Future-Proof Cloud Communications with Andrey Kovalenko, Co-Founder & CTO of Voximplant

(Intro Narrator at 00:00:03) Hello, my friends. Today, Joel is talking to Andrey, co-founder and CTO at Voximplant, and they discuss how Voximplant's communication platform as a service enables developers to easily build complex communication solutions, how Voximplant has been serverless since before serverless was a thing, and why individual contributors should think twice about whether or not they want to move into management. All of this right here, right now on the Modern CTO Podcast.

(Joel Beasley at 00:00:40) Here we go. This is the Modern CTO Podcast. Can you just start out with a brief overview from when you got started into technology to you starting Voximplant, your whole story?

(Andrey at 00:01:04) So when I was a small boy, I got my first PC at age 10, I think. And when I started at school, I found that programming is quite an interesting thing. And I usually wrote some small programs that my sister was required to write, but she didn't want to. After that, I went to the university. My first job was in second grade in university, and since then I'm in software development, doing all this stuff. So we started creating things in web telephony in 2007. The first product created was called Flash Phone. It was the first web soft phone ever to exist. We didn't have any idea how to make money off it, so we just gave our users an option to make some free calls to anywhere they wanted and didn't charge them at all because we didn't know how to do it. But spending that money gave us some interest from media and from enterprises who wanted to license our technology. So then we created a software business.

(Andrey at 00:02:20) We sold some software licenses, but it wasn't very good, to be honest, because we didn't want to, and we didn't have any idea how to sell it abroad. And because we are from Moscow, we just worked on inbound requests. We didn't make any active sales. And then we created a Zingaya service, which is a click-to-call solution from websites. It is still live. You can check it, but we recommend everybody to migrate to Voximplant because it's a newer solution. It's on more modern technologies than Zingaya. So that's the story short. And in 2013, we created Voximplant, a cloud communications platform which uses WebRTC. Well, it used Flash at the time. There was no WebRTC when we started. It appeared in 2011, I don't remember exactly. But when we started, there was no WebRTC. We used Flash for voice calls. And, well, it's better than nothing, but as a technology for real-time communications, it's not so good, to be honest. It was not so good because now, you know, Flash is off. There's no Flash at all.

(Joel Beasley at 00:03:23) I think it's good that there's no more Flash.

(Andrey at 00:03:27) Yeah, because it has some issues. It has some flaws and a lot of them, to be honest.

(Joel Beasley at 00:03:36) I actually got to interview one of the people from Adobe who, when they were making the transition from disk to software as a solution, like web platforms, it cost them like a billion dollars. It was a huge bet that they were making that, you know, the cloud was the right way to go. But ultimately, it was the right way to go, and it ended up working out pretty well for them.

(Andrey at 00:03:59) Yeah. Subscription-based services are usually better than one-time sales.

(Joel Beasley at 00:04:05) Yeah. So why did you even, the first telephony project you did, why did you do it? You could've just picked up a phone and called somebody.

(Andrey at 00:04:14) Well, because at first, by the time, I worked, and all of us, all the founders of Voximplant, we worked at a company that developed Voice over IP software, so we were quite familiar with it. And by that time, there was no such solution because we were students, basically. It was even before we graduated from the university. It was interesting for us. We just thought, why can't we do it? So we did. We never thought that it would evolve into such a project like Voximplant. It has 200 plus employees and all this stuff. It was just basically a great idea to start with, but we didn't have any idea how to make money from it. We just started as developers. We bought some server that was staying at home in front of us. We didn't even use a data center. So it really was a bad project that evolved into a great business.

(Joel Beasley at 00:05:10) So you three were working at a company together that was in this space, and then you decided to do this other project together?

(Andrey at 00:05:18) Yeah, exactly.

(Joel Beasley at 00:05:19) Tell me about the other two founders. Who are they? What do they do?

(Andrey at 00:05:22) One of them, Alexey, is our CEO right now, so he's responsible for all the company vision and general stuff. By that time, he was a developer as well. The third guy, he's more about sales and business development. Well, he started with us in the technical university, but he didn't really get much of it. So he's more about marketing, sales, and all this stuff.

(Joel Beasley at 00:05:48) So that's exciting. So they handled the sales side of things. Was he working sales at the other company too? Is that their background?

(Andrey at 00:05:56) He was working at the marketing department.

(Joel Beasley at 00:05:58) Yeah. So that's cool that you were able to combine your different skill sets in order to start a business.

(Andrey at 00:06:03) Yeah, that was great.

(Joel Beasley at 00:06:05) And now you've grown and expanded over the past 10 years. Now you're this big company. Man, that's gotta be a lot of stress growing that fast. Tell me about some of the things you learned along the way.

(Andrey at 00:06:17) It's quite easier to be a small company than a big one, right? So when you have 10 people or even 50 people, it's quite easy to manage. And at least you know what they do. Because when you have 200 plus people, you basically can't know them all by name. Because there are new hires going on constantly and some people leave. When there was our party recently, I met people and I couldn't tell the names of, like, half of them. So it's quite stressful to rely on people that you don't even know, compared to when you start business and, you know, there are three of you and you know each other for a lot of years. And it's quite more stressful to rely on people that you don't even know who they are, but you have to.

(Joel Beasley at 00:07:09) So you rely on systems and processes and your culture?

(Andrey at 00:07:13) Yeah. I rely on tech leads who I know pretty well, and I hope that they know what they do and they know who they hire and how they handle all these people.

(Joel Beasley at 00:07:26) What are your responsibilities like today? Are you writing code? Are you managing the technology organization?

(Andrey at 00:07:32) It's mostly management, to be honest, and all the architectural stuff. Because tech leads are good in their own field, but when it comes to overall architecture, I'm still in charge, and it's good for me because if I did 100% management, I wouldn't be happy at all. So I don't write code too much these days because I don't have time to do it properly and to take some major tasks, because I wouldn't be able to do them in time because I have many other things to do. But there are some small projects inside the company that I still maintain. One of them is our Python SDK that I maintain myself and some other internal tools, just to still be a techie. Because you can't be a CTO without writing any code at all. It's impossible.

(Joel Beasley at 00:08:22) Oh, yeah. So my background is I wrote code for 17 years pretty much every day, and then when the podcast started to get popular, I noticed that there were weeks I would go without writing code and I missed it a lot for about the first year. And then after the first year, I just enjoy so much understanding people and how they work and planning. And for me, my love of programming was because it was a way I could create. So I was creative and programming allowed me to create. But then I found that I could create with people and teams. And for some reason that just, to me, was more exciting. And so I feel bad for saying this, but I don't really miss writing code every day. I really, really enjoy getting to talk to great people like you and getting to grow a business and helping people, the next generation of leaders, helping them grow, sharing the mistakes that I've made, and watching people do well. You know?

(Andrey at 00:09:27) Well, it's personal, so it may be great for you, but I personally prefer to write at least some code because I like it.

(Joel Beasley at 00:09:36) Yeah. So I get to talk to a lot of CTOs, and, you know, we make lots of notes and things like that. And there's a certain type of CTO that really likes to code even later. And one of the things that they do, and I'm just sharing this with you, is they will make something called the Office of the CTO. And it'll be a small team of people that'll work on really cutting-edge projects that will probably fail, but if they succeed, they would really help the company, or things that the company needs. And it's just this small team that works directly with the CTO. And then the CTO will hire a manager, obviously somebody who knows how to code, somebody who knows how to structure code, knows how to at least be a great architect. And they'll hire that person to manage the technology side of the organization, the people, and the process-type things. And then they get to spend their time building really cool, unique, interesting things, and that's what they do.

(Andrey at 00:10:36) Well, to my point of view, it seems a bit egoistic, right? Just offloading all the stuff you don't like to other people and giving yourself tasks that you like. It may be good if the company is bigger than ours, I think. But for our size, I think it's a bit of overkill hiring another person to manage all the tech leads, all the development stuff. I think it's a bit early, to be honest.

(Joel Beasley at 00:11:05) Oh, yes. So it's definitely whatever you feel because you have all the variables. But I was just sharing it from a perspective of it's something some people do, and I didn't even know that was an option. And I've met people who were very unhappy as their company grew and they really didn't like management. And then for them to find out that that's an option and then do it makes them happy again. They're like, I didn't even think that that was an option. Are you allowed to do that? Yeah. You're allowed to do whatever you want, man. That's why you started a company, you know?

(Andrey at 00:11:40) Yeah. I wouldn't say that I'm unhappy, because I managed to find some time to write code. It's not for me, so I keep some balance.

(Joel Beasley at 00:11:48) Yes. Yeah, and it's important. So I'm curious, for people that don't understand, can you explain what serverless is?

(Andrey at 00:11:57) Serverless is a technology that allows a developer, you for example, to build some solution without managing any infrastructure at all. So Voximplant is a serverless communications platform, and we provide tools to create some call flows, some call scenarios, or messaging applications that are being executed in our cloud. You don't need to maintain any software, Asterisk, PBX, or any other open-source tool. You just write code directly in our IDE, in our cloud, and it's executed in our cloud. So you don't need to worry about scaling, about security, about connectivity to the public telephone network. We handle all that stuff. You just need to write code and think about what actually is important for your business, not how the telephony works.

(Joel Beasley at 00:12:57) Is it, do you have to be a, it's very code-based. You have to be a software developer to use it?

(Andrey at 00:13:02) Currently, yes. Currently, Voximplant requires JavaScript skills to create a call scenario because we have a JavaScript engine in our platform. Unfortunately, there are no other languages right now, but we're thinking about maybe adding other language options or maybe creating some low-code or no-code solution on top of Voximplant to make it easier for non-developers or for developers who don't like JavaScript because there are many types of developers who basically don't use it, like mobile devs or some people who hate JavaScript. There are so many of them.

(Joel Beasley at 00:13:42) Do you support anything like CoffeeScript?

(Andrey at 00:13:44) Currently, no. But we are planning to create tools that will allow developers to integrate Voximplant with their repositories, CI/CD functionality. And so they can add transpiling from CoffeeScript or from TypeScript into JavaScript.

(Joel Beasley at 00:14:01) Oh, that's cool.

(Andrey at 00:14:03) Yeah. But it should be available in the near future. I think in Q1.

(Joel Beasley at 00:14:07) So wait, let me get this right. By creating the ability for Voximplant to be part of their CI deployment, that's the ability you're creating. And then they could just write it in CoffeeScript, it would compile down, and then go into that.

(Andrey at 00:14:22) Yeah. I think there are compilers for very many languages these days to JavaScript. I think you can even compile C++ to JavaScript if you'd like to. But we don't recommend it. But you can.

(Joel Beasley at 00:14:36) Yes. You can do a lot of strange things with code that we don't recommend. I've seen people write code in the most strange ways. One of the most interesting ones was when there was some testing framework. I think it was Cucumber that came out, and they would dictate all of the rules in these large blocks of text. And to me, it was just so foreign the first time I saw it. And typically, my background recently is Ruby. So RSpec, using test systems in Ruby. And the way that you get to assert things and test things there makes a lot of sense to me. But writing it out with Cucumber, like plain English, writing the whole thing out, I was like, that is a whole other level.

(Andrey at 00:15:20) Yeah. And it's actually funny that when I interviewed QAs, many of them told me that they don't want to work with Cucumber at all. They want everything but Cucumber.

(Joel Beasley at 00:15:36) Yep, yep. Because it confused me when I saw it, and then I saw there was a whole book about it and there were people. But for me, it didn't click. Some other type of testing clicked. I would read Sandi Metz, I think that's her name. She was an author that would write a lot about writing tests in Ruby. And she would reference bikes a lot. She was really into biking, riding bicycles. And so a lot of her examples included that. And it was really easy. She explained everything super simply because there was a time when I didn't test, and then there was a time when I did, and it was a difficult transition. When you got started in software, did you start testing day one, line one, or did you write code for a while that you didn't test and then learn to write tested code?

(Andrey at 00:16:20) Well, of course we did some tests. But the first QA we hired was in 2015, maybe. So a few years after we started Voximplant, we hired our first proper one.

(Joel Beasley at 00:16:36) But you as a developer, when you first started writing code just in your personal life when you were younger, did you learn to write test-driven development from the beginning, or did you write other code and then figure it out?

(Andrey at 00:16:50) When I started developing, I was just writing code. I was a student. You shouldn't expect too much from a student.

(Joel Beasley at 00:17:03) Yeah. I mean, I learned to write tests when my project started to grow and people really depended on them to run their business. And then, you know, every time a bug would come in, we had learned to write a test for it. And then that's how we started building. And then we went around and you can get crazy with tests.

(Joel Beasley at 00:17:20) You can write too many tests, you can write fragile tests, you can write all sorts of things. So learning how to write tests the correct way, or the correct way for your project or use, is an art almost.

(Andre at 00:17:35) It's important to select proper type of tests, right? Because, for example, we started to write very many tests for one of our APIs, and then we found out that it's basically more time efficient to test them manually because writing an automated test for some complex solution may take weeks, while the manual test takes an hour. And even if you perform the test once a week, it's still more time efficient.

(Joel Beasley at 00:18:10) Yeah, we could talk about tests forever. A long time ago, I got to interview one of the people that was working on, like, some of the first—I think he was working on stock market software for the Netherlands, for Amsterdam over there. And they were working on some of the first testing, like the concepts of testing. They were working on the more modern ones when they moved from mainframes to more modern stacks.

(Joel Beasley at 00:18:42) And he was a much older guy. But it's so cool to be able to talk to somebody who's 60, 65, 70, and them talk about testing. I mean, I don't connect talking with my grandparents about software development.

(Andre at 00:18:59) Yeah. By the way, for stock market software, testing is quite important, right?

(Joel Beasley at 00:19:03) Yeah.

(Andre at 00:19:04) And not only should it work, it should work really quickly, really, really quickly.

(Joel Beasley at 00:19:11) Oh yeah. All right, so I want to talk a little bit about some leadership stuff. Is that okay?

(Joel Beasley at 00:19:17) Cool. So we get questions all the time from up and coming engineers that are moving into leadership, trying to transition and take on more responsibility, about how to do it, mistakes to avoid, things like that. But let's say that there's an engineer listening and they're writing code, they're doing some hiring, but they want to take that next step in leadership. What insight would you give them?

(Andre at 00:19:44) Well, they should think twice if they really want to. Because if you become a team leader or become a CTO, you will spend way less time writing code and doing all the engineering stuff. And if you really liked that and you will miss it, it may be an issue for you. And it's also important that leading a team and writing code is a completely different set of skills, right?

(Andre at 00:20:07) So if you feel that you're not very good in communications and all these soft skills and small talk, you need to train them because otherwise you would have an issue. You can't become a good leader just being a good engineer. You may be, but you need proper people that you work with because they should be quite self-assured, and they should know what they do. And they should respect you for your skills, not for your leadership skills. That's quite important to find the right people.

(Joel Beasley at 00:20:40) You just kind of get experience. You just roll with it. You go with the—I mean, your company is growing fast. You're having to do difficult things. You have these two business partners that you work with. They're doing difficult things. You're kind of all sharing and exploring together. So, you know, probably those two people had a large impact on your leadership style and your development there.

(Andre at 00:21:01) Yeah, probably. But I think we three are quite different people with different characters and different managing techniques, right? So I wouldn't say that we take any good ideas from each other. We try to do everything as we think it's proper to do. And I think I can say that we succeeded, right? Because we have 200-plus people, 215, as far as I remember. So if we were not that good, I think we would have failed by this stage.

(Joel Beasley at 00:21:36) Yeah. Well, I'm glad you didn't because it's fun to get to talk with you. So tell me a little bit about how you hire engineers. Like, what's your process for hiring engineers?

(Andre at 00:21:47) The first thing to hire a good engineer is to hire a good HR person, right? It's the first step. Because I remember, like, five or six years ago, it was enough to put a job description to a job board, and that's it. You get a lot of responses, and you can easily hire an engineer, and a good one.

(Andre at 00:22:11) But these days, you need to, like, pray. Or at first, you need to hire a good HR. They need to pray. And there is a lot of jobs to be done for HR. They tell me that they message a hundred people and get one or two responses these days. So it's immense.

(Joel Beasley at 00:22:31) Yeah. And so if people are listening that are in the telephony industry or what's the acronym? It's CPaaS. What does that stand for?

(Andre at 00:22:39) Communications Platform as a Service.

(Joel Beasley at 00:22:41) So you're hiring right now if people are engineers in that space and they're interested in working with a fast-growing company?

(Andre at 00:22:47) Yeah, we do. But an issue is that we have a development office in Moscow, Russia. So—

(Joel Beasley at 00:22:55) We have listeners there. We have listeners there.

(Andre at 00:22:58) Okay.

(Joel Beasley at 00:23:05) I know six words. I know how to say excellent, floor, table, side table, and then a bunch of really bad words.

(Andre at 00:23:16) Well, you should know them if you plan to visit Russia because you need to understand when someone is trying to offend you, right?

(Joel Beasley at 00:23:25) So in our last episode, when we were talking, you were telling me about AI being trained on different languages. What are you guys doing with AI?

(Andre at 00:23:35) We use AI currently for answering machine detection. So when you place an automated call, it's quite important to understand whether a person picked up the call or was it a voicemail or IVR. So we use a technology for that. Currently, we support mostly Russian language, but we recently hired a specialized team for that. So we're working on Spanish, English, and other languages these days.

(Joel Beasley at 00:24:08) Well, that's cool. How much data do you need to train a model to figure out if a person picked up or an answering machine picked up?

(Andre at 00:24:17) Well, we need about a few thousand, maybe 10,000, very short samples, right, of the first three or four seconds of each call. So it's not a very big amount of audio files, right? But you need a lot of samples, and they should be different. So you need to find thousands of people, thousands of different voicemails of each type because every carrier has its own template.

(Andre at 00:24:48) And actually, these days, it's even more complicated because—I don't remember, was it last year or the year before?—some carriers started introducing AI-powered answering bots for customers. So if you have a mobile phone from one MVNO carrier in Russia, they offer you this bot. They call it Oleg because it's the name of the founder of the company. And this bot picks up the phone and tries to communicate with a caller and check whether it's a person or some phone spam or stories and all the stuff that you don't like.

(Andre at 00:25:23) Because I don't know about the States, but in Russia, you can have up to 10 or 15 calls each day with some advertisements and other stuff that you really don't need. So it makes sense, but we try to make a solution that can sort of detect this type of voicemail. And, well, we don't aim to overcome it and to make it seem that we are a person, not a bot. We just want to detect it properly so our customers know whether their call was answered by a person or by a machine.

(Joel Beasley at 00:25:57) A few years ago—it was either Google, I think it was Google or Apple, probably Google—they did this experiment, and they showed a bot calling and booking. I think it was an appointment of some sort. And it was hilarious because it sounded like a human. It interacted. I think they made a restaurant reservation or booked some type of appointment. Have you been following that?

(Andre at 00:26:20) I don't think so. I'm not sure whether this project succeeded or not. I know for sure that it was available only in English when they launched it. I'm not sure about it these days. This voicemail bot actually works.

(Joel Beasley at 00:26:32) Yeah, they work. So I think it's gonna be amazing in the future. I'm gonna say, you know, "Hey, Alexa, do this or do that." And it will be able to interface with a human versus just like if it had an API to actually send me a pizza, you know, or an API to actually order me a book. For it to be able to interact like, "Hey, Alexa, call my grandmother and tell her this," you know, that'd be interesting.

(Andre at 00:26:55) Yeah. But you're unlikely to trick your grandmother. She'll know that it's not you.

(Joel Beasley at 00:27:03) That's right. The next step after they get it to where it sounds like a human and can correctly call and do these things is for you to then put your voice into it. And so now it can be you, and then we don't even need to exist as humans anymore. Machines can just run everything.

(Andre at 00:27:19) Actually, it's possible these days. At least, Yandex, a Russian tech company, created a "surround voice" solution in their text-to-speech engine. So you need a few hours, or a few dozen hours, I don't remember exactly, of your speech. And they will train their text-to-speech engine for your voice, and you will be able to use it to say anything. So now you—

(Andre at 00:27:44) You actually can create a voicebot that will pretend to be you and call your grandmother and discuss her daily things.

(Joel Beasley at 00:27:54) I remember in the early 2000s, those sound boards that you could go on the websites, you could click all the big buttons and make them say a different thing for prank calls.

(Andre at 00:28:03) Yeah. Things are different, right?

(Joel Beasley at 00:28:06) Things are very different. You could do so many illegal things. And, like, that's what kids did. Like, "Oh, I just hacked NASA's satellite." You know, there was a kid in my state who had done that.

(Joel Beasley at 00:28:18) And people really didn't know how to deal with it because computers—they really didn't understand them. But now they crack down hard. Like, somebody had done something where they imitated, hacked Elon Musk's account and said, "Donate crypto," or whatever. And I haven't followed up on the case, but I know that they were in a lot of trouble and likely going to jail.

(Joel Beasley at 00:28:42) And they were a kid. I was like, you know, if you're not mentally mature enough, like, we should be more lenient with the kids. You know?

(Andre at 00:28:51) Well, if you need to do something that costs somebody thousands of dollars, or even more, it should be somehow handled.

(Joel Beasley at 00:29:01) Oh yeah. You should definitely do something, but you shouldn't treat them the same as a 30-year-old adult because—I mean, you're over 30, right? And so there's a big, big mental change from 17 to 25, 26, 27. And then once you have a kid, there's another big change, because it goes from all your hobbies and your stuff to, like, how do you make these little humans grow up carefully?

(Joel Beasley at 00:29:30) And yeah. You have a little human, right?

(Andre at 00:29:33) I have two little humans.

(Joel Beasley at 00:29:35) Yeah. Boys, girls?

(Andre at 00:29:38) One boy and one girl. A boy of age seven and a girl of age five. I have different types of little humans.

(Joel Beasley at 00:29:44) What do they think of you?

(Andre at 00:29:48) Well, they are not allowed to think bad of me, right?

(Joel Beasley at 00:29:53) I love it.

(Andre at 00:29:54) And if they do, it's their issue, right?

(Joel Beasley at 00:29:57) Yeah. Do they understand that you're an entrepreneur? Do they—are you teaching them about money and things like that?

(Andre at 00:30:04) Like, yesterday or two days ago, I told my daughter that if she wants to steal some chocolate and her brother sees it, she should share with him so he wouldn't tell us that she ate it.

(Joel Beasley at 00:30:18) I love it. I love it. You're teaching her how to adapt in the real world.

(Andre at 00:30:22) Yeah. Exactly. Because, well, I can basically say, "Don't take it," right? But they're not that young, so I should give some real-world advice to them.

(Joel Beasley at 00:30:33) Yeah. There you go. This is great. Let's say people are listening and they're in the CPaaS world, right? They're in that world, or maybe they're thinking about getting into it. You know, why do people buy Voximplant? Why do they use it? Is it to build call centers? Is it to build automated one-off scripts? Like, how do your customers use it?

(Andre at 00:30:55) Well, there are different solutions, but there are some of them that are most popular, I think. The first one is automated calling campaigns. We don't recommend to use it for any advertisements, right? Make it clear. We obviously can't check all of them, but we think most of them understand us correctly and do only legitimate stuff. One of the recent projects—I'm really proud of it—it was for the Ministry of Health Care of some province in Spain, and it's an ongoing project, so it wasn't one-time. It was automated calling for COVID vaccination. So they really saved time for a lot of employees who could do other stuff rather than calling everyone and saying, "Yeah, you are invited for vaccination next Thursday." So, automatic call campaigns. The second solution that is quite popular is Cloud PBX. So you can build a Cloud PBX on top of Voximplant and embed it into your own, for example, CRM system. There are a few CRM vendors that have chosen Voximplant as a solution to embed telephony into their products. We are quite happy with them because they have a lot of small customers and they're quite consistent. It's great. Another popular solution these days is Dialogflow integration. Are you aware of Dialogflow?

(Joel Beasley at 00:32:17) No.

(Andre at 00:32:17) Dialogflow is a solution made by Google, which is basically a voicebot, right? So it's the one that we described before, but the other way. So it's designed to be called. So you make a call to a Dialogflow bot and it communicates with you in some way using some scenarios that were designed by a developer and performs some tasks or provides information. So it's basically a bot.

(Andre at 00:32:53) And so we found that of the customers that signed up for Voximplant in some months—I don't remember which one, like May—when we checked the stats, 10% of people used Dialogflow integration. So AI solutions are really trending these days, and that's why we are thinking of integrating them deeper into Voximplant.

(Joel Beasley at 00:33:14) Seeing the advancement of AI happen so rapidly over our lifetimes, what do you think of, like, Elon Musk and the neural implant?

(Andre at 00:33:22) I have read recently some news about a person making a tweet using a neural implant.

(Joel Beasley at 00:33:22) Oh, that's cool.

(Andre at 00:33:22) Yeah. So I'm not sure how this works, right? But I think that it's a great solution because there are many people who have no ability to communicate using conventional ways like speech, writing, and all this stuff, and it's really important. And if it makes writing faster, I would be for it. Because sometimes, I think that it takes quite a lot of time to write your ideas properly. And if it goes right from your brain to the laptop, it would be great.

(Joel Beasley at 00:34:05) That would be great. And then you just think, "Brain, laptop," and then money goes into your bank account, and you're good.

(Andre at 00:34:12) Yeah. Or from your bank account, and you're also good.

(Joel Beasley at 00:34:17) What else did you want to talk about today? We've got a couple other things. One of the questions that I wanted to ask is the difference between Voximplant and Twilio.

(Andre at 00:34:28) So there are a few points that we're different in. The first one is an architectural design. When we started Voximplant, we created this as a serverless platform from the very first day, right? So there was not even a word "serverless" these days, but we created a serverless platform.

(Andre at 00:34:46) So for basic solution, you don't need to have any server yourself. You can develop everything on Voximplant itself and it will work. In Twilio's case, it's the other way. So Twilio always communicates with some web servers to get instructions on what to do next. Later, they introduced Twilio Functions, which is, well, basically they host web servers themselves.

(Andre at 00:35:11) Right? And they communicate with their own solution. So it's more reliable than communicating with customer solution over the internet. For writing more complicated solutions, I think it's hard to use. To create something easier, it may be simpler to do with Twilio.

(Andre at 00:35:28) Right? If you need something really, really simple. But if you need some more sophisticated solution, doing this with Voximplant is really easier.

(Joel Beasley at 00:35:38) Is it a large part of your business where people move from, like, Asterisk or Twilio and come over and you have to help migrate them? Or is that really rare?

(Andre at 00:35:49) I don't think that it's quite an often case. Right? There are a few customers that perform migration from Asterisk or from Twilio. But the idea is that if you use Twilio Voximplant for more than, basically, an SMS solution or a full service, it's quite hard to migrate. Right?

(Andre at 00:36:15) Because you are integrated really deeply, and you use a lot of features from CPaaS platform. And you need to have really strong arguments to migrate. So it's hard.

(Joel Beasley at 00:36:29) It's good though, because while you won't get a lot of people that migrate often, one of the things is the people that build on your system will likely stay. It doesn't seem like there's a lot of migration in that industry.

(Andre at 00:36:42) Yeah. For example, there are other companies that created compatibility layers for their CPaaS platforms. There are some who work the same way as Twilio for easy migration, but we decided to go our way. And yes, we discussed creating a Twilio migration tool a lot of times, but we decided not to do it because it's not a good way to start. Right? Because if you convert all the Twilio XML documents to Voximplant scripts, it wouldn't be something that you could be able to maintain.

(Andre at 00:37:18) Right? Because it would be generated code that is basically a lot of letters. That's it. It's not a proper way to develop with Voximplant, so we decided not to create such a migration tool. We decided to explain to people how to work with Voximplant properly and create more examples, more solutions on our GitHub.

(Andre at 00:37:41) But we don't want people just to use some code generation tool for migration from Twilio.

(Joel Beasley at 00:37:49) Are you guys doing anything with low code or no code?

(Andre at 00:37:52) We have VoiceportKit, which is basically an old school no code contact center, but its functionality is limited to contact center. And you can't create all this stuff that you can create with Voximplant using this visual editor. But we are thinking about creating an old solution for Voximplant. We understand that these days, developers are very, very busy, really busy. And creating an old solution that can be used by a guy with not so extensive technical skills can be valuable for our product.

(Andre at 00:38:26) We think that it can move us to the next level. So we're going to create it next year because it may be a bit professional.

(Joel Beasley at 00:38:36) Oh, okay. So then you're gonna have to come back next year and share what you're doing.

(Andre at 00:38:40) Yeah. Exactly.

(Joel Beasley at 00:38:42) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn, or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.