about process, it's also about the product, right? How does a product involve a human to make sure everything is right before proceeding it to the next level? That all needs to be built as a user experience within the product. So there's a lot of learning that is happening within the AI world at this point of time. It's not just the model training and model doing all the work. How do we use that in our real life use cases?

Yeah, I agree with your sentiment on it's not going to replace the finance department. The way I've been thinking about it, it's going to make us more efficient and more productive, right? And that we still need to be the translators of the information in terms of how does value need to happen? How should people think about value for the organization? I can have the AI agent build me a discounted cashflow, but does it have the context, right? Can it tell people outside of finance? Well, what are the things that I need to think about to create exponentially more value?

Do I have the right discount rate? Do I have the right tax rate? Yeah, you can plug those in, but I think that's where it helps eliminate the stuff that maybe you don't want to do in your day. It can solve some of that and free up your time so that you spend more of your time doing the things that are the most productive, the most value adding for you and for the organization. At least that's what I've found. And there's been a tremendous opportunity for us to automate some of the minutiae, eliminate some of the manual tasks.

And people on the team have said, what happens with AI taking jobs? We haven't done any of that because there's more work. A lot of these organizations are growth-oriented organizations. So you can scale better by leveraging some of these tools and technologies. You can deliver better insights, better information, better guidance and support to everybody outside of finance. And so really, that's how I've been thinking about it. And I think we've been able to really transform that and make that happen.

Right. Adam, how are you seeing the whole AI transformation that is happening? How are you managing it? Is it top down right now currently in your organization? Or how is the whole AI adoption happening?

Yeah, so we have a strategic initiative at the corporate level around AI. So there's a lot of stuff happening centrally. You know, I feel like the IT department is really the feeder for the technology piece of it. And my team is thinking a lot about the use cases, right? What are the things that we think we could use this for? And those are more of like the bigger, what I'm calling maybe more home run type things. Then there's the element of just general education around the personal productivity with AI. How do you have it help you in your day-to-day stuff within Excel or within Outlook.

And so we have just some education that we're pushing out to the team so they can be aware of how to use and leverage these technologies in the daily productivity aspect of it. But we have to be careful too that AI is not a solution to every problem. There are other solutions out there and AI could be a solution, but it may not be the best, most efficient or effective solution. So we're thinking about it very strategically at the organizational level. How do we bring it in? How do we leverage it as a competitive advantage for us? And then how do we make ourselves better from it?

That's right. And I don't think there is going to be one AI that's going to solve all the organizational problems. Supply chain versus finance versus marketing, they all need different kinds of AI to solve their problems. That probably would be, for example, even if you take Microsoft Copilot, bigger complaint about it is that people still haven't used it to the point where they see a significant gain from it. Yes, it's helping in daily work and it's across the company, right? It's not just for finance teams, it's across the company, it is helping.

But what are the specialized tools that can really help the specific teams to get their work done? And the whole ecosystem is also evolving in that way. Even though I think at an organizational level, you need a few tools to do it, but again, within your function, you need different tools to do it.

I think you're right. I think there's a lot of opportunity in more niche or specialized functions, right? Like pricing and inventory management and things like that, that I think maybe a general AI isn't going to be able to help you with because there's such a specific nuance around that that you're just never going to get there with a generalist type AI. The generalist AI can help you with the, format my PowerPoint, right? Help me write this email. Those things are more suited for that, but then you need the specialized stuff, I think, to really drive home the value.

Right, how much ever we force them, right, how much ever we force a generic tool on a team, I think the productivity will at some point won't be anything better.

I think if you're not leveraging these tools at some point, you're going to be left behind. Right. And so I think this really is the wave of the future. I think you have to really understand how to embrace it and have an abundance mindset and a growth mindset as opposed to a scarcity mindset. Right. These things are going to make us better and faster. And we got to think about how do we leverage them. And it's not so much of I'm going to be better than anybody else if I use it. But if you don't, there's probably really a competitive disadvantage if you're not able to really capitalize on that.

It's at a different level. Even for a company that's not using it, they'll lose it to the competition. And even for the individuals, for executives, you need to be aware of where AI can really add value. So you have to have your own kind of experimentation going on with your processes, with your daily work, with your processes, with your team. At the same time, the teams need to also do it. It's for their competitive advantage when they go into the job market, saying that, I used AI to do this work, so I'm pretty good with it. So that's different levels of competitive advantage happening within an organization once they start using AI.

It's gotta be widespread, right? And I like the idea of being a continuous learner, right, and experimenting, right? So create space to experiment, create space to continuously learn, because it's changing very fast. Figure out how your IT department can help you, how other sources can help you learn about what's happening, and then try it out. I'm a big proponent of trial and error, right? I play around with the technology, I try to figure it out, I use it at home, I use it at work, right? For me, the best way to learn how I can use it for work is to use it for a number of different things and see what it can and can't do.

I work in finance, I work in AI and I haven't used probably like 80% of the products in the market yet. That makes me like, I always say three days of a week I'm very positive, happy on the clouds about AI and two days I'm just depressed that the kind of evolution that happening every day, there's new products in the market, it's kind of overwhelming.

It is, and it's hard to stay up to speed. So I remember like my early days with AI were, you know, pre-COVID, right? But it was like using chat GPT to generate a picture for me or something like that, right? It was very basic, you know? And I'm like, I didn't see a lot of potential. And at first they said, hey, it's going to really impact marketing and graphic design people and more of creative aspect of it. And I'm like, okay, I'm safe in finance, right? We're still going to do what we do. But as you've seen it evolve, it's like, hey, it's not... Not just for one thing or the other thing, it's really a holistic tool you can use in a number of different ways for anybody and kind of anything.

Right? Have you started calculating ROI on any of these AI initiatives?

Not in a sense, yes, right? So we have deployed some specific AI around specific processes where it's easy for me to quantify. Here's my cost associated with what I was doing before. Here's the time that I saved on it. And here was my cost to deploy the technology. We have a couple of those that are pretty clear, cut and dry. It's easy to follow the use case. On the personal productivity front, it's a little harder to quantify the ROI.

You know, if I save 30 minutes here and somebody on my team saves 30 minutes there and it records a meeting for us, those are harder to quantify. They are beneficial, but we're looking at, where do we deploy this in the best ways, right? What, when I think about our team's comprehensive time allocation, like where are the heavy hitters that consume a lot of our time and also have good use cases to be able to deploy the technology with as well?

Yeah. It's a... Now ROI becomes, because in finance ROI is the key, right? It's the key metric to sign up for any project or any initiative or any spend. Now, an ROI is not really clear with any of these AI implementations at this point of time for the reasons that we just discussed, right? Then how do we do it? What's a better way to prioritize the project if we can't really, and I know we can always say, hey, maybe we'll do a better job at calculating our Y and those numbers are just going to be guesswork. Right. What's the better way to do this?