This AI Platform Lets Anyone in Your Company Manage Operations Without a Single Engineer

Wiley Jones: DOS helps our customers and, you know, the industry adapt their applications to the strategy of their business as opposed to adapting their strategy into the shape of the applications.Sam Nadler: Built This Week, breaking it down. Built This Week, we show you how. A fresh idea, a clever tweak you locked in. Built this week.Sam Nadler: Hey, everyone, and welcome to Built This Week, the podcast where we share what we're building, how we're building it, and what it means for the world of AI and startups. I'm Sam Nadler, cofounder here at Rise Labs. And each and every week, I'm joined by my friend, cohost, and business partner, Jordan Metzner. How are doing today, Jordan?Jordan Metzner: Hey, Sam. How's it going? Another fine week in the land of AI. Lots going on. Some IPO news across the board from SpaceX, Anthropic, and AI.Lots of fun stuff going on and obviously really excited about today's guest as well.Sam Nadler: Yeah. Super excited about today's guest building a huge huge product. We love to dive into it. Before I introduce our guest, please like and subscribe. We have new episodes of Built This Week out every Friday.I think we just crossed the 27,000 subscriber mark on YouTube, and so your support is appreciated. And with that, I'd love to introduce our guest CEO, Wiley Jones of DOS. Wiley, if you don't mind, give me a quick intro of who you are and what you're building, and then we'll jump into DOS. Awesome.Wiley Jones: Thanks for having me on, guys. So my name is Wiley. I'm one of the founders of the company. We started DOS about four years ago. And I think the funny claim to, you know, AI nativeness on this is that we actually started the company, incorporated a month for ChatGPT.And I left my job in order to go start working on DOS the moment I saw GPT-two go to GPT-three and I started seeing some of these what would be emergent capabilities. And, of course, like all great, you know, Bay Area entrepreneurs now, I looked at it and said, great. Awesome. Amazing. Super intelligent AI.Let's let's go build more enterprise software. And no. But in in all seriousness, my my background's in hardware engineering. I had a lot of fun making hardware products. I loved making hardware products.Still miss it to this day. And the thing I kept running into over and over again was that enterprise software systems would force our business to operate in a way that I didn't expect I didn't expect we would need to. And ultimately, what the way we kind of described this as a one liner is that DOS helps our customers and you know, the industry adapt their applications to the strategy of their business as opposed to adapting their strategy into the shape of the applications. And this was the thing that I had experienced firsthand many times over and ultimately what we wanted to go solve with DOS.Sam Nadler: Amazing. I'm super excited to see it. We'll jump into overall kinda, how you founded the company, some of the customer base, but give us that quick demo so we you know, I this is to me, it's such a big product to tackle. I I would love to see a little bit of how it works.Wiley Jones: So there's a couple of things in here that I think are really important to note. The first is that it looks and feels like a lot of standard enterprise software. And the thing that we are really clear about is we've tried to make the primitives very simple so that we can build applications that model themselves around what customers need as opposed to forcing people into the shape of the application. So in the world of ERP or what we call ARP, you know, adaptive resource platform, we think a lot about procurement, inventory management, order management. We have these core applications, and what we'll a lot of times do is we'll be able to say, I have these entities and structures in here.I wanna be able to actually show our on hand inventory. I wanna see all of our locations and all the different inventory values we have across the business. Now, the reason this is so hard for so many other companies is that they actually have some set of unique behavior that's unique to their company versus what you would see inside of an existing, you know, off the shelf system. So what we're able to do is especially if I go look at, for example, master data, stock keeping units, very common example here. There will be a lot of components and capabilities in their product catalog that don't fit nicely into the data model that a vendor has provided to them.So if I go look at this, this is our demo instance of like a fake coffee company. Now they actually want to put in their stock keeping units in in the data model for it. They wanna actually say, hey. We actually have Grinds that we wanna track. We actually wanna go into here and say, here are all of my medium grind copies and here are the 25 variants of medium grind copies that we actually track.And so we actually have fundamentally created a dimensional data model. So the ability to take a dimensional data model that's highly composable, you matrix that over with, you know, very ergonomic user experiences around, let's say, creating a new SKU. It's just a form that you can go in and build similar to what you would do inside of Google Forms, just connect it into, you know, the database of the system. You combine all of this together with reporting capabilities and actually what we'll get into with JawSpot, and you have a really powerful adaptive suite of applications with best practices for procurement inventory manager order management all baked in. To put it really succinctly, we help our customers manage the flow of goods, dollars, and data.And the mechanism by which we do that is by taking applications and packaging them for them, but still allowing them the ergonomics and composability that they need if they wanna make it their own. A lot of times people say, you know, you you kind of took the best concepts of Airtable, Ritual, and Zapier. You cross sectioned them, made it great, halted at ERP. A couple of fun things to add just in terms of like, you know, the demo ability of this. So here's a table of sales orders.Here's a million records. I can actually go in and type my name. And over the course of, I don't know, however long it was, a couple hundred milliseconds, it actually went through and searched and found all of the items that referred to me. And it's direct full search on a million order, you know, system. And it was almost instant.The ability to do that kind of really powerful search and, query capability on top of a composable data model, very difficult to do technically. And so a lot of times you'll see, you know, incident spinners and, you know, half wheel of death while you're running queries. Now, the thing that really emerges as like the magic for the system is that not only do you have the performance and capabilities of this at your fingertips from like a real raw database performance level, but you actually can give that to an agent that can drive these experiences. So a lot of times people will ask simple questions like, tell me about my last 10 wholesale, you know, orders. And you can see what it's doing here, you know.It's like it's Dotspot. It's it's going and actually doing what a person would do. I'm gonna go look at all your sales orders. I'm gonna go full pull all the detail information out, figure out how I can go, you know, construct this. And it's going and doing its job.The thing that's much more interesting is that because our data model all lives at the application layer, we can ask it really complicated questions. Tell me how our SKUs work. How do they connect to our product catalog? How do we connect these to different sales channels? And when we process sales orders across those sales channels, what are the ways we do it?How are the integrations functioning? What workflows are actually doing, you know, nondeterministic AI based parsing. And then how does this all get fulfilled and and where you get depleted as a result of this? Create me a very clear description diagram. Walk me through the actual business process and give me a concise summary of all this.It's a much harder question. Yeah. A little bit.Jordan Metzner: A little bit. Now The thing that's The answer is like Strawberry. Yeah. It'sWiley Jones: two. Right? Meaning Yeah.Jordan Metzner: Right? I will not.Wiley Jones: I'll go back to our last wholesale one. It it, you know, it answered it. It does what you'd expect. Here's all the stuff that's in here. Pretty normal.You know? It it went in and pulled all the information. Now if we look at how it's doing this, we actually have a full agent experience that we built under the hood here, very similar to what you would see inside of a cloud code or inside of your, you know, inside of cursor or any really any agent product. It's actually just fanning out and doing the work a person would do. I'm gonna go grab a bunch of stuff across the system.I'm gonna figure out how I work. I'm gonna go pull together all of my workflows. I'm gonna go pull together my data model. I'm gonna go sample data out of the tables and I'm gonna go try to describe the system just like a person would describe a code base or just like an agent would describe a code base. And the the beauty of this is that every single time the models get better, DOS bot gets better.And we don't really have to continue investing in the system, you know, the system infrastructure and and the the prompt engineering here because we built the agent experience to interact directly with the data model because we pushed the data model out of the code base and into the application layer. So I'll let this run for a minute while we, you know, keep looking around at a lot of things. But, hopefully, that was a fun demo of, you know, kind of the the broad capabilities of the system. Any questions while we let this guy continue to run?Sam Nadler: You can't interview thousands of candidates. Interviews to Can. Run interviews twenty four seven across roles, across time zones, across languages. No scheduling, no bottlenecks.Jordan Metzner: Yeah. That's great. Awesome awesome demo. I I love the commentary on, like, the great UI. But and I saw some of the data come coming in, looks like, from Shopify.But maybe you can give us an idea of kind of, you know, maybe, you know, what type of customers have adapted DOS and like where if they've really found kind of the best adaptability. Obviously, like, you know, every company has an ERP of some sort or another maybe. So maybe the TAM is, you know every company ever but I'm sure there's some that have really found some some great success here.Wiley Jones: Yeah. I mean investors love hearing that that's the TAM but the reality of it is that you wanna get really really really specific. So the thing that I would describe here is that it's really important to actually highly concentrate your ICP in order to deeply understand the customers in that smaller kind of, you know, micro segment so that you can deliver the right amount of value to them that will change their that will actually change the way that they work. So in the example of these businesses, we have a very similar story that we'll tell to a very large company that's making hundreds of millions of dollars of revenue versus companies that are making tens of millions of dollars of revenue. And the thing that's I would say most distinct though is that you actually have to understand where they're at in their maturity journey.So very small organizations, the way that they think about value is like they're like very binary. I wanna win. I wanna go grow revenue. I need to actually hit these goals and targets so that the company can continue on the trajectory it's on. When bigger organizations talk to us, that customer profile is a lot more conflated with the maturity of the businesses enterprise systems.So like a business that is exactly like the coffee company that, you know, that I'm demoing but they're doing a couple 100 millions of dollars in revenue or a billion dollars in revenue. Because they, as an enterprise, are enterprise shaped, they behave very similar to a company in a completely different industry, even more so than a company in their own industry that's smaller. And this is I think that good is I think especially important in business systems like ERP is that they become indistinguishable from operating the company, you know, the the the investment in these systems. And so we have to really clearly determine early in our sales cycle is what what side of that line does our customer or does this prospect fall on? Do they have a capability roadmap for their enterprise systems?If they do, where do we fit into that? They already probably have a point of view about data warehousing or they have a point of view about how they'll be using a CRM or what their finance and accounting solution is. And we need to understand where we fit in the potential journey of or in in from a potential solution standpoint in their capability roadmap more than what we would in the smaller segment if we're looking at that 20 mil to let's say a 100 mil in revenue. Those companies are like very much binary. I want to win.I need to go into sales channels and grow revenue. I need to introduce new products. The question from a 100 mil to 500 mil ends up looking a lot more like how do I build systems that I can in can produce dividends and compound for me from a business standpoint over a number of years? So yes, the TAM is every single company but the whole thing the way that we segment it down today is really trying to identify that company that is thinking much more about how do I win in my market for physical operation companies 20 to a couple 100,000,000 revenue.Jordan Metzner: Awesome. Go ahead, Sam.Sam Nadler: What about yeah. What about switching costs? Like, I know, you know, it's the the sales cycles are really long. You know, some of the some of the players in the space are institutional. Have you found it really challenging to get, you know, your ICPs to switch over?Wiley Jones: No. Switching costs is usually not well, like, I think there's two ways to answer this. Number one is that if you're fighting switching costs, then you are not qualifying customers properly. Like, if you're fighting the switching cost, this is not a qualified customer or this is not a qualified it's not an because if they're evaluating solutions but then saying, oh, I won't switch, they're not actually evaluating. Like that, you know, it's like trying to a real estate agent being like, yeah, switching cost out of home is so high.It's like, well, not if you're talking to someone who wants to leave their home and buy a new home. And I I think that's one of the big problems here is it's like the switching cost is very high but people do want to leave. They want to move and they want to migrate because they're looking for capabilities. They're looking for new systems. And so the better question though I find a lot of times is how do you find the people that are in market?And the people there's, you know, let's say five to 10% of the market wants to make a move every year, ballpark. How do you actually make sure that you're in that conversation and in consideration of the people who are making those switches. Or the people who are new entrants in the market, they have a business that's grown from 5 mil to 50 mil in the last eighteen months. Rapid acceleration and growth. Okay.They're thinking about the systems that are gonna take them from 50 to 200 over next five years. How do we make sure that we are one of the vendors that they consider? And so there's the binary, you know, zero to one, this is our first time buying a solution. And then there are people who are saying the business is growing, part of my current systems no longer serve serve my needs. Or there are people who are saying the business has had material changes in some sort of structure, sold the company.We introduced a new product line that's changed the way that we go and actually go to market and the shape of our IT infrastructure has to change as a result. So like, we wanna be in consideration when the business changes so that they can adopt new systems. And so switching cost I think is only relevant in a scenario where someone has no material change. But I don't think you should be pursuing people anyways.Sam Nadler: Yeah. Fair point. Great answer. Jordan, do you wanna jump to new news or any more questions?Jordan Metzner: No. I mean, I think we can jump into the news. I I obviously, I mean, we could talk about enterprise software all day long. But no, I thought it was an awesome demo and just showing kind of the power of how AI into the ERP obviously kinda changes the way people interact with their business. And I think that's an exciting kinda new opportunity and new greenfield space especially as kind of the older classic companies probably will will take a while to to adapt.But yeah, let's let's wrap up your demo here, Wiley.Wiley Jones: Yeah. So just super quick because I took a minute to run, you know, asking you to make a shorter description of this. But it it actually is able to describe here not just like a, you know, tell me how my orders are doing, but it actually is able to self introspect and say, here's how your product catalog connects to your SKUs and how your SKUs connect to your sales channels and how we process different orders across different channels and actually how our order processing workflow works all the way from sales order to scanning and leaving the warehouse. And it's like, you know, line by line in here and we can actually go in and look very specifically at the workflows that drive this. We can see in here for third party shipping, they're using ShipStation to go and actually purchase the shipping labels.We're actually generating dummy labels to then for pickup and local delivery because they don't actually have to pay for the the shipping label but they wanna put something on it so people know who it's going to. Transfer orders, things like that, same such situation. So very, very detailed precise understanding of the system behavior and it's not reading documentation. It's actually reading its own implementation. So that's one thing that's really unique and possible as a result of the architecture of the product.Jordan Metzner: That's great. So this is, you know, essentially tribal knowledge for lack of a better term. Right? Like, maybe somebody knows all of this, but probably a lot of people know pieces of this, but maybe nobody knows all of this altogether or maybe one or two do. And this is kind of exposing that tribal knowledge in a transparent format.Wiley Jones: It's tribal knowledge but it's also implicit. I would say system behaviors that are made to be explicit. So a common way of describing this is if you were to ask someone who has used I don't know. Like if you've ever used any social media product and you ask people, how do you go post on Twitter or how do you go post a photo on Instagram? Well, they would say, you know, I go hit the button, I hit post and then I can actually go and hit like and I can I can take all these actions and if I hit delete then it'll delete it?You know how the system works and how the system functions. You have a kind of a mental model of the life cycle of how a post works or how, you know, liking and commenting works. The complexity of what I just showed on here is much more complex than like the simple understanding of I post a picture, it shows up in some speed. And so it allows us to take complex rich workflows and data models and represent them in a manner that's legible so that people don't have to just remember how the thing works. It's it's less tribal knowledge in how does our company handle x y z process and it's more taking implicit system behavior and making it explicit.Jordan Metzner: That's great. That's great. Awesome demo too. Awesome. Great job.Thanks, Wiley, so much. Alright, Sam. Let's jump into the news.Sam Nadler: Let's do the news. Alright, guys. I thought this was a fun article. Robinhood is allowing users to use AI agents to trade stocks. I thought it'd be fun to riff on for a couple minutes.I personally am not a huge user of Robinhood. Jordan, I know you you use it. I have used AI to analyze my portfolio just like one off snapshots just to get its opinion. Would love to hear, would you use an agent to to trade on your behalf? What parameters would you put around it?Something that excites you. Jordan, if you don't mind, I'll tee you up and let you go first.Jordan Metzner: Yeah. Well, I mean, if you remember, Sam, like, few months back, we did the episode where we used AI to trade on on Calci. Right? I'm sure a lot of people are using AI to evaluate their portfolio like you said. And I you know, if you also recall, we had kinda we had the CPL of Public on and they've announced some new like AI kind of trading features as well that are kinda cool, kind of like make your portfolio.But you know, everybody's looking for alpha. Right? And so the question is is like, can AI bring you alpha? Well, if everybody has AI, you know, doesn't that just kind of phrase everyone's tied together? So I'm not really sure how trading with AI on Robinhood will will gain you like, you know, additional alpha that wasn't previously available.There's probably some things that can be done to help you manage your portfolio. So like, you know, instead of buying an S and P index fund, you could maybe like outright buy the S and P stocks individually and then, you know, rebalance on a daily basis or something like that. And I'm sure there's some other type of features that you could do that might be like better for like tax loss harvesting and stuff. But I don't know. It doesn't it doesn't seem to me that like launching this is going to give you a bunch of better performance.And if it did, then obviously then the entire financial markets would change. Yeah. So I don't know. What do you think, Wiley?Wiley Jones: I think there's a there's the like the, you know, read the marketing, seems okay, seems reasonable. Because I think there's, you know, that take and then there's also the like everything is gambling now, is this just gambling?Sam Nadler: Everything is gambling now.Wiley Jones: So yeah. I I think that's I think that's kinda like the main thing with this is that it's it seems like a little more gambling oriented and making increasing the friction the increasing the ease and reducing the friction for people to have their desire to participate in financial markets made more ergonomic through an agent. Like, if I were to go in and say, hey, I don't like this company. Can you figure out how to short them? I don't know how to short them.Like, the agent would be able to figure it out for me. That seems like maybe beneficial, but also could be destructive.Jordan Metzner: I mean, yeah. If you don't know how to create a short, then maybe you shouldn't be You be shorting.Wiley Jones: Yeah. Yeah. It's like, I think it's kind of a point with this is like they're, you know, hopefully, Robin Hoods, as they're doing this, they're making skills, you know, to back all this. Right? Like, go go look at the Renaissance, look at the Citadel Strats and and go apply those for me.Okay. Great. Done. I don't know. I'm I'm generally, like, I think stuff like this is stupid but that's just my take.Jordan Metzner: No trading bot come into the DOS platform many times in. Yeah.Sam Nadler: Treasury treasury functionality. Yeah.Jordan Metzner: Yeah. Look, at least I know they're trying to be innovative in a space that, you know, obviously, historically was not innovative. You know, Robinhood's probably largest innovation was bringing down the cost of portrayed that emanated throughout the industry. So, know, you maybe this will be a dud as a feature, but, you know, maybe it'll carry on and we'll see kind of application of it across the board. So, you know, they're trying at least and so Yep.That's fun.Wiley Jones: I think it's think it's great because I think it gives peopleJordan Metzner: like, I think theWiley Jones: idea of having natural language interface and having agents take action on your behalf with respect to your finances and your ability to construct a portfolio, perfectly useful. I think that also highly destructive capabilities and we've already seen what that looks like in the world of Robinhood and I don't I don't love the idea of making it so that you have an even more frictionless path to destruction. We're gonnaJordan Metzner: see GameStop around too with this. Yeah.Wiley Jones: You know that. I'm just like mashing and saying eight eight more eight more. Yeah.Jordan Metzner: Yeah. Yeah. You know the model's gonna be biased towards something and then boom, you know, some Yeah. Yeah. Some random cheeseburger shop in Ohio is gonna be worth a few trillion or something.Yeah. Yeah.Wiley Jones: It's like when you ask Kodak's or Claude to review one another's work and you tell it that it's Claude, then it starts to be like, well, don't like Claude. And the same thing. Opening IPO is, you know, it's like, what what model are they using? And it'll tell them which stocks to pick as a result.Sam Nadler: Awesome, Wiley. Great episode. Right before we wrap up, where can people find you? Are you on X, LinkedIn?Wiley Jones: Yeah. I'm I'm on X. Just I think it's at wiley c w j. And, yeah, you can just find me also on LinkedIn. Very active in both places.Feel free to feel free to DM me. Always love talking about stuff like this.Jordan Metzner: And how about to sign up for DOS?Wiley Jones: For DOS signing up for DOS, just go to our website, book a demo, reach out to our sales team, reach out to me and I'll forward you along to someone. Literally reach out to anyone in our team and we will route you to a person who can tell you more about what Operations Cloud is, how the idea of, you know, taking procurement inventory management, order management package all up can help service for mid market company mid market companies that are moving physical products around the real world.Jordan Metzner: Awesome. Thanks, Wiley. That was an awesome episode. Thanks, Sam. See you guys.Sam Nadler: Thanks, everyone.