S4 | E12 | Why ERP Is Broken and How Doss Is Rebuilding It for the AI Era - Wiley Jones - Doss

Speaker: 00:00:03

Welcome to the Think Data podcast brought to you in partnership with Mydataworks. If you want to stay up to date with the latest breakthroughs and trends in the world of data and artificial intelligence, and if you're curious about some of the strategies that companies and founders use to launch data and AI products, then you're in the right place. Our aim is to bring together a diverse lineup of fantastic guests from the founders, through to accomplished leaders and product owners at some of the most fascinating data and AI companies worldwide. They will each offer you their own unique insight into what it takes to launch and scale a great data business. Thanks for tuning in and I hope you enjoy the episode. Welcome to the Think Data podcast and on today's show I have the pleasure of speaking to Wiley Jones. He is the CEO and co-founder of DOS. They are a super interesting business. They're a Series B hyper-growth startup in the world of ERP. Typically, that space is seen as quite legacy, normally associated with older, more clunky companies. But DOS is building the modern operations stack for physical product companies. So it's going to be AI native, really fast to deploy. And it's ultimately designed to replace the chaos of spreadsheets and outdated ERP. So yeah, really good to have you on. And obviously, congrats on the Series B. That's super exciting. And I mentioned at the top, obviously, CEO, co-founder, but not many people have heard of you. So what's the backstory to Wiley and what led you to launch DOS?

Speaker: 00:01:44

Great to be on. Thanks for having me, Alex.

Speaker: 00:01:45

Pleasure.

Speaker: 00:01:46

So DOS is a, I think, an exercise in pulling a thread and the world continues to unravel. And you realize you're... onto a much larger thing than what you maybe started out to work on. So my background is in hardware engineering. I worked as a mechanical engineer, an electrical engineer. I wrote a bunch of firmware. And in order to make physical products, you have to track the information of the supply chain. And so I spent the better part of a few years actually living in China, working in manufacturing, making products, getting them shipped out to people around the world. over the course of, you know, the next few companies I worked at, I kept seeing the same problem, which, you know, I would say for a bit of a flippant point of view is that it's a miracle that anything gets built at all. It really is. It's a miracle that products end up in the grocery stores or they end up in consumers' hands or that Amazon package ends up on someone's doorstep. And a lot of that miracle is in part to the information systems and basically what a lot of people would call the ERP that power procurement, inventory management, order management, all the pieces of the supply chain fulfillment. And so seeing how hard it was to build and architect these systems around what our company actually did, it really made me want to go start at DOS. And obviously, I think one of the biggest catalysts for this, and obviously the reason I quit my job and started the company was I saw the advancements that were coming with generative AI back in 2022. And it became really clear that businesses were going to be able to adapt their applications to serve the strategy of their company. You know, nowadays we have a fun term for that, vibe coding.

Speaker: 00:03:34

You know,

Speaker: 00:03:35

you're making the application meet the vibe of your strategy, whatever it is that you want it to be. In traditional ERP, that looks like, you know, multiple quarters, if not years of implementation. And that is a...

Speaker: 00:03:48

been a fundamental assumption that is now being revisited and challenged in the market and the industry and that's what we focus on yeah super interesting i love the fact you went to china and kind of uh as in your own admission you know almost like scratched anything and how does anything ever get shit but what i find wild is some of the largest tech companies on the planet which obviously i won't name names here are in the erp space you know they are vendors that have grown, multi-billion dollar profits, huge organizations. Obviously, you're competing, not necessarily directly, but how have they got this so wrong where it's almost allowed an opportunity for someone like DOS and obviously you and your super smart co-founders to get in on this?

Speaker: 00:04:29

Yeah, I think there's like actually a kind of a base assumption that you're making, which I would, I don't want to say challenge, but I would clarify a bit, which is that it is legacy software that is currently in the market. But legacy is a good word. Like people forget that. It means that companies did something right. And the fact that this software is powering the critical operations of the world's largest companies, it means that they understand something about how businesses should operate. And they've encoded that into an application. And where they got a few things wrong. I think is a little bit more of a consequence of the architecture that they were working with and the tools they were working with at the time. The mainframe era, IBM computers back in the 1960s and 1970s, we look at that and say, man, this is a clunky piece of tech. It's like, yeah, but in the 1960s, the alternative was people crunching those numbers by hand. And, you know, and the same thing we can say so much of the genealogy of these systems. Our view on this is that all we're doing is we're just pushing the frontier forward. And there's kind of this, you know, archaeological survey we have to go through and look through the large enterprises of the world and say, where are your outdated systems no longer serving how your business needs to be adaptive and agile? And it sounds really boring to say it that way, but it's really just how do we help you make more money, expand your visibility into your profits and your margins of your business, your efficiency and your leverage? and use technology to make better decisions. And we are just a part of this, you know, next chapter in the journey, which is how do we make these systems faster, easier to operate, more legible to humans, and now also more legible to agents and AI systems that can operate for people. So that's really, it's really, I look at it as almost just like taking the baton and continuing to move forward with some of these new architectures that are now possible.

Speaker: 00:06:21

Yeah, you touched on a few really good points. And I guess what these companies have as well in space is obviously the most the brand, the reputation that, you know, these organizations have got to a certain size because they're obviously doing something right. You know, they're, you know, they're obviously delivering something which is adding value, but obviously a lot of, you touched on a really interesting about legacy tech, legacy applications in these ops heavy organizations is, is as I think you said, right at the beginning is unpicking that thread where everything's kind of wrapped around it. So can you talk to me about how DOS... does it or how you do it which allows them to carry on operating but obviously if you're going to take ultimately rip out a legacy erp and the associated technology and infrastructure around it to go with a relatively new company how

Speaker: 00:07:10

do you convince them to do it and how does it work yeah so there are probably three or four really good analogies for this that all convey the same point which is that it has to be modular and it has to be composable. And those are two different things. Modular means that we can package up the concepts of let's call it order management. I have sales orders that I process from three or four different channels. I sell in retail stores, and they're not my retail stores, you know, it's a big box store where you'd go to like in the US, like a Costco. And then I sell online my own, you know, online presence, you go to, you know, nike.com and you buy from me. And then also I sell directly to businesses, I sell into, you know, these B2B channels. Order management as a concept means different things to different people. So I have this order management package, and that's a module. And inside of that, I have the ability to say, well, I'm a company that doesn't do B2B wholesale. And actually, the way I do my retail is I operate it myself. I do retail not through a third party, but I do it, you know, we run our own cafes, like one of our customers is a coffee company. And they have a really complex order management operation. The ability to combine composability and modularity allows you to go to businesses and actually meet the shape of their strategy. Instead of saying, your definition of order management in our order management module isn't the same as us. Therefore, you adapt your process or we deeply customize the software in a one-off manner for you that will not be able to be upgraded or maintained by anyone other than our team. That is the cardinal sin of these legacy systems. And... And it's, you know, I don't blame them for it. There wasn't really a way to solve this problem until we had JetBrains AI. And so what DOS has been able to do is make it so that businesses have well packaged, you know, let's call it 80 or 90 percent fidelity modules. And that last mile of customization can be done through a symbiotic relationship of our team, our customers, and basically AI automation that helps us understand the footprint in their business and model the application into that form factor. The combination of those two things is really a core differentiator. And it allows us to take risk from here's this whole massive surgical operation where we're going to pull, you know, five different parts of your business out and swap them out all at once. Instead, we say, OK, you have a freight and fulfillment problem. You have four different warehouses and we want to be able to fulfill from them across one really clear network of fulfillment. All right, let's go and put that part of the system in and we'll keep the other pieces intact until we can go in and extract each one individually.

Speaker: 00:09:49

Yeah, interesting. So yeah, as opposed to attacking it all together, it's actually picking on those individual use cases. And then do you find over time with those companies, then other opportunities present itself anyway, naturally, because ultimately, you've got a freight and fulfillment problem, but equally, we're now seeing the potential purchasing issue or a potential supply chain problem.

Speaker: 00:10:10

Yeah, there's two parts to that. Part one is like, you have to be a good partner to your customers.

Speaker: 00:10:16

Yeah, they have to trust you, right?

Speaker: 00:10:18

Yeah, it's trust, but it's also like trust and competence. I think, especially folks on the other side listening to this, especially leaders and people who have spent time managing teams and building teams, the moment you sense competence in the parties around you, you go, great, let me give you more work. Let me give you more things to do. Because everyone has this infinite list of, you know, What are my priorities? I have to choose 15 things that I can't do today. The moment they sense that they can trust us with critical operations in their company, they immediately come back and say, great, we solved this problem. What about these 15 other things? Because that's not, you know, for our coffee customer, they're focused on how do we go and expand our brand? How do we build better products? How do we go in and source the best coffee? They're not thinking, how do we optimize our footprint of our data platform? And how do we make sure that our fulfillment network is operating at the exact way it should? They want partners who they can trust to go deliver on those things. So immediately delivering with a competence partnership, they just go and push more work back to us. The second part is because the product is composable and modular, we actually get to say yes. A lot of companies will say, that would be great, but that's not what we do. Or that would be great, but... we're focused on x y and z that is much more in order management and unless you have this shape and form factor um you know we can't actually match the pattern of the puzzle piece that you're describing and so we've given ourselves the right to say yes to when people ask us to solve a problem what

Speaker: 00:11:50

a good position to be in right you kind of have that confidence and strength and conviction no way i know egon flips i'd know when to say no you know kind of yes yeah Because I guess as a startup, it's easy, not easy, but it's you want to say yes, right? Because if yes leads to revenue, leads to opportunity. But saying no is probably just as staying in the right lane is also where you're going to get more credibility from the market.

Speaker: 00:12:11

Yeah. A couple of notes on that that I think are super important for us is like we wanted to prove really early in the company's journey that we could be horizontal and sell a product that worked to a wide group of people, not just like stakeholders, but and not just geography, but. the individual applications. We have a company we work with that's based in Israel that's a demolition business. And then on the other hand, we work with a coffee company. And those companies have almost nothing in common. And the product can support that. Now, you will not see our organization going back out and selling to a bunch of demolition companies right now. What we did early on was we proved out and felt out the shape of what was possible inside of the system and from a product standpoint. And then from a go-to-market standpoint, we have controlled and closed the funnel a lot by going and saying, here is our ideal customer profile. Here is the shape and form factor and firmographics of those businesses. And we can deliver an outsized amount of value into that market very quickly. And that's where we want to go build our brand legibility. But on the backside of it, we know the product can support these 15 other things. We're just not going to focus on that right now because... We want to be laser focused on go to market and extremely unopinionated and flexible on the product.

Speaker: 00:13:30

Yeah, no, no, I think it makes complete sense. And there's no surprise you've had the success you have. And I'm keen to drill down on this kind of term that's quite widely used and in some senses quite probably overused, AI powered. You know, people are talking about there's AI. And I know you said three years ago you saw this opportunity that may previously have not been there for you because of generative AI. But how does... the AI work specifically within DOS? Because obviously what you're talking about here, there's legacy technologies, there's legacy, you've got a demolition company, you've got a coffee company, they all got different setups and different use cases. But where does the AI come into this and how does that work in practice?

Speaker: 00:14:10

Yeah. So I'll explain, I'll answer your question with a little bit of a history point of view. There were three clear generations of enterprise applications. When I say enterprise applications, I just mean systems that people use to do work in a commercial setting. So, you know, the government, businesses, things like that. I'm not talking like Instagram and Facebook. I'm talking more like, you know, Salesforce. And the first generation of them was mainframes. You had a big room with basically a room-sized computer. And, you know, the U.S. Census and Wells Fargo used it to crunch numbers. And those were basically the only people who could afford them. And then we moved from mainframes to PCs and you had networked PCs in an office building or in, you know, some local networks and maybe some global networks. And people were deploying, you know, networked applications. And then we moved it all to the cloud. Each time we went through those generations, new companies get built. because there's a new architecture that enables new things. So IBM, SAP on the mainframe era, Oracle, PeopleSoft, Siebel systems in the client-server PC era, and then ultimately into the cloud, you see every time these applications actually have to get rewritten. Siebel moving to Salesforce, PeopleSoft moving to Workday, Oracle and NetSuite, and then ultimately they bought NetSuite back. Those companies had to go in and rewrite their entire system to take advantage of the new architecture. And what a lot of companies are saying, they're like, we're AI powered or we're AI native. I think the market is really going to clearly identify what AI native means in a few years, just like we the Internet. It's it would be like a company in 1995 saying we're Internet native and they run all of their servers on prem. They do everything. You know, they're still using CD ROMs to go install software. They're like, but we have a website. And so we're we're Internet native. Right. And there's a lot of companies doing that right now with AI. And, you know, they'll say, oh, we're AI native or I'm fluent in Chinese. And then I pull out my phone and use Google Translate to translate Chinese. And then I read it back to you. I'm like, no, no, I'm not fluent. I'm going somewhere else for the knowledge. And I think that's where when we think about AI native, that was the view that we brought to it as well, which is that the intelligence actually has to be embedded inside of its understanding of how the system functions. This is why coding applications have done such an amazing job of being AI native, because the agentic workflows can directly interact with the application. They're not interacting with it through a GUI or not interacting with it through a manner in which a human was designed to interact with it. They're interacting with it in a way that they want to interact with it, which is like through what looks like a Linux file system. And so we built our product mostly focused on the agent user experience, not the human user experience. that is going to completely transform the way that work gets done because we can basically have, I would call it digital labor running 20 in the background of our product. And it's, you know, orders of magnitude cheaper and honestly a lot better at it than humans are. And that's amazing for all of us. But that's only possible if you rewrite your application into this Gen 4 architecture to take advantage of the fact that you now have the ability to have human level intelligence. running 24-7 inside of a computer.

Speaker: 00:17:33

Yeah, no, it's really interesting. It's that kind of, people talk about kind of human in the loop, don't they? But I think it's all around that, you know, embedding that into the core functionality of the system as opposed to saying we're going full AI native. It's like, that's just not the message here is basically what you're saying.

Speaker: 00:17:50

Yeah, it's architectural. It's the same way that someone would say, you know, like we're cloud native and then they go in and they say, oh, but... In order to update our software, you still have to go in and install the floppy disk or the CD-ROM. Cloud-native means that you have a completely unique architecture where the system is running in a multi-tenant server that's in someone else's data center. It's the exact same thing with AI-native. There is a fundamental architectural rewrite of these products that is going to have to have, absolutely will have to occur.

Speaker: 00:18:20

Yeah, I can agree more. And I'm keen to focus on that. I think you touched on it earlier about... go to market, you know, you've been around for a few years. You've obviously, you've spoken to obviously a lot of prospects, customers, you've had different reiterations of the product. Is there a, I suppose in that journey, is there a bit of customer feedback or product feedback that you've had, which really helped shape what the product looks like now? Is there one, you don't need to necessarily name names, but anything that stands out and makes you think, oh yeah, that really helped us.

Speaker: 00:18:51

I think someone said that a product market fit is always one, you know, degree away from the place where you thought it would be.

Speaker: 00:18:59

Interesting.

Speaker: 00:19:01

It's not where you think it's going to be. It's just right next to it.

Speaker: 00:19:04

Yeah.

Speaker: 00:19:05

And I think that being really curious about what your customers believe, the first step, and that was really where we spent a lot of time when we first brought the product to market. We were thinking of it in terms of these modules and these composable building blocks. And then we realized that our customers wanted us to do a lot of the thinking for them. And they wanted us to do a lot of the leading practices and the best practices. And that made sense. The thing that was surprising, like super, super surprising, was that the capabilities they were asking for were actually already possible in a lot of the tools they were using. It was, it was that the ergonomics of those capabilities was very poor. And The way that we were able to make that simple and legible to them, and a lot of it being powered by an AI-native Gen 4 architecture where they can ask the system not, how does my order is doing? Or tell me about, you know, go find me my latest orders that are doing X, Y, and Z. Instead, they can say. How do you work, system? Explain yourself to me, which is a lot of times what people want. They want to be able to understand the provenance of why something came to be. Our ability to show that and deliver that to them, I didn't realize that was the killer feature.

Speaker: 00:20:25

And then,

Speaker: 00:20:27

but it took time for us to realize this when we'd be doing these demos. We're like, look at how great the performance of X, Y, and Z is. And then they're like, yeah, yeah. But the fact that the system understands itself is... I've never seen that before. And I have, you know, a public company CIO, I show them this and they go, wait, wait, that would have taken a team of consultants multiple days of work to answer the question of how the system is wired up, because they're aware of the data inside of them, but they're not aware of their own implementation. That was like a really big moment for us when we realized we had crossed that, you know, that.

Speaker: 00:21:01

interesting it goes back to your point earlier about kind of the architecture and how it's embedded into that as opposed to trying to make it all around ar natives actually it's understanding its environment it's understanding it's got that human input it's yeah it's very smart it's funny isn't it because you you know sometimes those eureka moments were never supposed to be that eureka moment right you think oh okay hang on a minute we were actually on to something so yeah i've never heard that that's really interesting really interesting yeah

Speaker: 00:21:25

it's um was it David Foster Wallace, right? Like, this is water, the fish going around and asking, how's the water today? And the other fish is like, what are you talking about? Right? And it's a lot of that. It's that I think it's very easy, especially for early stage founders. And we fell very much into this of like, you're so consumed in the thing that you're doing, and you're staring at it that you begin dissolves into the background. And you just accept that that's the way that things are. And other people when they look at it, you know, we're like, oh, that's just the product. And they're like, no, no, no, that's the special thing. And we're like, well, that's just a bit. But really, I think having the deep curiosity to just put it in front of people and be like, hey, react to this. What do you think? What's weird? What's cool? And being super open about what they're describing, not to say, go follow every single thing they're saying, but integrating it into your worldview. That was transformative for us.

Speaker: 00:22:15

Yeah, you're completely right. When you look at the same thing over and over again, it's... it just all blends into one but that's no i've never had that answer and it's um you know super impressive that you built something almost so advanced that you were like well actually hang on a minute that's actually what we should be doubling down on did that impact your kind of wider go-to-market motion as well then after that did it make you think hang on a minute we need to really ensure that messaging comes through yeah

Speaker: 00:22:41

it it changed it completely it changed it mostly because we spent so much time originally focusing on how to communicate capabilities around industry specific go accomplish xyz thing to make my revenue improve to make my profit and you know visibility into my margins improve decrease the risk for my business and that works in a local setting when you're talking to an individual on a specific team or you're talking to managers inside of a procurement function or whatever it is but making the like idea really, really simple to people in a single sentence required that positioning and messaging for our go to market team. It required this clarity, which took basically two and a half years of trying to sell the product and having success in certain form factors. Yeah, it took two and a half years for us to get that clarity, which is you needed applications that adapt to the strategy of your company. Yeah. AI and basically our Gen 4 architecture makes that possible and it's never been possible before. And here, here's how we can show you. And like that, that kernel of truth took so long to get to. And once we got to it, then it was now here's how we price and package this. Here's how we deliver value. Here is how we actually run a sales process to show you and make you believe that fundamental truth. And so, but that took a long time. And starting with the, getting the message and the vision of clarity at that point, perfectly dialed in. If you don't have that, that should be the you're searching for on gta And then once you have it, then you put all of your energy and weight behind driving pipeline against that.

Speaker: 00:24:23

No, fascinating. I suppose the million dollar question now is obviously what's next? You know, series B, you know, obviously everyone, the assumption is growth, but for you and the wider team at DOS, what's kind of on the roadmap here? What can people expect to see from you guys?

Speaker: 00:24:39

Well, growth for sure. That's the big one. I think it's also just focus. like really being focused. How do we win this year? How do we win next year? The main reason I think that growth needs to be tempered with focus right now is that we're in a period of transition where new technologies are being introduced and people need to re-underwrite their view of the world pretty fast, pretty often. I don't know about you, but I probably... In the last 12 months, I've changed my view on a few fundamental things like twice. And I don't know if I had done that in the last five years leading up to the last couple, the last year, right?

Speaker: 00:25:18

You're completely right. For me, it's been, I got so attracted by all the shiny new, as a recruiter, all the shiny new AI tools. And actually what ended up happening was... create more of a bottleneck and actually what we've done now is strip right back and go hang on a minute what what do we really need and actually there's you're right we're in this inflection point now that everyone's got the latest thing on their phone the latest and actually how often what's the engagement levels like what's the productivity like and actually no one can really talk about roi at the moment so i think it's yeah i do i completely agree and i've just had that epiphany probably a couple of weeks ago i think hang on a minute do we really need this

Speaker: 00:25:55

Right. Yeah. And like. People are going to have to adapt to a new way of working in our own company, outside of our company, you know, in the market, the way our customers are going to expect to buy things, the way that they're going to expect the product to behave. And so part of it is that the organization needs to be laser focused on the right thing. That is where we need to be growing because growth in the wrong direction is very, very dangerous. growth along this you know streamline of the where where the world is heading is objectively what you want to do. But my job is to keep my head up and on a swivel and pay paying attention to where the things are going and steer the company towards that. It's your team's focus towards that.

Speaker: 00:26:39

Yeah.

Speaker: 00:26:40

If I would just said, hey, high into the right, put numbers on the board, make, you know, make number go up. It's not clear that we are actually selling the right things to the right people building the right products the right way. Someone said, you know, uh, you know, in this case, growth a lot of times is growing revenue or things like that, run customers. That's a measurement of success. That's not success. And so I think for us, it's going to be tempering growth with focus and that focus being steered with clarity around what's happening in the world.

Speaker: 00:27:17

Such a refreshing view because I think, you know, fresh off the back of investment, organizations can get quite carried away, you know, going to hire the best folk, going, we're going to add all these new features to our tool. We're going to, you know, we're going to target other sectors, which we never really initially thought of. And actually it says a lot about your vision and the success you've had, just doubling down and being mindful of the landscape you're selling into. That has changed forever and actually how people receive information, and how people want to. engage with ai that is going to continually change so i'm guessing that ties into what you said at the beginning is it all has to tie into the embedded kind of human still as a central point but obviously complementing with these ai features yep yeah

Speaker: 00:28:01

people people's role in their own jobs are getting i think pushed up and everyone is going to have to operate in a much more managerial way people will be evaluated a lot more on their judgment and not their throughput you know, the like, do you do this task? It's actually, can you orchestrate the things around you to do tasks and building products that allow people to get better at that for them to, you know, I think also we have the benefit of that these large AI companies, the open AIs and the anthropics of the world, they're really leading the charge on educating the market on how to work in this new world. And so that's been also refreshing that there are some great companies that are basically doing a lot of the consumer education for us.

Speaker: 00:28:45

Yeah, what time to be in this space, right? I think it's the once-in-a-lifetime opportunity in terms of that innovation, the wholesale shift of what we're seeing now. I can't liken it. You talked about .com earlier. It's like, you know, it's bigger than that, isn't it? It's just everything kind of is happening at such an exponential rate, which is exciting. But look, for you and DOS, I wish you all the best of luck, Wiley. It's obviously a fantastic product. It's obviously the adoptions growing exponentially as well. But yeah, you were so... clear and articulate with the vision and the why, which is, uh, which sometimes lost on, um, some people, but no, thanks so much for coming on. It's been a really interesting.

Speaker: 00:29:24

Of course, Alex,

Speaker: 00:29:24

thanks for having me on. I really appreciate it. Thanks again. Appreciate it.