X.ai Case Study: Artificial intelligence that actually improves your life and business

I interview entrepreneurs about how they built their businesses for an audience of real entrepreneurs. In fact, as proof of it, one of our listeners is joining me here today. His name is Dennis Mortensen and he created x.ai.

Any time people say automation doesn’t work, one exception seems to be the software Dennis created. Here’s an example of what it does: You want to book a meeting with somebody so you end up going back and forth to find a time that works. When you settle on a time, you realize you’re in different time zones so you have to start the whole process over again.

Your business isn’t going down because of this pain point, but your focus is definitely being split by it, your professionalism with your contacts, with your clients, with the people you do business with is suffering because of it.

Well, Dennis had this bright idea to say, “You know, I think this is where artificial intelligence can actually help. I want to find out how he did it.

Dennis Mortensen

Dennis Mortensen

x.ai

Dennis Mortensen is the founder of x.ai, AI software that solves the hassle of meeting scheduling.

roll-angle

Full Interview Transcript

Andrew: Hey there, freedom fighters. My name is Andrew Warner. I’m the founder of Mixergy where I interview entrepreneurs about how they built their businesses for an audience of real entrepreneurs. In fact, as proof of it, one of our listeners is joining me here today. His name is Dennis Mortensen, and he created x.ai. Here’s what it does.

Dennis: Thanks so much for having me.

Andrew: Thanks so much for being here. This is like every time I say that chatbots don’t work, automation doesn’t work, the one exception that other people bring up or that I bring up myself is your software. Here’s what Dennis’ software does. You know when you want to book a meeting with somebody, what you’ve got to do is kind of go back and forth and say, “How about Monday at 10:00 a.m.?” And then they say, “Well, Pacific or Eastern?” You go, “Oh, I didn’t realize you moved back to the East Coast. It’s Pacific time.” They say, “Well, no, sorry, Pacific time doesn’t work for me. How about this other time?” So you say, “Yes.”

By the time it gets to them, you’ve booked that time with someone else, and so when they say, “Yes, I could take that time,” you have to say, “No, sorry, I booked it with someone else. It’s kind of a problem.” And you go back and forth. Not a huge deal, right? Your business isn’t going down because of it, but your focus is definitely being distracted by it, your professionalism with your contacts, with your clients, with the people you do business with is suffering because of it.

Well, Dennis had this bright idea to say, “You know, I think this is where artificial intelligence can actually help. I think if you just email this fake person, Amy Ingram, I think this like robot named Amy,” and she also has different names, “can organize this for you.” Go back and forth based on your preferences and what’s going on your calendar and schedule it for you. And that’s what that software does. Dennis, I see you’re little uncomfortable as I give that description. I’m going to find out . . . You’re not.

Dennis: I think it’s pretty solid. Spot on. We are hyper-focused on trying to kind of solve this one particular thing. And I think the way you described the pain is kind of spot on, even as you suggest, which I’m sure most people will be uncomfortable with that it’s not really a major deal. You know what? You’re right. But that’s because we don’t think of it as a major deal, right?

So you and we will do our taxes in the next couple of weeks and it’s a major chore and we’re not looking forward to it, but that’s because we’re doing that in one sitting. So I’ll do about two hours of prep, I’ll do about an hour and a half with my CPA, and then I’ll cry a little bit, and there will be a sum of four hours and I’ll have some memory of that pain from last year.

Setting up meetings comes really in increments of 40 seconds. So you don’t really remember that 40-second of pain. It’s just that you and me will do about 1,000 meetings a year. It takes about eight emails, including the invite, so three and a half emails that we send for both of us, and it takes about a couple of minutes to kind of compose the email. So we don’t really see it, but if we sum it up and looked at it in aggregate, it is way beyond what, say, you and me spend on our taxes. So you’re absolutely right.

Andrew: I agree with that. And what I’m curious about then is when people don’t feel the pain, how do you get them to care about the solution, right? If I feel the deep pain of taxes and I drive through an H&R Block or I was actually running through taxes and I saw several signs like these that were great, “H&R block, we welcome procrastinators.” If I felt the pain of, “Why did I procrastinate? Oh, if I don’t pay my taxes, I’m really screwed.” All that pain and I walked past a solution, I immediately feel it and I go and I sign up.

I wonder with you when there isn’t that deep pain that people feel so badly that they can’t wait for a solution, how you convert them into customers. Let me just quickly say to this, my interview is sponsored by two companies. The first will help you hire phenomenal developers. It’s called Toptal. The second will help you host your website right. It’s called HostGator.

So, Dennis, I have found that solving pain is the best way to create a great product that people will sign up for. What do you say about that? How do you handle the fact that people don’t feel it acutely over and over in a big way?

Dennis: We certainly don’t have much pushback when we go pitch investors, the enterprise, the busy professional for where if I asked hand on heart, “Do you like setting up meetings?” The response is always, “No. I fucking hate it. If I could somehow escape it, that will be awesome.” But going from, “That will be awesome,” to changing your habits tomorrow, that is not necessarily easy.

So typically what happens with products like this, not just for me, but for anybody is that you let on to those that are in the most painful where, “Yes, I’m not really fond of changing my habits because yesterday was not completely shitty, but I’m in so much pain that I’ll do whatever to see some of it just disappeared.” So you’ll see us very easily latch on to say SDRs and AEs and recruiters and entrepreneurs and founders and what have you for where we do so many meetings that, “Help. I’ll try anything.” Almost like any other little productivity tool for where why wouldn’t 100% of the people who got email have some sort of text expander? Why are you writing the same thing over and over again? I say that should be 100% penetration on that. You and me probably use it because we write a lot of stuff, but . . .

Andrew: I see. You’re saying that for salespeople who need to get on calls, that is a pain that they feel a lot and that’s what you can bring up for them for investors who do have . . . people who have lots of scheduling and where the scheduling determines their money or their interaction with their customers, they’re good customers for you. On your website, though, on the homepage, you just say, “Scheduling sucks.” Do you then target salespeople differently, enterprise customers differently to make sure that they feel the pain that you’re solving?

Dennis: So, again, any company or usually when you and I kind of take a step back and look at our ventures and in our ventures trying to kind of figure out what particular segment should we kind of go attack here. And as we try to go and attack that segment is either some sort of vertical that we’re trying to attack, say, some particular industry or some particular profession, but for us, that’s not really the case. So we try to attack anybody who does a large amount of meetings and a large amount of meetings that are probably external. And in that setting, we tend to be a good fit. It’s just that certain professions tend to have a lot of meetings that are [inaudible 00:06:32]

Andrew: How do you target them? So how do you . . . Give me an example of a group that you target, and then what do you do to target them?

Dennis: The particular targeting for where you and me go buy some segments on Facebook or Google or some other channel, it’s certainly a little bit more tricky for where you can only really buy against the existing kind of dimensions that we just talked about. So how do I find these particular people? And I think in that case, would have worked well for us have been content. So people who are . . .

Andrew: That’s what I figured.

Dennis: . . . in pain, they’re not necessarily sure even how to describe it, but they’re suddenly on the lookout. So I would say if I go look at all of the channels that we’ve done experiments in over the years, content have worked quite well as trying just to kind of talk about the pain that we’re in, and even removing some of the anxiety that comes with changing your habits as if, “I used to do this myself. I’m a little bit nervous about whether there’s some sort of social stigma attached to me not doing this given he is one of my leads. I want to treat him nicely.”

And in that trying to really build up enough content where people can see that, “You know what? I think this is going to be normal.” Remember, there’s been plenty other tech for where it was not normal, like talking on your phone in Whole Foods. You used to be an asshole if you did that. If you don’t do it today, how are you doing your shopping? It used to be a kind of a rude . . .

Andrew: I get it.

Dennis: . . . to ask for email, right? Because we’re supposed to send letters.

Andrew: I get it, by the way, and I see some of that in your marketing. So I’ve seen some of your posts. I don’t see you specifically in your blog post say, “Here is 10 tricks for getting more appointments with your clients.” What I do see a lot of is, “Here’s how artificial is going to change your life. Artificial Intelligence is going to change your life. Here’s how we see this as a weirdo thing, but it’s going to work.”

Like for example, I’m looking at the Trello blog right now. You published a few articles on Trello, you personally. And one of them is, “Artificial intelligence is about to make us all managers, but are we ready?” Or there’s another one where you talk about how artificial intelligence can help us manage our time somewhere else. It’s stuff like that, and that then to the people who are into futuristic stuff who go to Product Hunt, they read this and say, “I’ve been looking for a solution, but also I’m willing to try something because it’s new,” and they sign up.

I see you nodding. I want to find out how you built up this business, how you got into artificial intelligence, how much money you raised. But let’s like talk about how big it is to give people a sense of whether this is just like a new thing that’s starting out and you’re hoping to make well or something that’s worked. How many customers do you have? Give me some metric that shows me how big you are.

Dennis: Let me take a step back and then I’ll give you a non-answer, but you can keep kind of pushing me here. But to give you some perspective, there’s a little bit above 10 billion formal meetings being set up in the U.S. alone every year by some 80, 90 million knowledge workers, people who comes into an office, touch a computer in some way, shape or form.

Andrew: Okay.

Dennis: And that is probably pessimistic. That’s just about two and a half meetings per week. And I think most people would agree that we probably do more. So that’s what we are attacking here. We set up hundreds of thousands of meetings, but even as I say that and I kind of half-pat myself on the back side saying, “That’s not too shabby.” It is still tiny. That means there’s actually no single vendor in market that really had made any dent. So the vast majority of what being scheduled today is being scheduled by you and me, people . . .

Andrew: People still. But you know what? In my world, what I see is a lot of Acuity Scheduling, a lot of Calendly, a lot of . . . What’s the other one? There’s a whole suite of them. I’m trying to figure out, are you . . . You’ve definitely raised more money than all those. Do you have more revenue than them, more paying customers than them? Can you give me any sense of that?

Dennis: I think there’s a few folks in place and there’s folks on one end for where you work, say, the direct insert and/or the one-on-one, so the Calendly, Acuity, and so on so forth.

Andrew: Those are all the software where you give someone a link to a calendar where they pick the dates from what’s available to you.

Dennis: Exactly. Andrew shares his availability with Dennis under a certain set of constraints and I picked that today, Tuesday, noon, EST we’re supposed to kind of meet up. Then there’s the kind of getting you me and our colleagues, buddies together, Doodle style, here’s a voting recognition, go figure out when we can almost make it and somebody needs to kind of make a compromise in that setting. And then there’s all the other meetings where’s you, me and another guy, you, me and the candidate, you, me and my sales rep needs to speak to the customer, and all of those things where, “Oh, I need to meet the product design team.”

Andrew: Dennis, now we’re getting into how the products are different. What I’m curious about is the size. Can you give me any sense of size other than number of meetings because that doesn’t register to us? How many users do you have? What revenue?

Dennis: Yeah. In that setting? I think that’s really three actors, Doodle in one end, Calendly another end, x.ai in the middle. I think we are all making somewhere between $1 million to $20 million. None of us are making anything that is really kind of significant . . .

Andrew: Are you doing over 10 million in revenue yet?

Dennis: No, we’re not.

Andrew: No, you’re not. Okay. All right. It gives me a sense of scale. Let’s understand how you got . . . By the way, I still love this so much. As we were freaking talking, I signed up. It took me all of a minute. I was going to actually just put my contact information in and then tee it up for later in the interview to say, “I signed up.” And then it just happened too fast so I didn’t even get a chance to like walk us through together.

The URL is x.ai. So that’s what I used to sign up. I signed up before. In 2017, I had a huge problem. I’m going to bring it up in the interview later on, but first let’s go back in time and then work our way to 2017 when I signed up, and then what happened as we grew. You are a guy who . . . A serial entrepreneur is actually the accurate word to describe you. Most serial entrepreneurs have dabbled in a bunch of different things that didn’t work out. You actually have created stuff. I want to know a little bit about those businesses and a little bit about your past. Let’s go back to even before you were fully officially an entrepreneur. You’re a guy who had dad and uncles who were entrepreneurial. You watched them work really hard. How hard? Give me a sense of how hard and what they did because it made you said, “I don’t want to be an entrepreneur”?

Dennis: The funny thing is I’ve seen it real up close, my dad, uncles, cousins, what have you. And there’s certainly a picture of me seeing my dad getting home late, having dinner in the living room, falling asleep in front of the telly, working six days a week. For the 18 years I spent at home not [as in 00:13:34] once in a blue moon my dad will kind of work on Saturday and that was normal. As in, Sunday was the only day off. And I just looked at that from afar thinking, “I’m not so sure I’m ready to going to sign up for that. So how about I just go take my CS degree and go work for IBM, and it’s going to be awesome.”

Remember I’m Danish, so awesome means you get home at 4:00, you get your seven weeks of mandatory vacation, and I’m going to be living the good life. And somehow it just didn’t play out like that. And I’m sure that we can talk about exactly how do you go craft a new entrepreneur. I certainly think that having them being exposed to that environment for years on end is a good way. So the original plan to kind of go work for IBM didn’t really pan out. And I started my first venture right out of college.

Andrew: In fact, even before you said, “Look, I didn’t want to be an entrepreneur.” This is what you told our producer, Arie Desormeaux. “The problem was, I was . . . ” And this is a quote, “I was always hungry for money.” And I get that. Right? Even as a kid it’s not like you’re trying to like have everyone else’s money, you’re not trying to, I don’t know what, be Mr. Burns. But there is like this passion for accumulation, and so you started to have these small businesses. Like what? I’m looking at my notes here. I see a few.

Dennis: So you’re actually right. So kids want a new BMX, some new sneakers. We all have things we want now, not next week, not this summer and not at my birthday. I want them now. And the only kind of real income we have is birthday and Christmas. Those are the two paydays you have as an 11-year-old.

Andrew: Right. And everything else, you don’t get to pick what you get, you don’t get to get the good stuff. Uh-huh.

Dennis: Exactly. So I bought fruit from my dad wholesale and would go door to door and sell it. And if there’s anything, I certainly learned a ton when I took my computer science degree, and I’m very happy that I did. But of all of the skills that I’ve picked up, perhaps what I picked up between, I don’t know, 9 and 14 where I knocked on probably short of 10,000 doors, that the sales gene is very difficult to kind of pick up and call us. As in that’s something for where most people are just not super comfortable. It’s kind of like you can’t really learn how to date well. You probably just do enough of it until you stumble into some really good setting. And that 10,000 little sales calls that I did when I was super young, selling fruit door to door was paying for my sneakers, my new BMX and what have you, that was a really solid entrepreneurial training.

Andrew: Fair to say, I’ve got here in my notes, you knocked on 5,000 to 10,000 doors since you were 11 years old.

Dennis: Yeah.

Andrew: And you got . . . Our producer said, “Tell me what you learned from this.” And one of the things you learned was to just flex and grow your rejection muscle. What do you mean by that?

Dennis: I think it’s almost difficult to not get attached to some level of disappointment. When you ask somebody to also love what you love when that is a product you want to sell, some idea what you have. And when you get that rejection, most people just somehow find it very hard to see beyond the rejection and not learn from, “What is it that he does not like?” or “Why is it that I shouldn’t have asked this particular person this particular question?”

And if you do it hundreds of times a day, thousands of times a year, and year by year, you suddenly pick up on some idea of they can’t all be yes. It’s just the acknowledgement of the fact that if I do this, they’ll be a large amount of no’s and it’s going to be normal. It’s nothing to do with me. That’s just how the game is being played. And I think whatever muscle you train, that rejection muscle is perhaps not scientifically proven, but I think we all will have some version of a rejection muscle and some of us just work it better than others.

And if you are not used to it, you should somehow figure out, “How do I train that so that I can come into work tomorrow?” And especially as entrepreneurs, right, where it’s just one long string of rejections so that you’re equally excited on the next call or the next pitch or the next customer or the next hire. And that level of stamina, having picked it up early, it’s certainly something I didn’t really think about when I was super young. All I thought about was that pair of sneakers, and if it takes me another 80 doors, I’ll do 80 doors. Which I could see and visualize the sneakers, and the sneakers arrived. Sometimes it was 90 doors or 50 doors, and you kind of perfect your little pitch. So I actually do think if you can somehow early on figure out, “Where do I go train that? What is the gym I go to?” And for me it was door-to-door sales selling fruit.

Andrew: And you also, fair to say that . . . Well, here’s a quote that I got from you. “Being good at sales is more important than being an engineer, designer, data scientists, etc.” Fair?

Dennis: I think if you just say better without context that seems unfair. If we suggest what is the primary skill you need as a founder.

Andrew: Yes.

Dennis: I certainly believe coming in with a core skill. And you can see many times that the companies built up around that core skill which you have, that is wonderful. But if you’re unable to sell, I am typically skeptical of you having long-term success if it’s something that you’re not willing to learn. So of all the things I could pick, if I could only pick one primary skill as a founder, it will be sales.

Andrew: I’m with you. I agree, because you could always persuade a developer to come join you and other developers to come join you, but if you’re bad at sales, it’s just you. If you’re turning people off, it’s just you. If you’re hurt, every time somebody says no to you and you can’t then recover and go find somebody else, you’re really in trouble. Just to round this out before we move on to your better more mature entrepreneur . . . or not better, but your later stage entrepreneurial experience. Your dad was doing what?

Dennis: He sold fruit door to door for a while.

Andrew: Wow.

Dennis: Thirty years, 40 years perhaps.

Andrew: Wow. What a painful thing to do. All right, let me come back to the companies that you started. x.ai is your fifth company, you had four before, three of them had exits. One of them was painful, very, very painful. We’ll cover those a little bit and then we’ll get into how x.ai became this model of how to do artificial intelligence right.

My first sponsor is a company called HostGator. You know what? Usually, Dennis, what people do with interviews is they try to get a big audience. Let me give you another suggestion. Like if all you had was a HostGator account, here’s what I would suggest that you do and you want to build your business. Do interviews with people that you want to get to know as like business development. But if you call them up and you say, “Hey, can I talk to you to see if there’s an opportunity for us to do business?” they’re not going to want to talk to you.

Instead, what you do is you say, “I’m going to do nothing but business development interviews.” And let me give you an example. There’s someone who wants to do . . . There’s someone in my audience who wants to get lawyers to do lead gen for them. Now, calling up lawyers and saying, “Can I talk to you about how I can sell you lead gen,” is kind of a pain because first of all, they don’t really think about words like lead gen, they don’t talk to strangers like that.

So what instead he decided to do is do an interview program, where he would just call up lawyers to understand how they got their clients, to understand how they service their clients, and then put it online. It’s very flattering for them. They spend an hour now, they get a little bit more Google juice, a little bit more attention online. Fantastic.

Here’s what happens at the end of the call. They often say, “By the way, what do you do?” He says, “Well, I do lead gen for companies like yours.” “What does that mean?” “Well, what it means is, you know how you’re always are looking for clients and in the old days you might put your name in the phone book? Well, guess what? That doesn’t work anymore, but Google is way more powerful than that. It doesn’t cost that much. Facebook is incredibly more powerful and doesn’t cost that much. If you like I can talk to you right now about how I can get you leads, and many of them will convert into clients and I know because I’ve done it for others.” Boom, now you’ve got a sale.

So if you’re out there and you’re trying to do Biz Dev whether it’s you as the entrepreneur who’s doing this and you want to round out your relationships with people in your industry, or you’ve got a sales team, and maybe you say, “Look, one hour a week.” What you’re going to do is you’re going to get to know the people in your space by interviewing them. Many of them are going to turn into customers. All of them are going to teach you more about your customers.

All you have to do is go to hostgator.com/mixergy, get a brand new website, give it its own site, its own name, its own field so that when you call clients up it doesn’t feel like you’re clearly shoving some sales proposition down their throats. And then you launch your site. Right? That’s all it is. Hostgator.com/mixergy will make it so easy for you that if you get that middle option, really inexpensive, you can launch dozens, in fact, more unlimited websites all for the same low price which means that other people on your team who see that this works can start to do these kind of informational interviews that they publish online and then they help get more business and help bring more relationships to you.

That’s one of many uses, Dennis, of HostGator. If anyone wants to get inexpensive hosting done right that will scale up with your business, I urge them to go to hostgator.com/mixergy. By throwing the /mixergy at the end, let’s be honest, Dennis, they give me credit which will help me with my sponsors so incredibly I’m appreciative of it. And number two, they’re going to get the lowest price available from HostGator and they’ll be attached as one of my customers which means my team and I will always stand behind them. Hostgator.com/mixergy. How’s that for sales, dude?

Dennis: I like it. I like it.

Andrew: Right.

Dennis: There’s many things you can fake in business. Enthusiasm is probably not one of them.

Andrew: No. I wish I could. Sometimes I really do wish that I could. Sometimes I just don’t have the energy to do a good promotional but it just sucks and I have to be honest and call myself out. Other times it just works.

Let’s talk about the one that worked the best for you, the business that worked the best for x.ai, and then the one that was the most painful. My sense is, tell me if I’m right. My sense is that the one that did the best was Visual Revenue.

Dennis: That certainly did very well. And we ran that from beginning to end over 1,059 days, raised only a small seed and small A round. I still owned more than half the company personally when we did the exit. We had found a really good niche and really kind of leaned into it. So three fantastic years.

Andrew: What did it do? This is . . . You know the ads that go underneath blog posts and other . . . Well, how would you describe it?

Dennis: So we did predictive analytics for editors. So if you go to cnn.com or Forbes or The Guardian, Time, Le Monde, wherever you read your news, and you see that set of stories that they picked for the day, the editorial ones, there will be a big ass story in the hero spot, some to the right, some below. And if you look at CNN, they’ll be about 150 stories on that homepage. They’ll update it about 200 times a day. How do you do that? If you go back before us, mostly through gut feel and Diet Coke and that is not extremely data-driven.

So what we came up with was some algorithm for where we could do recommendations to editors on what stories to carry, where, for how long, when you kill them, what new stories to put in place. That said, the company that acquired us, Outbrain, they did the kind of recommended articles below stories on the article pages, but we did all front page recommendations to editors on only editorial content.

Andrew: Oh, front page only and only editorial.

Dennis: Yeah.

Andrew: Got it. And then Outbrain acquired you and what they do is they do those ads, those advertorial type content, those square blocks underneath the article that says, “Someone from San Francisco just earned $1 million,” or, “Obama wants you to have these tax savings,” you click it and then it’s an ad for something. Got it. And so how much do they acquire you for and why would they acquire you?

Dennis: So number is not public, but they would acquire us because they had a really good, stronghold of the article page. As in, they are still the primary leader in the space. Go to any kind of article today and you’ll see a set of links below and the likelihood of it being Outbrain is very high. They didn’t have any exposure, really, to the front page. And the front page could be the homepage itself or /sports/political/something else. And those were the very pages that we provided insight on. So that was kind of an inroad into the other half of any kind of news destination.

Andrew: Okay. Did that one recover what you lost when you went bust with the previous business or the one . . . Actually, wait. This is two businesses before, right? This is Evonax?

Dennis: Yeah.

Andrew: Evonax . . . So just to set the stage for people. You leave school, you start Canvas Interactive, it’s an interactive agency, right? It gets acquired?

Dennis: Yep. I’m one of those kids who got out of the right side of the dot-com boom. We sold in April 2000. So when people say, “What a horrific time.” I lean back and say, “No, I kind of liked it.”

Andrew: What did you get acquired for? Did you get cash or equity?

Dennis: A mix and we got acquired for about $11 million. And I owned about . . . We were bootstrapped, so I owned a little bit about 40%.

Andrew: Wow.

Dennis: So we didn’t take any financing. So I took a fair amount of money out of that. And what I did then was that I put all of it into my next venture.

Andrew: About $5 million . . .

Dennis: Yep.

Andrew: . . . into the next business which was a way for people to book restaurant reservations, right?

Dennis: Which seems so normal today, right, in 2019. Where else would I do it? Or suddenly you wouldn’t call it. That seems silly. But back then that was kind of radical. And it was perhaps a little bit too early. And again, you’ve heard all these stories where the ideas themselves aren’t necessarily bad. They just tend to be a little bit too early, as in the timing wasn’t perfect. And the funny thing is, even to this day, I have this kind of entrepreneurial lust for where, “You know what? Once I’m done here, I want to go back. I’m going to play that game one more time just to show people that it was a good idea.”

Andrew: The game of letting people book restaurant reservations.

Dennis: Yeah.

Andrew: And this . . .

Dennis: It’s too late now because . . .

Andrew: I’m looking at an old version of the site. This is like, I could get pizza. And this was in Denmark, right?

Dennis: Yeah.

Andrew: I could . . . It’s more than just booking reservations, though, right? It seems like I can even order from there.

Dennis: Yeah. So what we did still sounds like a good idea in my mind, but it was kind of what we died on. So if you go to Seamless or Grubhub or any one of those vendors today what you really have is a marketplace for where I need Thai, so I’ll go find a set of Thai places, I’ll order from a place. If it’s late, if it’s not good, if it’s anything which doesn’t really fit my liking, I blame the Thai place, not Seamless. I might even give them a one star. So Seamless survive, the Thai place might not.

What we did, though, is that we rebranded it so that you would order from us, it will come out in our packaging, and we could have a slightly higher margin, but we will take all the blame. And that meant we ramped up revenue rapidly, but we ran this kind of Domino’s type thing for where if it wasn’t to your liking, we would give you a refund. So as we kind of ramped up revenue, it cost me really because I couldn’t get the quality right. And I thought, “You know what? I’ll solve that at scale. So I keep taking my money, I’ll keep putting it into it, and then at some point, these two lines will cross.” They did not.

Andrew: And I could see then you had to close up. You said you lost everything.

Dennis: Yep.

Andrew: I don’t want to make this into a hero story where nothing goes bad for you. Let’s take it to that moment where you knew this wasn’t working out. Take me to the darkest point. How was that for you?

Dennis: Here’s the funny thing. Three months prior, a major competitor in Europe called Just Eat. It’s probably a £2 billion, £3 billion company on the London Stock Exchange today were in kind of acquisition negotiations with us which I declined, as in, “I don’t think this is going to be a good fit. I have a good thing going. I can outrun you guys.” Not years before, like, three months before. So it’s not like I didn’t have any opportunities, I kept running it forward until I just literally ran out of cash.

And on that, honestly, the way it worked is I was just in my bed, in my underwear with my mobile phone looking up at the ceiling and I called the founder of Just Eat and said, “Guys, you win. You can buy the assets for next to nothing. I’ll keep my laptop and I’ll keep my phone and I’m going to buy a pair of tickets to the Caribbean and do a reset.” And three months later, it kind of went belly up. I got a little bit of money just enough to get me to the Caribbean.

Andrew: And that was it. And you ended up in the Caribbean. And did you feel like a failure? Was there a period there where you felt like, “I thought I was smart. I thought I had it all. And then maybe I’m not”? Was there any doubt?

Dennis: I wish I could give you the kind of raw version of, “No, I never felt any of that.” But you know what? It just sting a little, as in, that one was a nice apartment. That was a nice antique pool table. That was a good setting me my wife and my two young kids I kind of put in place. My youngest was like nine weeks. And basically, had a memory of, “Oh, I think we’re set for life.” And then wake up one morning, “No, we’re not even set for next week.” And that did sting a little bit, but it didn’t sting for long. As in, really.

And I’ve certainly come to kind of look at it like that over the years, perhaps not that day but it was a really good learning as this idea of you and me really just being professional footballers. As in, we play a set of games, some of them we win. We certainly go on the field trying to win. Some of them we lose, we learn a little bit, we come back stronger. But it’s really just not a reflection on you as a human being.

Andrew: How do you get to that state? I don’t think I could get to that state very easily.

Dennis: And I think you’ll see . . . I’m certainly not the only person, but certainly suggest we see many entrepreneurs end up in one of those two buckets, right? The bucket for where my life’s worth or the whole way I feel about where I’m at depends on the success of my startup at this moment.

Andrew: Yes.

Dennis: And then there’s people for where I’ve been able to somehow just turn this into a game for where I am going to emotionally detached from its success. It doesn’t mean that I’m not playing it to the best of my ability, but no matter what happens, I can go home now. Those two girls, they still love me. As in, this is not the end of the day. That took a little bit of time. But I think perhaps that experience was the one that solidified it for me for where, “Dennis, it’s just a game. Don’t think about it too much. Take all your skills, go play some more games. And if anything, here’s the way you should perhaps think about it was slightly more rational.”

So if you’re a VC, you run some funds, you invest in 20 companies, you expect most of them to not work out, some go sideways, one or two kind of win. Well that’s because you can do this in parallel, but that means you and me as entrepreneurs, we probably need to run a similar fund strategy. I’m just running the 50-year fund for where I can do 10 investments, but it’s going to take me a long time. And I have to accept that in my fund, you know what, many of them won’t work, and that is absolutely okay.

Andrew: That’s a very reasonable way to think about it. It’s really hard, though, when you have to feed your kids, when you have to take care of a house for your family. How were you able to do that?

Dennis: That is the part for where if I even just take it one step back . . . So I remember, this is a funny story. I’m sure my wife don’t want me to say it, but still. So being a Danish and probably what most my American friends will call a complete socialist, was not the case. It’s some setting for where she would buy the pill for three months, I would buy the pill for three months. I was running my first startup and I’m just so busy. She comes back and say, “Hey, Dennis, if you don’t buy the pill, out comes kids.” And I tried to give her the whole kind of rattle. “First, I know how this works. Second, is not that easy.” Five days late I get us back on track and it was too late. So I had to . . .

Andrew: Okay.

Dennis: So I couldn’t even blame anybody. I knew what was about to happen. I’ve been warned. And now I’m 26, she was 23 which certainly compared to our parents is kind of normal, compared to kind of our generation is a little bit young, right? Where I have friends who end up getting their kids first day in their 40s, right? So I was actually just a little bit scared that when you get this kid, what you used to be will disappear. They’ll be some new version of yourself, but a version that probably wouldn’t be able to do the 50-year fund. So you need to kind of now figure out what that new version looks like. But that was, at least for me, just not the case.

Andrew: But then wait. How were you able then to feed your family?

Dennis: You just go frugal. So when I sold the assets, I sold it for just enough for where here’s a little bit of money in my hand, you won’t survive in Manhattan for long, so go find a place for where you can survive a little bit longer.

Andrew: And that’s what you did?

Dennis: Yeah.

Andrew: You went to a place . . . Where was that?

Dennis: Trinidad, Tobago.

Andrew: Got it. So you went to Trinidad where . . . And you lived there for how long?

Dennis: Not for long, but just long enough where, “Hey, I can live here on whatever money I have in my hand and not be afraid. We can feed ourselves.” And you look around thinking, “Do I fit in here? Probably not. But I got a responsibility to feed those kids.” And you just go with it.

Andrew: And you were starting to say how much it costs you to feed the family, like . . .

Dennis: It’s just a 10th of what it cost to feed your family in Manhattan. Right?

Andrew: Yeah, yeah.

Dennis: So don’t move to Manhattan. Go move to . . .

Andrew: I had a similar situation when I was in Argentina. I lived it up and it was $1,000 a month for rent. And then the food was super inexpensive. So I get how you can kind of pause time by living like that. I want to quickly go into the next business. The next business was Indextools. It was acquired by Yahoo. It was actually called . . . What was it called? It was called Tensa, right, was the name of the company. Am I pronouncing it right?

Dennis: Yeah. That was the mother company, but the tool and product and us doing business was as Indextools.

Andrew: Indextools, web analytics software with the focus on delivering highly customizable and scalable analytics platform. Where did you get the money for that or did you just bootstrap that one?

Dennis: Entirely bootstrapped. So moved from London, England to Budapest, somehow sold my wife on that, spent a good four years there. And remember, this was before they joined the EU. So just like I have actually very little understanding of what coastal cities in the U.S. and how they compare to mid-U.S. because I’ve only seen SF and Seattle and LA and Boston and New York and DC. Actually, I don’t really know what’s in the middle, but I’m sure it’s not very similar to Manhattan. And in Europe, just like Western Europe, it doesn’t look like Eastern Europe. And people say, “Sure, sure. It’s all Europe.” No, it’s not. In Eastern Europe, they look east. So that means they speak Hungarian. First-second language is not English. It’s Russia. And that means that was certainly interesting. So we moved there and spent a good four years building up that company entirely bootstrapped and then we sold it to Yahoo. Martin, Peter, me and Charlie were the owners of the company.

Andrew: Okay. And again, you were able to do this because you kept your costs low. And this is soon after you got back from Trinidad. It seems like you stopped in England and then quickly went to a cheaper place. All right. We’re going to get to x.ai right after I say to people, look, you’re going to listen to Dennis, I’m going to talk about my second sponsor, you’re going to listen to Dennis talk about how he’s got this business that is focused on artificial intelligence.

Now, some people listening to me might say, “You know what? I think our business could use some artificial intelligence too. I could see some opportunities for how we would fit in,” or, frankly, anything that’s a little bit exotic and a little bit cutting edge and maybe you don’t have the expertise in it, like, maybe blockchain or I don’t know what it is. There are tons of different things that are brand new.

All you have to do if you need that kind of development chops, if you need somebody to come in and advise you on it, somebody to help add that kind of intelligence to your business, if you go to Toptal you get the best of the best. If for nothing else you get to consult with them, or you get to hire them on a part-time basis or a project-based basis to see, can you incorporate some of this intelligence, some of these superpowers into your software? And if you can, does it make a change in your business for the positive? If it does, you can hire them and keep working long-term. If it doesn’t, you say, “Hey, you know what? It’s been great Toptal. I think this developer needs to go back. I’m sorry. We’re not going to get to work with them.”

That’s the beauty of Toptal. Best of the best developers, so when you get an idea of how you can supercharge your business or just bringing another incredible developer into your team, all you have to do is go to toptal.com/mixergy. When you do, you’re going to get an unbelievable offer that I can’t believe they still keep up for us, but they’re not offering it to anybody else. It’s 80 hours of Toptal developer credit when you pay for your first 80 hours in addition to a no-risk trial period. All you have to do is go to toptal.com/mixergy. That’s top as in top of your head, tal as in talent .com/mixergy. One final time toptal.com/mixergy.

Okay. You had this idea for solving the problem of scheduling. And the first thing that you did was not sit down to code. You wanted to understand . . . Actually, you know what, before I even go to that, I love this thing that I’ve got to get to. You have a list of hate. Tell me about the list of hate and give me an example what would be on the list of hate.

Dennis: So that sounds very sad, right, as a [meanest 00:41:38].

Andrew: No.

Dennis: You should get your life back on track. But I’m certainly not a believer of this idea that you and me can get together at my apartment next weekend, we’ll get a couple of beers, pizzas, work the whiteboard and come at the end of the evening you and me know exactly what we’re going to spend in the next five years on. That seems just completely unrealistic to me that somehow that particular evening will provide some epiphany, there’ll be kind of a guideline for the next five years.

What I find way more realistic is to slowly but surely keep a list in your pocket, which I call my list of hate of all the things that you stumble into, that you dislike so much that you have to put them on list so that one day you come back, take a look and potentially kind of remove that particular pain. And I just keep that list on my smartphone. I never look at it, but whenever I spot something like standing in line at Citibank thinking, “You know what? This is shit. I should start a retail bank.” And that becomes a one-liner and your kind of the list of hate. And then a few days later something else kind of happens for where, “This can’t be how it’s supposed to work,” and you add that to your list of page.

And then one day, you end up in some exits and you go back to your list which is now 15, 20 pages long of one-liners, two-liners, many items being repeated over and over again because you’re stumbling into them often and you start to say, “Oh, this is interesting.” And that is four years in the making, not one evening and a couple of pizzas, but four years in the making. And that list tends to be the catalyst for what’s next.

Andrew: Here’s why I love that. I think it sounds sad or mean spirit or whatever. I have found after doing over 1,500 interviews here on Mixergy that the surest way to success is to find a pain and address that pain. You don’t even have to solve it. If you show a little bit of emotional connection with the person who’s experiencing it, they love you often enough to pay you to move you to the next level. That’s how bad a pain is. And so on your list of pain was booking and scheduling meetings. And in order to figure out how good this was, you decided to run an experiment with someone who was not even a co-founder, not really part of any business. Tell me about what this experiment was and how it worked out.

Dennis: So the first thing I did was just to at least see, is this a personal fantasy of mine for where really what you just want to be, Dennis, is SVP of whatever over at IBM and you want a human assistant? If that is the case, spend some of your money on getting a human assistant then. But what I thought I could do was walk over to my VP engineering, Alex, and say, “Hey, Alex, I have this idea. How about you and me play each other’s assistant for the next couple of months.” So that means we set up an email account. Any meeting you want to have set up, you use that email account and I will act as your assistant. And we call the email account Amy. And what we’ll find out is two things. One, is there any real joy in handing this over to somebody else or is it one of those fantasies that you have, right?

Andrew: Yeah.

Dennis: Like, “I want to go live on the beach,” then you spend three months on the beach figuring out, “I don’t want to live on the beach.” So is this just a fantasy? Two, can you actually schedule meetings with very little context? As in, can I as a human just immediately started working for Alex without many instructions? And in those couple of months, we figured out two things. One, to begin with, the joy of asking somebody else to do this for you. “Hey, Amy, set something up for Andrew and I first week of May when he’s back in Manhattan on 200 Broadway, please.” That joy of clicking send and immediately clicking archive, that was great.

The ability to schedule it was fairly simple. And the funny thing is that we immediately picked up on the fact that even on a data set of one, we started to see things kind of replicate. It’s not like Dennis will do only unique meetings. Many of them just look alike. All these interviews or candidate kind of screening calls, they’re all the same. Even his little kind of sentences are the same. I think he used some sort of text expander, probably. So it just started to kind of replicate.

And then, of course, the really funny thing was, I fucking hated scheduling meetings for Alex. As in me being his assistant was the worst because now I have to do it. They’re not even my meetings. There’s no joy of imagining, “Okay, at least I’ll meet this person.” So that we did for a couple of months and it worked very well. It was a very good experiment.

Andrew: And it showed you that, look, hiring a bunch of people to do this is not going to be the fun solution, it’s not going to be a solid solution, but it can be automated because Alex is asking for meetings in a consistent way. We’re not like randomly saying, “Can you . . . ” And I’m trying to think of some random thing. If it’s shopping, for example, it might be a little bit more random. If it’s all the other tasks that I asked my assistant to do, like I’m about to ask her to pay for my health insurance, she’s going to have to deal with all the different health issues that go, all the different bureaucratic issues to figure out who to pay for it. That’s going to be infinitely crazy. Here, it’s solid, it’s clear, it’s repeatable. I get it. You decided, “You know what? This makes sense.” Did you then expand it to other people before deciding to sit down and code it?

Dennis: Yeah. Then we ran a step two in the experiment where perhaps Alex and I know each other too well. So let me go hire a full-time assistant. I’m going to ring up 30 of my kind of friends, tell them, “I know you’re a middle management. I know you don’t have an assistant, but you know what? I hire one for you. Here’s the email address and here’s the requests that you can do. Only ask for this and go wild.” And we ran that for a month, short of two months just to kind of see, “Hey, can Barbara deal with all these friends of mine for where I know with 100% certainty she do not know them? She doesn’t know any of the quirks but can the very text which they kind of put in place be enough for her to kind of manage their calendar?” That also worked well. They also all loved it. And that was the kind of catalyst for us to kind of put the team back together, raise a little bit of capital and get started on the whole thing.

Andrew: How much capital did you raise?

Dennis: All in all, about $40 million.

Andrew: Forty, overall. In the beginning?

Dennis: Seed round, $2 million.

Andrew: And for the first version, now that you’ve got $2 million and you’ve got a problem that you know exists, a solution that you think is cookie cutter enough to start an artificial intelligence journey on, what did you code up? What did that first . . . What kind of intelligence was in that first version?

Dennis: So the funny thing is plenty of people who will either pitch me or just generally kind of come up and have a chat with me, they have this fantasy of, “On Monday, we get together, we talk about it. And on Tuesday, we download TensorFlow.” And you know what? You’re not. You download . . .

Andrew: On Tuesday, you download what?

Dennis: TensorFlow or whatever machine learning framework that you’d want to work on. So that is just not how it works. So the first thing we did was that we built what you would probably describe as a data annotation console. So what we needed was a data set. So this kind of feeling of this being repeatable, I needed to kind of quantify that. So the very first thing that we built was a ability for a set of humans to do this task in a way that could label the task as they did it. So we did that in the first three months for where . . .

Andrew: Humans would actually sit and do the booking, but they needed to also tag the requests in a way that allowed your machines in the future to understand how people request.

Dennis: Yeah.

Andrew: What did you do for that? You created your own email software or did you use something else?

Dennis: Yeah.

Andrew: You did.

Dennis: Yeah. You should think of it really as some sort of kind of Excel and Gmail amalgamated into one solution for where you should be able to both reply and mark this whole thing. And the only outcome that we promised off of our seed round was one for where we will come back to you, dear investors, come the end of this first year and tell you that this is indeed tractable, as in, what we think is possible is indeed possible. You won’t get MVP, you won’t get customers, you won’t get revenue, you won’t get anything but a thumbs up or a thumbs down.”

Andrew: It’s like a science project. Can we actually categorize all the ways that people request meetings and request changes to meetings in some kind of ordered finite way?

Dennis: And I think you hit on all the right keywords here for where the job was really one for where you need to be able to define the very universe that you exist in. So just like the self-driving car will not have a model of the real world will have a very simplistic model of the universe it exists in—cars, bicycles, pedestrians, and so on so forth. We also needed some finite version of the universe that we exist in, probably a set of intentions and entities. And once you have that, you need to make sure that there’s no kind of long tail attached to it. You need to know that if there’s something in this universe then I know it’s in it, and there’s something outside of it, like you started to talk about Chelsea Football Club to Amy, I know it’s got nothing to do with meeting scheduling and I can ignore it.

Andrew: Got it.

Dennis: And that was almost the first year of what we worked on.

Andrew: Dennis, I went back and looked at the first version of the site. It’s interesting how little there is to say about artificial intelligence, about anything, and a lot to say about you. Your photo prominently displayed. High res shots of you, your about page. It seems like what you’re doing is saying, “Look, this is the project I want to go on. You have to understand I’m an experienced entrepreneur. You’re trusting my track record here when you’re investing in this.” Am I right?

Dennis: To the [last v 00:52:11], yes, for where not many people are really in this day and age for the cost of setting up a new venture, interested in doing the $2 million for a thumbs up. “Dennis, I think I can get a lot of MVPs for $2 million. So how about I go buy that instead?”

Andrew: Right.

Dennis: So it’s suddenly a bet on me and the team being confident enough for where if they figured it out, it could be a good bet, but you really making a bet of the team and you know that there’ll be nothing real tangible come the end of the first milestone.

Andrew: Right. As opposed to going in and saying, “Can we get a bunch of people to let our assistants do their booking for them?” And now that we’ve got a bunch of clients, our next question is, “Can we find a computerized solution for booking for them?” Why didn’t you decide to do it that way?

Dennis: So we certainly . . .

Andrew: It’s almost [like 00:53:02] for granted you could do that, the coding.

Dennis: So we were certainly never interested to build a company for where the very foundation was based on a set of human schedulers. Even if those human schedulers were powered by 80% software, we wanted something that was infinitely scalable, something for where the whole idea of setting up a meeting become so normal, like, you wouldn’t brag to me about you saving your files in the cloud. That’s just where you save your files in the year 2019. We wanted this to be as scalable and as normal and only the 100% software solution we kind of work in that. So we leaned into that with the idea that we’re going to try making it work and if not, we’ll die in that.

Andrew: Okay. You were able to see that there is a finite way that people ask for meetings, you were able to book it with people. When you switched to software, who wrote that software and what was some of the difficulty involved in getting it actually up and running?

Dennis: The software and the team I brought over my VP engineering for my last venture, one of my senior engineers brought in a lead data scientist and that was the core of that kind of initial team. And I think what we might not have fully appreciated was that step between having some override even if minimal, and then go into that new setting of no ability to any override. Just like that dude in the Waymo self-driving car, he only really do a few overrides a day. Yeah. But those are the important ones. Those are the ones where he’s not hitting a pedestrian. And it looks so minimal, that we should be safe to kind of let the AI loose.

Andrew: You’re just saying is . . . What we’re getting to is the software, you guys wrote it, it worked, it made sense. You had people check the software and override it sometimes, but it was so small that you said, “You know what?” January 16, 2017 was the day that you’re going to do what?

Dennis: Pull the safety [belt 00:55:22] out and let the machine do all of its own [space 00:55:27]. Even as we look at it, as in, if this is a wrong one, it needs to figure out how to self-correct. As in, I make a prediction for where I assume that you’re a mandatory or Samantha’s optional, or it’s next Thursday, or you’re traveling, or whatever it might be, but I needs to let the machine do the incorrect predictions and see what happens. And you know what? It blew up in our face. It looked so solid.

Andrew: What happened?

Dennis: January 16th, 0800 EST, we flipped the switch, and now the machine is on his own. And that was a day for where the product went from being, “Hey, pretty solid.” As in, “You actually haven’t heard anybody complain about those Waymo cars not being solid? Yeah, because they are, because whenever they’re not, somebody will override it.” We went from the, “Nice. I like it. Dennis, you got a bright future here. This is going to be awesome.” to, “Shit. This is probably broken. I don’t think you’ll ever make it. This doesn’t work. Massive churn. I loved you yesterday, now, I love you a whole lot less.” And we then thought, “Okay, there’s going to be a tough period now and we’ll slowly but surely figure out these edge cases.”

Andrew: Right.

Dennis: You know what? That took me and the team 18, 19 months.

Andrew: Wow.

Dennis: It was brutal.

Andrew: There was one point soon after this you went to the bathroom at WeWork with your laptop or your phone and you did what?

Dennis: So here’s what I tend to tell people. As an entrepreneur, and we have our own kind of little routine. So you got two states, the state for where you are super fucking ecstatic, high fives and the whole thing.

Andrew: Yeah.

Dennis: Or you sit alone in a WeWork bathroom in the dark crying. Those are the two states, and you have to kind of figure out, “Which state am I in?” And you know what? Some days, you’re in a state for where you sit alone in that WeWork bathroom, you might just be crying on the inside, but that was really hard. What we did do, though, as the slightly experienced entrepreneurs was to at least assume this might be harder than we imagined. So let’s make sure that we have some capital that allow for enough runway for where if this is going to be a little bit longer than expected, we won’t die along the way.

Andrew: Eighteen months. So, by the way, I signed up just a couple of weeks I think after you guys made the switch maybe four or six weeks, somewhere around there. And part of what drew me to x.ai was that there were humans behind the scenes. And that’s the way I was interacting with the software too and in retrospect I had no idea this was what was happening. And what I would do was I still used one of the calendar type software. And then with Amy, when there was something weird, that’s when I would bring Amy in. Amy is the name that you guys gave your artificial intelligence.

So I would say something like, “Amy, please schedule an appointment with Dennis for Friday.” And I liked that I could do that. The problem with those calendar booking systems, which I love. My favorite is Acuity Scheduling. The problem that I have with that is if I ever need to do something weird with someone, like, I only want to talk to you next Thursday, I can’t do that. If I only want to talk to you for 10 minutes, I can’t do it unless I fire up a new calendar.

With x.ai, I realized I could say, “I need to talk to five people on Thursday 10 minutes each.” I could introduce Amy, I think it was like amy@x.ai and say, “Amy, please book a meeting for next Thursday for 10 minutes with Steve.” And then boom, “That’s it. I’m done. There’s nothing for me to do.” And that was the dream. And then I would throw in a couple of extra things. Now does that work? Now can I do that?

Dennis: Yeah. And I think you described it very well. So there’s really two types of meetings which you described. They’re the ones that we describe as meeting types. So if you’re a salesperson who’ll do demo calls, who’ll do kind of X number of qualification calls, you’ll have a set number of meeting types, and whatever your job profession, those will be fixed. And we provide the very same kind of scheduling pages and so on and so forth. But then there’s also the ad hoc meetings for where sure for this particular meeting I could go in and create a new pace with some specific constraints that seems overly cumbersome because I really just want these two or three meetings to be within these constraints. So it’s want to describe those on the fly.

Andrew: Yes.

Dennis: And that is certainly the allure of what we’ve created and what I think is our competitive advantage. And that is something for where we have learned a ton outside of just being able to provide the accuracy that people kind of have come to expect, but also the design of it. As in, we found out a ton of things along the way for where if you and me go engineer some iPhone app, there’s a ton of best practices for how to kind of put that in place. What best practices do we have for intelligent agents? Actually not many.

Andrew: Yeah, none.

Dennis: So there’s a ton things we’ve learned, like, I’ll give you a few of them. Don’t pretend to be a human, but do a job so well that you still thank the machine come the end of it. Be very willing to ask your boss for what next step to make. Don’t assume you know best. For example, I give you a little kind of design detail for where if you deal with humans, you allow them to apply a little bit of elasticity to your preferences, say, your assistant. Yeah, you like to be home by 6:00. If somehow Dennis is in the Bay Area and the meeting could only be at 6:30 and I only come every year and a half, hey, just make it happen.

But you what? When Amy and Andrew did that, our two agents, people immediately went for, “Hey, this is broken.” We spent like eight months, multiple [PSDs 01:01:37] trying to figure out how do you apply elasticity to your common sense to an agent. They don’t think like that. So for that, we ended up with a lot of these, what we call perceived errors where the agent did exactly what he’s supposed to, exactly what you want upon explanation, but on first view, “This thing is wrong.”

Andrew: Because the agent would say, “Dennis is only here for a short period of time. Andrew should have some flexibility for that. And we’re just booked Dennis for 6:30.” And Andrew sees Dennis at 6:30, he goes, “No, I’ve got to pick up my kids every day after school.” Why are you doing that? And so how do you solve that?

Dennis: So what we’ve done now is that we have these decision points for where we will go back and say, “Hey, Dennis is only in for this one day before he flies back to Hong Kong. Do you want . . . I will schedule the meeting at 6:30. And if you don’t want me to do that, let me know.” So we have this kind of design principles where, “I’ll take the action and now I’ll let you know why I took it and the reasoning behind it,” but you have a say in it for where you can say, “You know what? I see what you’re trying to do here, but it’s not going to work.” “Okay, I’ll go back and reschedule it.”

Andrew: You know, that literally is what happened with your interview. Where are you in the world?

Dennis: New York.

Andrew: You’re in New York. So my assistant knew that you are so important to me that she . . . I don’t book interviews for 9:00 a.m. because I do want to take the kids to school, and they could just go like have a screwed up morning and I don’t want to have to rush them just so I could get here by 9:00 so I start my interviews at 10:00. My assistant, Andrea, knew you were that important to me, she said, “Screw it. I’m doing it at 9:00 a.m.” And that’s the type of thing that you’re talking about that Amy can . . . You know what else? I always when I would book with Amy, I would always have a line that says, “Amy is a chatbot or a bot but I’m really human.” I kind of in retrospect maybe you shouldn’t have used the name Amy. I think you let us use different names for the assistant, right?

Dennis: Amy and Andrew, yes.

Andrew: Amy and Andrew. I almost want to just call it “x.ai, please do this for me.” without using the person’s name so that I don’t give the impression that this is a human but let them know this is an automated system.

Dennis: So we also in the early days, and I’m sure you can imagine how the geek inside of us took some sort of pride in running 1,000 daily touring tests and winning some of them for where there was people in this kind of network that never figured out that this was machine. And we just kind of run around the office thinking, “Hey, this is pretty awesome. But you know what? You haven’t won anything.”

That’s just an unfair amount of risk in you fooling a person or them now imagining that this is a human and do things where a machine will obviously not be able to do what you’re just trying to make it do. So we’ve been way more kind of visible in our design and the way we communicate in this being a machine. That doesn’t mean that you remove the ability for the flexibility that we just talked about. It just means one for where you shouldn’t hide the fact that this is a machine.

Andrew: Got it. Okay. And you were starting to say a couple of other things that you learned as you did this.

Dennis: Some of the other things that we learned with . . . Again, many of these, in hindsight, are just so obvious where you’ll say, “Dennis, why didn’t you figure this out on day one?” So in any setting where you got two parties who sometimes are in conflict. Say on eBay, you got sellers and buyers and eBay make all their money on the sellers. As in, that’s the vast majority of their revenue. But you know what? Who should they cater to? Many times you immediately think, “I should cater to the very people who pay me money.” But for eBay, they figured out, “You know what? I need just to cater to all the buyers, because you know what? If I bring buyers, the sellers don’t give a shit. If they make money, they’re happy.”

Andrew: Right.

Dennis: We started out catering too much to the host, our customer, to you, Andrew. And you know what? We just spend all our time just making your guests happy even if I have to kind of ask Andrew just one additional question. And that took us actually a little bit of time to flip around to figure out is all about guest delight. If your guests are happy, you’re happy.

Andrew: Right. I wouldn’t mind being bothered, but if a guest told me, “Hey, Andrew, your automated system is kind of janky or it’s awkward,” I would feel guilty and I feel like bad for introducing them. Got it. Yeah.

Dennis: So we completely flipped that around. But again, now when I say it out loud, it seems like, “Dennis, you should have known on day one.” But I think many times you end up just falling in love with your customer because that’s what you always do and try to do good by them, but do good by them means be kind to their guests.

Andrew: All right. One last one. I love this stuff. This is the type of stuff that I could read for hours. Is there one other one that sticks out for you that you’ve learned about this?

Dennis: Really what’s happening right now is actually two things at the same time. And I’m not sure everybody immediately sees that. So we seeing the introduction of a new UI and . . . I’m 46, so I’ve seen myself take a CS degree on the command line as the primary UI. I’ve seen my mom being introduced to Word over graphical user interface of Windows and what have you. I’ve seen my kids grow up on a touch interface on their mobile devices. But right now this very moment, the conversational UI is about to be introduced.

Andrew: Yes.

Dennis: That is just a UI. There’s no intelligence baked into it. It’s just a new way of interacting with compute. Asking Alexa what the time is, Citibank texting me whether I did buy that bagel, yes or no. Just a good new UI. And what happens when a new UI arise because it happens so rarely, as in, we’ve seen three or four over the last three or four decades, then you fall in love with it, or I suddenly fell in love with it as a technologist. But many times what happens is that certain things will carry over to the new UI, other things will stay put. So you’re not editing your photos on your mobile phone. You’re using a big ass screen and Photoshop and 27 inches of screen real estate and that should just stay put. So for us, we moved everything into the conversational UI. And you know what? That was sexy, but not everything belongs there. And I’ll take the most obvious.

Andrew: Yeah, go ahead. Go. That’s what I was going to ask for.

Dennis: So we imagined . . . So if you hired a human assistance, the first thing you will do is you’ll sit her down, have a Diet Coke, give her the lay of the land. “Hey, I got two kids. I like meeting Suzanne from 10:00. I like to leave a little bit early on Friday. Don’t pack and give me a little bit of a breather time. I like people not to call me. Can you get their number, then I’ll call them?” No. It’s kind of just do the verbal kind of introduction. We did the same you know first inclination. You just email Amy, tell her what you like and we’ll pick up on that. People don’t want that. What they want, “Just give me a set of sliders. I will pick when I want my meetings.”

Andrew: Yeah.

Dennis: You know what? Because that is not a bad UI for that particular question. And then we figured out also that there’s actually the opportunity to introduce a button when needed directly in the dialogue. It doesn’t have to be all text. You don’t have to say, “Yes, next Wednesday is okay.” You can just click Yes, that’s okay. That’s not illegal, Dennis. But for me, it felt like cheating.

Andrew: Right.

Dennis: Yeah.

Andrew: You’re saying when you email one of my customers to say, “Can you make it on Monday at 10:00 a.m.?” you don’t need them to hit Reply. You could just have a yes or no or whatever. Right. And it does feel like cheating. You know what I’ve got to say, Dennis, about you? I feel like you guys captured the world’s imagination. People in the tech community were talking about this a lot. Then you disappeared.

I would love to see you talk more at conferences about this because conversational UI now has expanded beyond where it was when you were starting out, right? To text messages, to Alexa, to etc. Your insights are so helpful for other people that I feel like you’ve got a chance to lead the way, to lead other developers in this and remind them, by the way, “We exist as a way of scheduling.” The people would show up at conferences to learn how to program their artificial intelligence, bots, or whatever, they would use your software, number one.

I’d love . . . I know you said that content is great for you, and I agree. The content is better than Facebook for you or bunch of ads. I’d love to see even more. I’d love to see somebody sit down with a pad of paper next to you all day and just as you’re talking through what you’ve uncovered write it down and turn it into a blog post for you. This is cutting edge stuff that people who talk about it are talking about it from this hypothetical . . . they’re talking about it the way that you might have when you first started. They think about it is all happening via text and forget about buttons. They think about it as this intelligence software, instead of saying, “It’s not yet fully there.”

Dennis: It’s very easy.

Andrew: What do you think about that?

Dennis: We should. And I feel guilty even just by have you kind of read it back to me, but you invite us and we certainly . . . I think anybody should be a fan of this idea of falling in love with the pain, not the solution. And we certainly in the early days fell a little bit in love with the idea of AI, but have come around full circle to really just lean into the pain, as in . . .

Andrew: Go ahead. What is the pain today then? If you would have fallen in love with the pain versus the solution, the solution was artificial intelligence, falling in love with the pain would have meant what?

Dennis: Meeting scheduling. I’m just a guy who scheduled meetings.

Andrew: Got it.

Dennis: If you pull out your phone right now and look at those 17 meetings for next week and I didn’t put them there, I got work to do.

Andrew: And then you would have been happier to put, which I didn’t know that you had, a calendar on a webpage because sometimes that does work.

Dennis: Yeah.

Andrew: Got it. Got it.

Dennis: For some meetings, really, I just want to expose my availability. And it might be some context for where I know this is by far the best way to make something happen between you and I know. And then there’s other ways in other meetings where, you know what, “For this one, I actually want Amy to kind of nudge them a little bit because I want this on the calendar.” And we need to just make sure that we provide both options and that they can interact because it might have started on a web page, but then somehow one of your kids got sick and you need to reschedule it and you just want to quickly email in and say, “Hey, Amy, can you move this to next week,” and you don’t really want to send links to stuff like that, just move it to next week.

Andrew: I’m trying to think, did we ask you to do an interview or did you ask us? I think it was us asking you.

Dennis: Yeah.

Andrew: I would like to see you going out there and talking more about this stuff. I would love to see . . . Like I said, just get somebody with a notepad to just sit and write down what you’re uncovering and take some of this. Freaking brilliant. All right, Dennis, I’m back on the software. So one of the reason I left was there was these issues and I said, “Okay. I can’t use it. It’s a problem for my customers.” I’m now in the process of going through it. I now stopped at a place where there’s a video because I don’t want to interrupt my conversation with you to watch this video. Should I just go to the next page and come back to the video later?

Dennis: Yeah.

Andrew: Yeah, I’m going to hit the Watch Later, that’s what I’m going to do. It’s a YouTube video so I can hit Watch Later on it to go through. Let me ask you this. If somebody wants to try this right now, I never asked this to some of our guest, is there a URL or something? Do you want to create a URL so we can see if people are using this and what they think of it?

Dennis: You should just go straight to x.ai. And I think there’s two reasons. There’s obviously the very kind of commercial reason for why I want to sell the next customer, but forget about that for a second. If people are still kind of not on the fence, but a little bit unsure about everybody says AI. I see it in every single kind of publication I get, but if you want to get some tangible kind of experience with . . . But what does it translate into? Just go sign up, there’s a free trial, try to set up a set of meetings and get an opinion about, “Ah, so this is where we’re at in Jan 2019.” And if I do well, you might fall in love. Pay me $8 and we’ll be friends for life. But that’s actually not the point. The point is, if you want to be able to be a participant in the conversation, I think this is a good product to kind of get some experience with.

Andrew: You’re saying, look, even if you don’t think that scheduling is for you, go see what technology is like today in the world of artificial intelligence and for free, you get to try this book calendar appointments with yourself, with your sister, with your brother, see how that works. If you fall in love with it and you decide to pay eight bucks a month, great, go for it. If not, you’ve learned a little bit about artificial intelligence. I think that makes sense. x.ai is a great domain. I’m signed up now. I want to try this. I like the idea.

I can say this, Dennis. I can say, “Schedule a meeting with Dennis for tomorrow.” I can actually go outside of my thing. So if usually at 6:00, I’ve got to be out of the office, I can say, “Schedule a meeting with Dennis for an hour after 6:00 p.m.” and the software will know to do that. One more thing. This interview was supposed to end at 10:00. It ended at 10:30. I usually keep a buffer for interviews. Can I say that to x.ai to say, “For interviews, add a buffer of 30 minutes,” if it’s on this interview calendar?

Dennis: Yeah. And you can . . . So all of those kinds of little details that you and me don’t really think about when we just talk about it as human beings, Amy and Andrew are able to kind of pick up on.

Andrew: Amy and Andrew are the names of your artificial intelligence.

Dennis: Yeah. And can obviously move those to your domain so that it can be amy@netflix.com.

Andrew: Or amy@mixergy.com.

Dennis: Yeah.

Andrew: Right. Okay. All right. I’m back into trying it. It’s also cheaper. It was so much more expensive when I tried it before. And I’m happy. It was. You lowered the price?

Dennis: We started out . . . And this is actually . . . This is another kind of funny insights if you want it for where we had a dramatic elasticity in where people wanted the price to be. So plenty of products, most people will have some sort of idea of what it should cost. If I asked you what should be the cost of a laptop, you would give me some sort of range between cheap, 600, kind of expensive, 1,300, or some version of that.

Andrew: Right.

Dennis: But you and me kind of know what it’s supposed to cost. When we asked people, and we asked about 5,000 people before we set the first price, people started out with, “You know what? I bought an app in the app store yesterday. So anything above $1 is crazy talk.” Then we had people, “I used to have a human. If you sell for this for $500, $600 a month, it is dirt cheap.”

Andrew: Yeah.

Dennis: That was dramatic. So what we did is we ended up at 39. But you know what happened? We would sign customers like Nike or Stanford or Electronic Arts and really to see, “Hey, lovely.” But they turned it into a luxury. So in their mind at 39, “Oh, I’m going to give this to the people who deserve it.” It’s not like . . . It’s kind of like, you guys probably have Slack around the office. You don’t give that to people who deserve it. You should give it to everybody willy-nilly, right? Of course. And somehow at 39, people saw this as a mini-luxury. When we change the price down to eight, it became software. “Oh, eight? Anybody wants this? Take it.” And that really just changed everything.

Andrew: Yeah. I frankly for me, the way I think about it is I compare it to my assistant’s time and if it saves time, then it’s perfect. But you’re right. When I start to think about, “Do I give it to everyone else?” Well, I’ve got someone who just books four meetings a month, literally four meetings a month. Do we pay for him? I might say, “Yes,” Just to get it off my plate. Would Marisela, our COO, say yes? She might say, “Let’s save Andrew’s money.” Well, that’s because she’ll say yes. Screw it.

All right, I get it. I freaking love this interview. All right. For anyone who wants to go check it out, I just signed up x.ai, I get no money for this. I would be insulted if I got any money for talking about this. I’m not even necessarily recommending it. I’m saying, I’m going to go back in and try this again because scheduling still is a pain in my ass. My butt, excuse me.

I want to thank the two sponsors who made this interview happen. The first will host your website right, it’s called HostGator. Check them out at hostgator.com/mixergy. The second will help you hire great developers. Like if you decide, “I want to have artificial intelligence on my team. I want to have blockchain on my team. I want to have a consultant to help me think through this technology or somebody to code it up for me,” all you have to do is go to . . . Oh, wait. To toptal.com/mixergy, toptal.com/mixergy.

And finally, I love chatbots. I love bots. I still think that for most purposes, competing with x.ai and trying to have artificial intelligence, we are not there for most places. I created a chatbot that you guys can see for yourselves. No artificial intelligence, just guiding you through an experience that happens via chat instead of through email. If you want to check that out, go to botacademy.com, botacademy.com.

Dennis, I could’ve talked to you for a lot longer. I know that I took up more of your time than we agreed to, but screw it. It was freaking worth it and I’m glad that you were willing to sit and stay here. Thank you.

Dennis: This was fun. Time well spent.

Andrew: Same here. Thank you, Dennis. Thank you all. Bye, everyone.

Who should we feature on Mixergy? Let us know who you think would make a great interviewee.

x