Omer Khan [00:00:03]:
Welcome to The SaaS Podcast. I'm your host, Omer Khan. AI has changed the playbook for building and growing SaaS. Every week I talk to founders who are writing the new one. My guest today is Zhong Xu, the co-founder of Deliverect. They connect delivery apps like Uber Eats and DoorDash to a single system for restaurants. Zhong's already had a successful exit. His last startup merged with Lightspeed, which IPO'd.
Omer Khan [00:00:28]:
And today Deliverect serves over 80,000 restaurants across 50 countries and is approaching $100 million in ARR. But when they started, they signed up 100 restaurants and manually processed every single order before writing a single line of code. Then COVID hit and growth exploded overnight. They opened up 10 offices in just a single quarter. And now Zhong is racing to build an AI layer for his business before the whole industry gets commoditized. Zhong, welcome to the show.
Zhong Xu [00:01:03]:
Thank you, Omer. Thank you for having me.
Omer Khan [00:01:06]:
So for people who aren't familiar with Deliverect, can you tell us what does the product do? Who's it for?
Zhong Xu [00:01:13]:
We started Deliverect around 7 years ago. Deliverect is really the operating center or nerve center for restaurants as well retailers to manage the digital sales channels,, right? What do I mean is if you think about a restaurant today, you know, we, you have customers from a Burger King to Popeyes to Little Caesars. They, you know, run all their sales almost 80, 90% digital through one side digital marketplaces as Uber Eats, DoorDash, you know, Just Eat, Grubhub, as well even through their Instagram, TikTok, website, catering, their kiosks, their app, and so on and so on, right? So every restaurant today, we asked before, has so many digital channels. And so you need an operating system to manage the menus, the promotions, the customers in a centralized way. And so today, you know, we do that for more than 80,000 restaurants around the world. I know, and we are growing really, really fast because just in Q4 2025, last quarter, we added over 8,500 restaurants in just a quarter. That's quite insane. I think that's more than most restaurant platforms public companies do in a quarter.
Zhong Xu [00:02:26]:
And so for us, it's really helping them to navigate the ever-changing digital world and make sure they are future-proof themselves.
Omer Khan [00:02:34]:
Great. And this is probably an oversimplification, but would it be fair to say that you're almost the interface between the restaurant and the kitchen and the way that customers order and trying to reduce all that complexity in between while still giving them plenty of choice in terms of how they order or interact with the restaurant.
Zhong Xu [00:02:57]:
Yeah, exactly. But if you put it simply, think about, you know, everyone has used Uber Eats or DoorDash. Once you see a restaurant, have you ever wondered where does the menu and the pictures, the pricing come from, right? Yeah. So that's what we do. We push it through DoorDash so that it's alive. You know, one item is really out of stock because it's sold out. We make sure that you can't order it. But even like when you press the button and do the checkout and really said, I want the food, what happens then? It's not always DoorDash doing it, right? So what happens is we get these orders and we inject that in the restaurant systems.
Zhong Xu [00:03:36]:
It can be their point of sale systems, their internal ERP systems, and make sure we track the whole journey of, uh, cooking, being prepped, being packaged in the bag., and then afterwards we even get a rider to deliver it. So, you know, something very simple, you press a button and you get food is actually digitally or technically quite complex.
Omer Khan [00:03:58]:
Yeah. Yeah. Um, and give us a sense of the size of the business. Where are you in terms of revenue, customers, size of team?
Zhong Xu [00:04:04]:
Yeah, I mean, we are a, a global company, uh, over 450 people in 13 offices, uh, around the world. Literally, I, I woke, wake up from Sydney to, you know, we have office in Mexico City, New York, uh, Atlanta. So it's really global. Today we process more than orders for more than 80,000 restaurants around the world. And look, soon we'll cross the 100,000. So that's going to be a milestone. But even more important, I think we, in August, we processed in total more than a billion orders for these businesses. A billion orders.
Zhong Xu [00:04:38]:
It's quite a lot in dollar value that went through our system. That was almost, you know, $20, $25 billion that went through our system. So, and you know, that's true the history of our journey, but this year we're doing more than $500 million. So it's, we're getting to a decent scale where, you know, I think we are really the backbone of digital food. On that, we're the global leader and we're really, really excited what's next.
Omer Khan [00:05:03]:
And where are you in terms of revenue ballpark?
Zhong Xu [00:05:07]:
So this year we'll probably, you know, getting or breaking the $100 million in committed ARR. It's really exciting. I think that's, you know, that's really the marker for any company, right? You want to make sure that you get to size and really be sustainable as we go.
Omer Khan [00:05:25]:
Now you have a really interesting journey in terms of how you got to Deliverect. And the thing that stuck out for me was that you said your first gig was when you were 17 years old, you started building websites for restaurants because your, was it your father was selling point-of-sale systems to restaurants?
Zhong Xu [00:05:48]:
Yeah, exactly. I mean, he was the first generation of immigrants from China to Belgium, where we grew up. You know, back then, he, you know, nobody wanted to give him a job, although he has a PhD or a master's in material science. And so he, to survive, he randomly went to Asian restaurants to wash the dishes, right? And so what happens is one of these owners said, hey, "You're a smart guy. Can't you make a system for me?" So what he did is he bought a book, learned C++ in 30 days, and built a point of sale for Chinese restaurant. Still has it. But you can't imagine having a dad like that. So I think from age 14, he was like, "Man, when are you going to work though? You can't like just, you know, you need to do stuff with your life." I turned 15, he was like, "You really, really need to do something." So when I turned 16, I went to study, you know, Dreamweaver back then for people that I was very, very old to create websites.
Zhong Xu [00:06:45]:
Today you don't need to do that. And between my 16 and 18, I made over 1,000 very, very ugly, I would say, Chinese websites. So, you know, that really taught me that, you know, the entrepreneurship side of it. And that's also it's possible to, you know, at that young age to make money, right? To, you know, do something that has an impact for these restaurants. So that's how I started my journey.
Omer Khan [00:07:08]:
That is awesome. And did you also, after that, build your own point of sale system?
Zhong Xu [00:07:17]:
Yeah, so I got, I would say, the microbe to really do software. So I studied software engineering, created mobile apps before the iPhone existed. Back then was the Nokia N95 was a smartphone of our age. And so I did a lot of that. And One thing that we really saw is when Steve Jobs launched the iPhone, that was such a game-changing moment that I thought, wow, we need to do something about it. You know, and when he came out with the iPad, I connected with, you know, the historical, you know, knowledge I have of the restaurant industry with the fact that you can have a new medium where things are cloud-based and you just have commodity hardware instead of, you know, installing bulky hardware, you know, driving every time, couple hours to install stuff., right? And so that's really inspired us to create one of the first iPad point of sale out there. You know, basically me and my co-founder, we coded, you know, the complete iOS app in 9 months' time while working a full-time job, basically during traffic jam and all the holidays and weekends. And yeah, we released that to the world and that company, you know, is one of the biggest point of sale companies now in the world.
Zhong Xu [00:08:34]:
As we merged that in 2014 with a company called Lightspeed, and we IPO'd that in 2019. So I think Lightspeed today has around 150,000, 160,000, you know, restaurant retailers around the world, and that has been a great journey for us.
Omer Khan [00:08:51]:
That's awesome. So tell me where the idea for Deliverect came from.
Zhong Xu [00:08:54]:
So most of our ideas, I think people always ask me, you know, how do these ideas happen? Most of the time I just listen to customers, right? In the case of, you know, POS iOS or the iPad policy, it was literally because I went to restaurants to do support and, you know, they called me to drive 3 hours and then to fix something. I did something very small and just, you know, wasting so much time. I thought there was something better. In the case of Deliverect, it was really, again, listening to restaurants. So, you know, I visited obviously a lot of restaurants, and when Lightspeed IPO'd, I was going to take a long sabbatical.. But all these restaurants said, hey, you can't do that. There's a bigger issue. And I really said, what's the issue do you guys have? I already solved all the technical stuff for you guys.
Zhong Xu [00:09:37]:
And for them was really the move from physical to digital. They said, what's going to happen if what's happened in retail is going to happen in restaurants? What happens if, you know, there's these platforms, Uber just starting, that's going to take from now a couple percent volume to 50, 20, 30, 40, 50% of my volume. What happens in the future where, you know, you have home automation devices that listen to you and just orders? What happens if you have a smart fridge that sees there's no beer anymore, that auto-replenish that? What happens if there's drones and robots? I mean, they went actually quite far. But in hindsight, you know, I thought, hey, look, it sounds so futuristic, but in a decade, that's probably the case, right? Fast forward 7 years, I think we, we are very much in, in that world. And so that really pushed us to create a platform to help these restaurants to make that transition because at the end of the day, we wanna make sure the restaurant can, you know, do what they like and that's cooking and, you know, serving their customers.
Omer Khan [00:10:39]:
So there, there are 4 co-founders of Deliverect, is that right?
Zhong Xu [00:10:45]:
Yeah. So in, in case of POSISS, we were just with 2. The reason we got more is we learned that Although, you know, we can make the products and certain bits, you need other elements, right? So we have one co-founder that was great in integrations. You know, Yan and I are great in strategy and making the product. One's great in integrations. And then we have our other co-founder, Jerome, that's our president now. That was a key element that could scale a go-to-market side of things, right? Because one of the things I learned through the years is the idea doesn't really matter that much. And, you know, I know for a lot of people making product is hard.
Zhong Xu [00:11:24]:
Making product is not that hard, right? What's really, really hard is distribution. How do you get whatever you're making that is good or at least at par in as many hands as you can, right? And so to nail that, that's where you sometimes need people that can build a go-to-market engine, right? For me before when I, you know, were young, I always thought you build it, they will come and sales is just, you know, they just talk. But actually, Sales is really an engineering process. So you need, you know, people that can engineer that sales side of things to make sure you can scale.
Omer Khan [00:11:58]:
So tell me about how you got started. So you've seen this opportunity, this idea. What did that first version of the product look like? What problem did you initially pick and say, I'm going to go and solve?
Zhong Xu [00:12:12]:
So this is a really big one where we learned that MVP is really MVP. I'll tell you why. We learned from our first company with POS iOS, We coded 9 months, doesn't sound a lot, but actually we went so overboard in features. We had drag and drop tables, we could like put pictures on the product and, you know, we could probably have launched it in 3 months or 2 months, but then we wanted to do it perfect. And so when we started Livrec, we really said, look, let's not make this mistake. The MVP is, you know, what if I'm gonna do it? So literally I sold to a couple restaurants and I was behind the scenes each time when the order came in. I was manually inputting in their system, right? Without them knowing it. And that was the automated way.
Zhong Xu [00:12:57]:
So we could launch from day one. Second time when we saw, hey, it was more than we got to, you know, 50 to 100 customers, we're like, okay, let's make sure we can map it. Then we start coding, but it was very rudimentary Excel sheets. You know, you need to think burger A and then we map that to, you know, what's in their point of sale so that it, that could get injected without all the whistles and, you know, with all the features we have now with making menus and automatically getting injections. So the MVP is almost no product. Like, and that's what I want to emphasize, you know, to listeners, like really do not overengineer it because you actually have no idea what you're building anyways. Like until you get scale, what are you building? Like, so get to market as quick as you can. Maybe an analogy, I was talking to one of the largest investors in the world that invests a lot of AI companies.
Zhong Xu [00:13:51]:
And I asked them, hey, what's the difference between classic software companies and AI? And his answer was actually enlightening. It's like, it's still iterations. The difference is before, even if you guys are working agile, you release every month 5, 6 times, right? That's a lot., but these guys can release 50 times, 60 times. So they're beating you by 10x because they can so quickly code these things, right? So that's really a lesson learned, you know, iteration and MVP is key.
Omer Khan [00:14:26]:
Yeah, yeah. So tell me, so basically I think it's like the Wizard of Oz MVP, right? It was what we'd call that where you are sort of selling a product on the front end to customers, but Everything behind is not code, it's human delivering on that. And was it just you doing that for like the first 50 customers with all these orders coming in?
Zhong Xu [00:14:49]:
We were just with 2, we were just literally doing that, you know, quickly we wrote a script because we're software engineers. But the first couple of days, you know, I think the first week we did it manually. And look, it's really a big lesson is don't always try to do things that scale, do things that don't scale. Because, you know, let's automate when we have volume, you know, let's make the product when we have validated that more than one customer wants it, you know, let's do this, you know, once you have a certain scale. And so you still can convince the customer, right? So because if the pain is there, you know, there is certain amount he's willing to pay, although it's not great, he's still wanting to do it. So, you know, in that case, we said, hey, what if you pay us just, you know, $50 a month. Is that, you know, worth it? And because that's worth it, you can easily see the value if you can automate and grow that, right?
Omer Khan [00:15:43]:
So that's how you start. What did you show those first 10 customers? Like what was the sales pitch? Was there a demo? There's kind of not much code right now at that point.
Zhong Xu [00:15:55]:
I don't think, I do think it's easier if it's the second time, there's some trust involved, but even if it's your first time, you can find your maybe mom and pop shop that says, hey, trust me, I'll I'm gonna do it, right? Or, you know, your friend's store that has it to prove your MVP, right? Basically on trust and like, hey, look, we have done this a lot, you know, we can bring these products, let me help you to do that. You know, and we will start with one store, we'll then go, you know, they have 100 stores, maybe we scale as we go and you measure it like that. And the only requirement I ask for them is, okay, you're gonna get a large discount, I just want you to be my reference case. Right? I want to make videos and you're going to tell me how great automation is. It was all manual. Say it's all, you know, quality is better, no more mistakes, etc., etc., right? So, you know, that's how you do it in the beginning.
Omer Khan [00:16:49]:
I love that. And so at what point did you actually build the product? How far did you take this? And what was the What was the signal or, you know, that told you this is enough, we need to have a product now?
Zhong Xu [00:17:08]:
I think that's quite quick. First of all, I start talking to a lot of restaurants, right? As we serve them, I think I talked to in the early days in the first 3 months almost to 200-300 restaurants. Like literally you just go out there, you just, you know, it's not that hard to talk to them, but you need to, you know, every day you can talk to at least 10, right? So, and especially if they have changed. So you talk to them and I validated that the issue is quite everywhere. And I even validated that across countries because like, is it a really thing for one country or is it everywhere? So we realized this problem is really, really important. And because of our background in point of sale, we understand, we understood the complexity of doing it, that it's not just building an add-on, it's, you know, you need a different platform because a point of sale is very hard to update the software because you can't just do an update on Friday evening because, you know, it's payment and there's hundreds of customers. But then in the reality, in the software world, like Uber Eats will release their software on Friday evening because they just don't, they move way quicker. So we need to build a platform above it.
Zhong Xu [00:18:11]:
And so we started coding. And every time I think I start a company, I code the product, we code the product ourselves. I think in this case it took us, you know, 2, 3 months. We had something very quick script in a week so we could get customer online, but then we thought, hey, we need menu management, we need to make sure they can manage their locations, we need to make sure there's quality, et cetera, et cetera. And that took us 3 to 6 months, right?
Omer Khan [00:18:36]:
So that's really the start of things. What did you learn from your previous experience about, you know, feeling like you spent too long building the MVP, you took a very different approach with the MVP with Deliverect. But when it came to building the product, what did you do differently to make sure that there wasn't feature creep or a bunch of things that you really didn't need?
Zhong Xu [00:19:02]:
Actually, it was reverse because once you know there is product-market fit, the first time we put a lot of feature in, but we didn't think about scaling. Because if we're honest, when we started POS, we're like, yeah, well, you know, in best case we get 5,000 restaurants, right? And what's the maximum order? So to give you an idea, early days we didn't even have a cloud database. We had a server at my home that was running SQLite. Like that's not made for anything, but in the early days up to 1,000 was perfect because it's super quick, was in memory, but it was not scalable. And so we did learn like, hey, if you build this, And you find the product market fit, think about architecture and scalability because changing these things afterwards is really painful. So we did do the right way, making sure that we have redundancies, we build in multi-server fallbacks and so on, right? So that's what's really important because when it goes, then it really goes. You actually don't have time to refactor all your code.
Omer Khan [00:20:03]:
So it becomes spaghetti and you don't want that. What would you say to a founder who is in the early stages and is maybe spending too long building that first version of the product? And their rationale is, if I screw up, I'm not going to get a second shot with these customers.
Zhong Xu [00:20:23]:
So I have to make sure my product is perfect. That's, look, there's no such thing as you build and it will come. Everyone's like, even we have product managers that like, but John, when we release this feature, there will be hundreds of customers using it. It'd be perfect. I said, I guarantee you there will be no one, maybe one. Really don't assume that there's that avalanche. It's iteration, right? Be happy if you have one or two that gives you feedback. And worst case, you lose them.
Zhong Xu [00:20:50]:
Especially the early days, it's not bad as in the later years. Look, today we can't make these mistakes. We run largest companies around the world because if you lose trust on one of these changes, you're not going to get back. But who cares in the early days? If you talk about restaurant, there's what, 10 million restaurants? Well, 10 million shots, right? So there's no real risk towards it, right? And so make sure that these customers are co-building, make sure they're also evangelists, right? Tell them like, hey, I'm building this together with you. I want you to be my ambassador. I want you to challenge me. Here's my WhatsApp, here's my phone number. Keep a very tight loop and use them actually as your beta customers, right? And that works really well.
Omer Khan [00:21:34]:
They want to help. So were you selling to single location and multi-location businesses in those early days?
Zhong Xu [00:21:42]:
Yeah, well, in the first company, not because we were doing just SME, we didn't know better. In the case of Deliverect, we thought about, hey, you know, it's hard to scale single location, so we need to make sure it was future-proofing as well for larger locations. But again, you know, because it's the same domain or same field and then it's my second company, we did have unfair advantage where I knew all the point of sale vendors and I had their numbers. I knew, you know, most of the customers that, you know, we needed in the early days. So the trust side, you know, allows you to go a lot quicker, right? I'm not sure if I could have built Deliverect without the previous experience and network we built before, right? So it's sometimes chicken and egg and you need something to start.
Omer Khan [00:22:27]:
Yeah, we were talking about this earlier, the fact that you, when you started building these websites at 17 for restaurants, and by the time you founded Deliverect, you had like what, 12, 13 years of experience in the restaurant industry. And so you have deep domain expertise, you have relationships, you've built trust. And I'm curious, like how much, you've kind of partly answered that question, but how important was that versus a great product and innovation and all of that stuff?
Zhong Xu [00:23:04]:
Look, great products always need it, but like if you build something great, there will be others that will also do it, right? There's no idea that's like only limited to you, especially if the problem is big enough. And so, having that network and distribution and getting as quick as you can to the customers, that defines who wins, nothing else, right? I'll tell you why. Like for example, when we started our iPad point of sales, we were actually before Square launched, right? So think about it, that was the, you know, and you know, when we launched and Square launched, we were like, why the hell did we make it so complex? These guys just have 3 buttons and a payment thing. That's it. We could have done that in a month instead of 9 months trying to do drag and drop tables and coloring and, you know, user interface changes, allowing the customers to customize everything. So, you know, having that and getting distribution and getting it to the customer first, that's what defines winning. Not, I'm sorry, not your great product.
Omer Khan [00:24:05]:
It matters, but, you know, most innovation you know, don't win unless you have distribution. So for a founder who is starting out today in the early stages, trying to get traction, and now they're realizing I should have started 13 years ago like Zhong and built relationships and all of this stuff. What's the next best thing that they can do or you would do if you were going into a new NewSpace?
Zhong Xu [00:24:34]:
Well, if I were to go NewSpace, I would talk to as many people as I can. I would learn, right? So, you know, the reality is you don't know what you don't know. There's a lot of things that, you know, to make a business really successful, you need to have some secrets. What do I mean? It's like you need to know something that whatever you think it's like that, it's a book, you know, Peter Thiel, Zero to One. It's like where you believe it's not true and everyone thinks something else. I'll give you an example in Deliverect. Everyone in the world, all the investors said a point of sale will do what you do. And I knew that it was not true because I built one of the largest point of sale companies out there.
Zhong Xu [00:25:14]:
I was trying to make this in the company. I'm like, this does not work, doesn't fit, it's not gonna scale. So, but because everyone thought, hey, this doesn't make sense as a standalone company. People didn't do it, right? Besides obviously all the point-of-sale companies, they get it and they're like, okay, let's partner with Zone because we're never going to do this in-house. So this is where you need to have deep domain expertise and understanding the pain that sets you apart, right? While everyone else thinks, hey, why would you build this thing?
Omer Khan [00:25:45]:
There's no point. Yeah, you've talked about distribution several times and the importance of this. And, um, you know, these days we hear a lot about AI-first startups that are getting to, you know, the first million, 10 million, like super fast. But for you guys, I mean, back in 2017 when you started this business, um, I think you got to like over 10 million in ARR in like the first 2 and a half years.
Zhong Xu [00:26:15]:
Right.
Omer Khan [00:26:15]:
So can you talk a little bit about what drove that? You mentioned the, You know, kind of thinking about the GTM engine early on, but what helped you get that level of growth so quickly?
Zhong Xu [00:26:26]:
So one of the things we did was, if you think about in the first company was App Store being distribution, you put on App Store, you had. In this case, we were like, okay, you know, we know restaurant, we can call them, but that's not quick enough. So we partnered with a lot of point-of-sale companies, right? Because of my background, I know that they needed us and they have obviously a lot of restaurants. So because we partnered with them, they could distribute us. So instead of signing, you know, customer by customer in the same time frame, I can probably sign 10 point-of-sale companies that in their own signs, you know, every month maybe 100 customers, right? So that partnership motion was very important to that success. You know, to be fair, it was not always easy because some of these point-of-sale vendors were competitors. So when I called the CEO, most of them all let me meet them. But then I think one really let me wait more than 1.5 hours in the launch.
Zhong Xu [00:27:24]:
I'm like, what happened? He's like, zone. That's the only payback you will have for competing so hard with me. But so that helps, right? And so, and as we go, of course, we also had COVID that accelerated everything. So that's helped the awareness or getting what normally would take more years to adopt for these restaurants in a very short time. So timing is crucial. It's the perfect storm. And when there's a perfect storm, you try to ride a tsunami and try to not fall off.
Omer Khan [00:27:57]:
That's basically what you do. Yeah. So I mean, it sounds like it was a sort of a delicate relationship with these POS systems. So how did you kind of pitch these partnerships? I often hear about founders who know that they should go and build partnerships. But often getting those partners signed up, on board, committed, knowing how to sell your product or promote your product, it's in the execution that things fall apart. And so what kind of, I guess, challenges did you face and how did you overcome those?
Zhong Xu [00:28:37]:
I think the benefit is they did trust me, although we competed heavily. They knew for a fact Zhong is not going to build another point of sale. So they was like, okay, he's not going to compete outright with us. So more is like, hey, look, let's partner. And then look, it's more making sure there's no channel conflict, right? I think what's really hard for early stage companies is you sign a partner, they sign, and you want also these end customers, and you start creating these channel conflicts. Oh, this is mine, that was not through you. So most of the time when we assign a partner, even if the customer came direct to us and they have certain interaction with the partner, we always attribute to the partner, right? So you don't want to create conflicts. You don't want to create a situation where they feel like you are not honest.
Zhong Xu [00:29:23]:
So make sure they are successful because when they are successful, you are successful, right? Create a win-win situation that works. And for the rest, like it's this following up, right? So it's almost like, you know, Today, probably less because you don't have so many salespeople, but normally you could even, you know, go to the office of these companies. You know, you work literally beside the sales guys and look, you give them some, you know, gift vouchers if they sign a deal that brings to you as like, thank you and so on, right? So it's all these tactics you can do to get, you know, really the salespeople from these partner companies engaged, right? They need to like you. That's it. You know, it's not because their boss says sell it, they will sell it. It's, you know, take them out, show them fun, show them that you mean it, show them that you care, you know, give them some spiffs, give them some gift card, give them something that makes something worth it.
Omer Khan [00:30:16]:
What about the integration with the, you know, the DoorDashes, the Uber Eats and these kinds of people? Were there any challenges there or was it You know, what was it like building those integrations and getting them up and running?
Zhong Xu [00:30:31]:
I think we have a great relationship because we always honor them, but that was not always easy for them. It was like, hey, we want to maybe, you know, have these own, you know, direct relationship. But I explained that it's really, really fragmented world, right? People think about, oh, restaurant point of sales, but there's 10,000 point of sale out there at least. So managing all these systems, you know, it's not worth it for them. Yes, they can integrate to one or two big ones, but never the long tail. For Uber or DoorDash, they obviously want a long tail, but you know, the maybe famous Chinese restaurant is definitely not using, you know, one of the point of sales, you know, it's gonna be a local guy supporting it. So having that reach was important. And as well, look, it just took a long time where we maintained our relationships.
Zhong Xu [00:31:18]:
We also got investors on board that had great relationship with them so that they knew that we are not, We are not bad intended, that we are really doing things for the right cause. And now, of course, we're quite big with big sizes. They understand that and there's contracting in place, right? But it takes time to win trust. And sometimes it's literally convincing one after other that you're doing the right thing. Most of them, they said, hey, I didn't believe you would succeed, but look, you proved it wrong.
Omer Khan [00:31:51]:
Let's talk a little bit about pricing. Can you just explain how that works? You talked about initially anything was okay, right? I mean, hey, does $50 a month sound good? Okay, great, we'll do that. But how did that pricing evolve in those first few years?
Zhong Xu [00:32:08]:
I do believe you need to sell value, so don't just give it for free. If a customer is not paying anything, it's not a customer. So, you know, also they will not give feedback. They don't feel, you know, people actually, you think, oh, they're gonna give you more feedback. They are actually scared. They're like, I'm not paying anything. I'm not gonna call them 5 times a day because they feel guilty. So they don't do anything.
Zhong Xu [00:32:31]:
So make sure you do charge a fee. And what you can do is, you know, you need a discount, charge a high fee, but then discount it backwards and say, hey, the first 6 months it's flatlined, or What we always did is like, hey, the first, you know, you pay for 1 year, but you're going to get 1.5 year or 1 year and 9 months for free, right? Why? Because, you know, you still can recognize the AR, although because at the end of the year you can have a nice bump, but then you give them enough comfort that you're doing something, but also ask something in return. You know, if we get any discount, we always ask them to be a reference case. You know, and most people like that. It's like, hey, we want to come make a video, promote your business, but like, yeah, tell something about us. Want to make a use case. I want you to, you know, intro me to 4 or 5 of your, you know, friends that are in the industry. And most people do that gladly, right? And that creates your distribution as well, your social proof that whatever you're doing is working.
Omer Khan [00:33:30]:
Yeah, I love that. I mean, you're very intentional about every customer I'm going to give some value to them, but I'm also going to get some value, which is going to ultimately help me improve my own distribution.
Zhong Xu [00:33:50]:
100%. But because the best salespeople are your customers. Like if you think about it in any business, if you have a peer that telling you this thing works, you're going to buy it, right? Especially if it's someone that you look up to, right? So get these early adopters in any markets and try to, you know, make sure they're reference and then you start that. Maybe one story why we, you know, that's why we also scale so quickly with Deliverect around the world. We are now in 50 countries. Literally, I opened, I think during COVID at a certain moment, 10 offices in one quarter. And people always said, why did you do that? And it was all virtual. I said, look, we learned one thing that You know, the idea that we have is good.
Zhong Xu [00:34:34]:
That means in every country in the world or every big region, there was someone doing it. And so I learned one thing that's really, really hard, even if you're a big, big company to enter a new market, if there's an existing incumbent with all the early adopters and actually, you know, made their name. So we said like, do this differently. What if we start everywhere? You know, we're competing at the same level. Worst case, worst case, you know, we need to close the country because we're number 3. If we're number 2, we can always survive. If we're number 1, we won. And so after 7 years in all the countries we are, or we are number 1 or number 2.
Zhong Xu [00:35:09]:
So that really allows you not to, you know, go and buy these companies to win the market, right?
Omer Khan [00:35:16]:
So getting these early adopters is key for everything. Let's talk a little bit about AI.
Zhong Xu [00:35:22]:
What's been the most significant impact that AI has had on your business so far? Well, that's the one thing that keeps me awake at night. I think it's transformational, right? It's, you know, internet, cloud, mobile, you know, and now it's AI. I think the difference is it's compressed in such a short time. The innovation that's happening so rapidly. If just think about it, like, you know, I think what the stat was beginning of the year was 5% of the global code was AI written by end of year 25%. If I look at Deliverect, I see that already 30%, 35% of code are, you know, guided but written by cursor, cloud, and so on. And so the change is profound, right? And so what we are really focusing on is two sides. One, on the operation side, how can we run agents for restaurants to manage the marketing side of things, their menu side of things, their throughput, helping them getting better pricing.
Zhong Xu [00:36:25]:
And then the other side is, you know, getting ready for the age of agent economy. Commerce, right? Where I do think that's important. It's for us another channel, but where you're going to get home, you're hungry, and then they'll say, "Oh, you like this? Should I order it?" Like your Alexa or your Google Home or even your fridge or any tablet or any platform, right? So that's really the future. And I think it's coming really, really quick.
Omer Khan [00:36:53]:
So if you're in the software industry, it's exciting, but also scary, I would say. Yeah, definitely. I mean, huge, huge opportunities where you just described for Deliverect and how the whole restaurant experience or the food experience becomes so much better for consumers as well, right? But what do you think is the biggest threat that AI poses to your business? What's the part of that that keeps you up at night?
Zhong Xu [00:37:26]:
Well, I think we need to make sure we innovate, right? We need to make sure we are connected to these NCPs or these LLMs so that we are ahead. I think it's in any business stagnating, like, you know, what you don't want to be is we are the intelligent layer, right? Because we don't just connect the orders, we help them to do menus and so on. So if we don't disrupt ourselves with providing intelligence so that you have AI that auto creates this menu, auto rebalances orders or to change the order of items and so on, then someone else will. And so once you lose that intelligence, we will still exist because someone needs to do the piping, the connectivity. But the intelligent layer is where the value is. So yes, you need to have the infrastructure, but being infrastructure alone, nobody remembers, right? So you can make money, but that's not where the value is.
Omer Khan [00:38:19]:
And so being in that intelligence layer is crucial. So in a world where, you know, we talk about, hear a lot about vibe coding and it is a lot easier to build products these days. And then if you're exposing all of your data to LLMs or through MCPs or whatever, what do you see your moat being?
Zhong Xu [00:38:46]:
What's the most important kind of, you know, part of this for your business, the value? I think really guard really the data, make sure you can't do something wrong. There need to be some rules and we still verify, right? Imagine we manage the menus. Imagine LLM says, hey, great. Now, you know, because there's something happening in the world or it's really bad weather, let's triple the prices because, you know, the assignment was make me profit., right? Okay, give me a menu that makes me profit. Well, you know, they can do that, but then you're gonna get a public outlash, right? So how do you scope it? So today is a lot of these things we generate and then we ask permission from the human. But as we go, I think, you know, at a certain moment you will say, hey, run it automatically. There's no reason technically why you can't.
Omer Khan [00:39:35]:
You just wanna guardrail it that it doesn't do things that you didn't think about, right? Yeah. That's an interesting way to think about it. Is there another example that you can think of with that? I'm just curious, what else would be a concern with this sort of agentic model and making this data available?
Zhong Xu [00:40:00]:
Well, it's less the data. In our case, because we control restaurants, it could be overwhelming for them. If they actually generate too much orders, again, it's like, you know, they can maybe the LLM can reach out to certain companies to do advertising, right? And then they swamp the restaurants because look, they were not foreseeing so many orders, right? To tricking customers, maybe, you know, if you control the menu and pictures, you know, they can create random hallucinated promotions that, you know, is not the value because you just want to baits your customers to order these items, right? So I think there's part of it where we are not sure what it is because if you give them direction, it will execute, but will it execute in the ethical way that you want to? I think that's where, you know, a lot of testing and research need to be done. And so look, I use LLMs constantly and once in a while that thing hallucinate, right? It's still the case. You know, there's a lot of AI slop. And so do you want that in a real world where your brand is important, where your customer is there? So guardrailing that, you know, it's reputation. Reputation is very hard to win, very easy to lose, right?
Omer Khan [00:41:14]:
So, you know, you don't want that. Yeah. And just a silly little example today I just went through this morning was I wanted to just look at a page on my site and just see how is this optimized for SEO or AEO or any of these things. And so I had Claude look at it and just say, hey, look at this URL and tell me how I can improve this. And he came back with a bunch of issues. And I was like, oh. So I went to Claude Code and I said, okay, actually go through the source and tell me how you would fix this. And Claude Code comes back and says, I don't see those issues.
Omer Khan [00:41:49]:
I was like, what? So I go back to the first agent that I'd been working with and I kind of ask questions and it says, "Sorry about that. I should have looked at the source code before I made the recommendations." I was like, "So what were you looking at?" It's a very small example, but you can imagine how tiny little things like that could have huge consequences for somebody or some business.
Zhong Xu [00:42:14]:
Yeah, exactly. I think we are very much in the experimentation phase. I do think, look, AI will change a lot of our jobs and how the world works, but you need guardrails.
Omer Khan [00:42:25]:
I think a lot of cybersecurity and guardrailing companies will arise as we adopt more AI. All right, we should wrap up. So let's get on to the lightning round. I've got 7 quickfire questions for you. Sure. What's one of the best pieces of business advice you've received?
Zhong Xu [00:42:44]:
Timing is everything. And what do I mean with that is, people said like multiple times I succeeded, people say it's luck, but it's not luck. Sometimes you can, you know, timing is everything. For example, when you launch a product, if you don't see the wave happening, just wait, right? We could code it a point of sale app on Nokia N95s or, you know, on a Windows Pocket device, never would have taken off, right? So sometimes you just need to wait until the right time and medium is there. Similarly with Deliverect, if we waited too long to start, look, we didn't foresee COVID, but because we were just on time, we could ride that tsunami.
Omer Khan [00:43:20]:
So sometimes, you know, you need to go quick and sometimes you just need to wait.
Zhong Xu [00:43:24]:
What book would you recommend to our audience and why? There's a couple. I think The Hard Things About Hard Things is really good because it tells about, you know, the entrepreneurial journey. I think that that's a great one for people starting out. I think there's 3 books I would recommend: Zero to One, You know, you learn a lot from Peter Thiel, a lot of practices. If you're starting out in selling, it's the book Positioning. If you never read it, it's really, really good about how you need to get your customer view, you know, your product, your positioning is everything. And the last one is Crossing the Chasm. I think especially if you're an early-stage founder, you need to understand and don't get discouraged.
Omer Khan [00:44:05]:
There will be a lot of difficult times, but you need to move through the motions. Great books, all classics. What's one attribute or characteristic in your mind of a successful founder?
Zhong Xu [00:44:15]:
A chip on the shoulder. I would say grit, but it's a chip on the shoulder. I think most really successful founders have had something that they want to prove, you know, with all their might that, you know, people are wrong or whatever to prove themselves because else why would you do it? What's your favorite personal productivity tool or habit? Yeah, I use a lot of tools, but in these days I use Cloud Code. I actually use it to brainstorm. There's a great methodology that our engineering team used and I started using it called BMAT. So instead of, you know, you have agents that actually ask questions. And the reason why is I'm an ideator, but then, you know, I need someone to challenge me on my ideas. And so it's very hard to challenge yourself.
Zhong Xu [00:45:00]:
So, you know, AI is perfect for that. So it keeps asking questions to validate or destroy my ideas, and then it allows me to come to better conclusions. What's a new or crazy business idea you'd love to pursue if you had the time? I don't think I have a— I thought about this question, but I don't think I have any other mindset right now than running Deliverect. If I were to retire, I would maybe create, you know, run a vineyard, like, you know, make wine.
Omer Khan [00:45:29]:
Love it. What's an interesting or fun fact about you that most people don't know?
Zhong Xu [00:45:33]:
People don't know that I like to sit in traffic jams. I like to drive a lot. So because the first company I started, we coded half our times every day in traffic jams. We were sitting every day at least 2.5 hours in traffic jams. And so my co-founder was driving or we swapped around. I had 4G or 3G back then. So one was coding, the other was driving. But then if you think about it, 5 days, 3.5 hours, that's a lot of hours.
Zhong Xu [00:45:59]:
That's 16 hours, 17 hours you could work, right? So, and I still to this day, it's a great time to think.
Omer Khan [00:46:06]:
Because nothing is distracting. Yeah, actually, I love that too. I love to just get in the car and drive nowhere, come back with usually a better idea.
Zhong Xu [00:46:15]:
What's one of your most important passions outside of your work? I like driving. One of the things I do is I drive classic cars. So I've done last year the Mille Miglia. So long, long, real, you know, classic car races, or, you know, it's not really races often, but, you know, driving around, you know, just zooming out.
Omer Khan [00:46:36]:
How old is the car? How better? Love it. So Zhong, thank you so much for joining me. It's been a pleasure. Appreciate you sharing your lessons and story of building Deliverect so far. If people want to check out Deliverect, they can go to deliverect.com. And if folks want to get in touch with you, what's the best way for them to do that? Hit me up on LinkedIn. I'm quite responsive there. Sounds good.
Omer Khan [00:47:02]:
Thank you so much.
Zhong Xu [00:47:03]:
I appreciate you making the time and I wish you and the team the best of success.
Omer Khan [00:47:08]:
Thank you, Omer.
Zhong Xu [00:47:08]:
Appreciate the time.
Omer Khan [00:47:08]:
My pleasure.
