S2 Episode 5: Unveiling the Future of AI in No-Code Conversational Development

Welcome to the fifth episode of S2 of the LowCode Podcast! In this episode, we sit down with Braden Ream, the mastermind behind Voiceflow.

Join us as we delve into the world of conversational AI, exploring Voiceflow's evolution from a no-code platform for Alexa skills to a multi-channel assistant builder. Whether you're a conversation designer or simply intrigued by the future of AI in no-code conversational development, this episode promises a behind-the-scenes look at one of the industry's most promising companies.

Jesus Vargas: Hello everybody and welcome again to the Low-Code Podcast. Today we have Braden Reem with us. He's the founder of Voiceflow, which is a voice product in the no-code space. Pretty exciting, I would say pretty unique. Braden, thanks for being here with us today. And why don't you tell us a little bit more about what Voiceflow does.

‍

Braden Ream: Sure, yeah, excited to be here. So, Voiceflow is an easy way to build conversational assistance. So, both chat and voice used by over 100,000 teams now. So, think web chat, think WhatsApp assistance, messenger assistance, call centers, really any sort of automated conversation for any use case, you can use Voiceflow. Typically, we skew towards customer support, but we've also seen automated drive-throughs, we've seen... bedtime stories, we've seen games, sort of the whole plethora of use cases for assistance.

‍

Jesus Vargas: Did you start with voice and then went to text or text and then voice?

‍

Braden Ream: so we, yeah, I mean, you know, you can kind of tell from the name, right? So voice, voice flow was originally a no code way to build Alexa skills. And then we expanded to Google Actions, and then we expanded to chat and then web chat and then SMS. And you know, it's just sort of spiraled from there to, we basically cover everything now. So you can build conversational assistance for any channel. And then the cool thing is being able to build assistance that can talk over multiple channels as well. So, you know, imagine being able to call your bank. have a conversation there and then, you know, when you're on your web chat for the same bank, be able to have a consistent conversation over the two channels.

‍

Jesus Vargas: Has voice exploded in the way that you guys imagined or not yet?

‍

Braden Ream: No. So, you know, when we were initially building the company, we thought Alexa skills were going to become almost like the new websites. And you know, it didn't it didn't turn out that way. You know, I think I can sort of safely say after at least a couple of years, you know, maybe that still is the future, just hasn't happened at the trajectory that we thought. And so there's a whole bunch of reasons for that. 

‍

Jesus Vargas: Why do you think that has been the case?

‍

Braden Ream: I think voice as an interface is very unique in that it has a lot of advantages but also a lot of disadvantages. So the advantages of voice typically is its speed. Its speed is the ability to not have your hands. And so here's a good example, right? Let's say you're picking like a Netflix movie. If you know what movie you want, you know, you don't want to use that like horrible keyboard thing, you know, we have to use a remote to like, you know, type something out. It's horrible, right? you would just want to use your voice. Voice is like a really high bandwidth input as an interface if you know what you want. The problem is if you don't know what you want and you need output, now imagine, let's take that same Netflix scenario. Suddenly you say, hey, I want to watch an action movie and Netflix has to start listing out verbally all of the movies that they have, right? really bad experience and since you realize that voice has a really high data input and a really slow data output that creates a lot of cognitive overload, right? Like you know, a picture is a thousand words. Well, you know, you don't want to have the interface that has to say a thousand words. And so that interface limitation, like it got people really excited because using your voice is really cool. However, it's hard to build really high quality, like in deeply engaging apps that have a really bad output interface. So frankly speaking, you know, I think voice has been really successful and it's going to continue to be more successful as an input interface for a lot of like command and control type applications. Things like, hey, how much revenue did we do today? It's a single turn and you know exactly what you want. For those types of applications, it's really good, but you're not going to see a lot of big businesses built where, you know, the entire business is a single turn application. So it's really more of an additive interface to existing applications, at least today.

‍

Jesus Vargas: Have you seen any app out in the market that's like mainly voice-based? I don't, like every wheel you like, Siri and Google Assistant, have you seen like something like, wow, this is like a voice only or voice mainly solution?

‍

Braden Ream: Yes, so entertainment's actually been a really good stronghold for voice only applications. So there's a lot of companies that are doing a ton of revenue, like building big businesses, and they're just voice gaming. And voice gaming is fairly unique because voice as an interface is also a one to many interface. Meaning, like, so let's say, for example, you want to play, you know, you go over to your friend's house and you have to ask how many controllers they have, you know, if you want to play Mario Kart or something. If they don't have a physical peripheral interface, you know, the controller for you to use, you can't play. Cool thing with voice is it's infinitely scalable, right? You could have a thousand people technically all use the same smart speaker to play with. And so that's a really cool thing. Of course you also have, it's completely hands-free and that allows you to create games where you could be a little bit more active. Maybe it's like freeze dancing. Maybe it's like song quizzes, a really popular game where it's like... get your friends and family around a smart speaker, it's gonna play some music, you have to guess the song, the song title and artist. Stuff like that's been really, really popular and continues to be a main use case for voice. But in terms of standing on. 

‍

Jesus Vargas: Are these products being built in voice flow or just voice but not using your product?

‍

Braden Ream: It's a mix. I mean, we have a lot of games on Voiceflow for sure. Like our early, you know, Voice... I don't know if this is that still true. You know, at one point we were, you know, at one point our company's KPI was how many Alexa skills are hosted on Voiceflow, right? Which, you know, if you're an Alexa-skilled app or, you know, platform, like that seems like a good KPI. And, you know, I think we had like six or seven thousand at one point and there were only a hundred thousand Alexa skills. And so we had like six or seven percent of the entire ecosystem was on Voiceflow. So we were like... Fairly large, we're certainly the largest platform by leaps and bounds, if not like an order of magnitude. And yeah, we had a good mix of games. We never had like the number one or the number two games though, so unfortunately, we're probably, we have a couple in the top 100, but nothing that's doing crazy, crazy volume.

‍

Jesus Vargas: Did you start the company as a no code company or did you eventually found like the no code space and decided to jump into the bandwagon and become like the no code voice app? No code voice app.

‍

Braden Ream: Yeah, so how we started the company, it's actually kind of funny. We're huge fans of Webflow. In fact, Webflow's, we were the CEO and founder of Vlad. We were investor. his first angel investment. So before I knew and knew about Webflow. And so we had built like all of our websites. My co-founder had been using Webflow basically since the start. And so we had dropped out of college to start Voiceflow. And in all honesty, when we were coming up with a name for the company, We weren't sure, you never know if, we just needed a name for the project, right? We didn't think this was gonna be some sort of massive company. And so we just needed a name, right? And so we're like, I don't know, it's like Webflow for voice apps, right?

‍

Jesus Vargas: Oh, okay.

‍

Braden Ream: And that was the thesis behind the name in all honesty. And then it's funny now, cause we're seeing in lists next to Webflow is like a top no code tool. And it's kind of funny cause like, I wish we had a more distinct name than.

‍

Jesus Vargas: Yeah.

‍

Braden Ream: It's especially given we're in the same list, but a lot of projects just start small and they get bigger over time. We did not know that this was going to be the size of company it is today. I say that like it sounds like we're huge. We're not huge. We're about 50 people, but certainly,

‍

Jesus Vargas: Yeah, not small.

‍

Braden Ream: yeah, we're not like the three people we started with. Yeah.

‍

Jesus Vargas: Do you cater to the no-code builders or are you more focused on enterprise clients building support chatbots?

‍

Braden Ream: Yeah, it's a great question. We used to focus on no code as our primary market. We no longer do. And I think so no code. Like, when you think about no code, like kind of anything with a GUI application is no code. You know what I mean? Like there's no really good definition for no code. And so I think when you talk about no code, what it actually is, is you're appealing to the community of people who love this subset of tools, where they're like builders, they're like citizen developers, hobbyists, stuff like that. But like, I don't know, like you go to, like yeah, Airtable is a no code tool, but it's also like a like a GUI database. I don't know, you look at like, what's another good example? I mean, even like a lot of Oracle's tools, it has a GUI over top of a technical solution. So where do you draw the line for what is no code and what is not no code? And so I think really no code is just a community of people who love these, they're builders. There are people who like the tinker on the side. And they've like. created this community around the technology when really the community is not even about the technology in my opinion, it's about the people who just love to build things and they might not have a technical background though, or maybe they do, you know, low code, no code. And so that's kind of how we view that space now is like, I actually think it would be really hard to build a like, business that purely focuses on low code, right? Because like, that is such a diverse community with like so many different use cases and people want to build different things and like, You know, you can't charge a ton of money for people who are just doing this out of passion. And so at some point it becomes like a, you know, you end up attacking a specific business vertical. And at that point, are you actually no code anymore? Right. Cause like if you initially start, you can see what I'm saying. So, yeah, now we’re

‍

Jesus Vargas: I think that it’s a sign a sign, like a sign of maturity. And now that we're recording the second season of the podcast, interviewing founders like you, everybody or most of them start targeting the no code space, the builders and the no coders. And then eventually they outgrow maybe their own community or where they started. Because, because

‍

Braden Ream: Totally.

‍

Jesus Vargas: it's because the business needs a different client. Is it because the product is so robust that you no longer target that user base? What changes?

‍

Braden Ream: Totally. I think it's such a good point. If you start to define no code as a community and an archetype of someone who likes to be early adopters, they like to tinker with new technologies, they like to build things in their free time, most of them have day jobs, maybe some of them do freelancing and stuff, when you start to think of it as a community, it's probably one of the world's best communities to launch a new product into. right? Because this community is so willing to try your product out and see where it fits in the stack and how it compares all this kind of stuff. But you can't really build a huge business only targeting like no like if you said my entire business is focused on no code people it's like well, what you know, what's the app business application? I think actually, you know what, with the exception of maybe like Airtable or Zapier, like the connectors, you know, there's probably a wide enough surface area where you know, you could actually build a business there. But for most businesses like Voicelow, We launched into that community. They're incredibly supportive, gave us a lot of great feedback. And then we started to get pulled into contact centers. Then we started to get pulled into, like more business facing teams. And then from there it was like, oh, this is probably where we should build the business. And like, we actually now view ourselves not as a no-code tool, but a collaboration platform. And of course, like we're a GUI interface and we're still used to, people still build Alexa skills on Voiceflow, but now they... They also build WhatsApp assistants, they build a whole variety of things. And so, yeah, it's interesting how that pivot happens. And I really think it just happens from customer pull. And I think for founders who are in the no-code space, I think it's really important to know that like, you know, it's really hard to build a business that only caters to no-code. At some point, you need to figure out like, what's the business use case? And then you're bringing your no-code solution into that business use case.

‍

Jesus Vargas: And did you did you start getting clients in the corporate space the enterprise space and then building for them? Or did you started looking for that type of client and then building features for them?

‍

Braden Ream: Yeah, it was the former. So we actually just basically what we found was. We were talking to customers and we saw at one point a really big well-known telecom company was using voice flow and we're like, huh, what are they using us for? What Alexa Skill are they building? And then we talked to them and they said, well, we're actually not building an Alexa Skill. We're using this to prototype call centers because there's no good prototyping solution. We either have to ship it with real code, which is expensive, or we don't get the user test and we end up shipping really bad experiences. And so they're like, we're using your Alexa Skill building app. as a way to simulate when a customer calls our phone tree, right? And then just see how the experience goes. And so that was like a light bulb moment for us to start looking into call centers or called IVRs is the official like that's the automation layer that goes on top of the call center. You know, so when you call a bank or something, press one, press two, that kind of stuff. And we got pulled into that space. And then those people are adjacent to web chat. And then those people are adjacent SMS and WhatsApp. And you just start getting customer like customers are pulling you in. to all these different use cases. So, yeah, it’s kind of.

‍

Jesus Vargas: Does that mean that are you competing with Twilio or not really or in some cases with Texthun?

‍

Braden Ream: No, so Twilio does like all the, Twilio is very developer led. So like, you know, our positioning is really as a collaboration tool first and foremost. And so a lot of our customers will collaborate, design, prototype, build on Voiceflow, and then they might host it on Twilio, right? Like they have all the telephony services or, you know, whatever, same thing with like the NLU stuff. They're going to design, build on Voiceflow and then host on Dialogflow, right? Or IBM Watson or whatever their chat bot platform is. So we're really like, you know, When you think about what's unique about voiceover as a business, it's largely that we have focused basically the past three or four years now on doing nothing but team workflows and a really high focus on the creation experience, right? Because we haven't had to think about number provisioning or the underlying telephony or building a natural language understanding unit. There's all this stuff that if you're trying to go full stack, you have to do it. It gets... you end up being kind of okay at everything versus we've

‍

Jesus Vargas: Mm-hmm.

‍

Braden Ream: said we're going to be the absolute best like conversational system builder in the world by like such a wide margin that people will use us purely for the build portion of it and then export into whatever their other services.

‍

Jesus Vargas: Yeah, that makes sense. Now getting into AI, which is the topic of the year, how do you see voice and AI merging together? Because we're looking at ChatGPT 3 with text, it's great with images now. Voice is the next one. Is it there yet? Will it happen soon? Or do you see voice, AI, and voice flow all connecting?

‍

Braden Ream: Yeah, I mean, so I always encourage people, when you think about conversationally, voice and chat are the exact same technology stack with only one extra step, which is turning the text into speech or the speech into text, right? But so how the systems actually work though, right, is when you speak to a phone system or something, like some sort of voice bot, it's turning your speech into text and then it treats it like a chat bot, right, after that. And so there's only one, you know, there are more sophisticated systems that have, you know, extra layers or, you know, they might muddle things together, but for the, you know, simplicity, like that, that's generally how it is. Um, and so, you know, you like voice was already applied like GPT three, like you can do it at runtime in voice. So you can create like, you know, chat GPT style, like voice bots and stuff like that. I think that that's all really cool. Um, and like, like, so we work with about. I don't know if it's, yeah, like say roughly 10% of the Fortune 500 now, like a pretty good chunk of big businesses. They think it's really like large language models like chat, GPT, and all these kind of things are really interesting. However, their tendency to make stuff up is really problematic for an enterprise.

‍

Jesus Vargas: Okay.

‍

Braden Ream: You spend 100 years to build a business, you know, and a brand. and it only takes a couple interactions with a bot that makes stuff up to tear it down. So enterprises for the most part, like when you think about large language models, you should be breaking it up into two components. There's the runtime, which is what's actually happening when the app is executing and at creation. So we've already applied in VoiceSlow, tons of large language model stuff at creation, which means it's suggesting responses, it's suggesting prompts, it's suggesting sample data, all that kind of stuff. But that's... when it's being created so that when it actually gets pushed to runtime, it's everything that you've already designed. There's no surprises, right? The large language model is not creating it on the fly. And then you have the runtime component, which is where it is creating on the fly. And I don't think you're going to see it used, at least for customer support, for a couple of years. You can go play with this, by the way, today in Voicela. If you want to play around with WhatsApp and Chat GBT, I posted a video about this earlier today. It's really fun. But from a business use case. It just goes all over the place. Like I had an issue where we have our own Voicelow customer support bot, and we had a GPT-3 running with it. And it was funny. So I messaged our support bot and I said, Hey, I want to make a purchase. And it goes, sure. What do you want to buy? And I said, Nike Air Forces. And like, it was walking me through the purchase flow of like Nike Air Forces on the Voicelow website, which obviously we cannot do. And then at the very end, I go, did the, did the order get made? And it goes, yep, orders placed. And I was like, okay. Right. Like. If you were a customer in a big brand, that's really dangerous, right? It just made that up. And it knew to go through this whole purchase flow, it knew to ask for my shipping address and the size I wanted, the color I wanted. And then it had the audacity to tell me that the order was placed. So it's like, you know, that's really dangerous in the enterprise. So I think it's all really cool. I think you're going to see a ton of it on that creation side over the next couple of, you know, months, really, to be honest. Like we've adopted it super quickly. I think the runtime will take a couple of years, though.

‍

Jesus Vargas: That's interesting. Where is Voiceflow going to? I mean, you're already moving or moved to the enterprise level, but as a product, how do you decide which features to prioritize when you're building the product? Is it based on customer feedback? Is it based on any target client that you want to get? How do you decide, and where do you see the product evolve?

‍

Braden Ream: Yeah, I mean, this is such a dance of like, this question is an art, to be honest. And so certainly a lot of it is customer feedback, but if you only listen to customer feedback for your roadmap, you'd end up, you know, we, we, you know, we'd still be building Alexa skill features. We wouldn't have had like the business, you know, in intuition to move us. 

‍

Jesus Vargas: Yeah, you get stuck with a difference audience

‍

Braden Ream: Totally, right? And so, you know, you need to balance. Because at the end of the day, you want your customers to win. And so you want to listen to their feedback and build what they want. But if the business goes out of business, then, you know, that's not very helpful for customers. And so, you know, it's our job.

‍

Jesus Vargas: I think it's especially challenging for a product like yours, because if you look at, let's say, other NoCo tools, they look up to the bubble and the whatever that's already out there. So they have to get their feature parity with all the tools that are out there. But in your case, you're in a space that maybe there's not a lot out there. So it's more on your end to come up with new things, right? And then the customers can use, hopefully, and they'll pay for them.

‍

Braden Ream: Yeah, it depends. So it depends how you define us. Like, you know, the true like ICP, like the customer profile that we target is what's called a conversation designer. These are professional UX designers who focus on like conversational automation. For them, we're like the best thing since sliced bread. We're the first tool to like be dedicated to them and, you know, trying to build a tool that's like the first collaboration tool If you view us like a Twilio, we're horribly under parity when it comes to features, right? It really depends how you view us. Like we actually had a chatbot builder review site, say Voicelo is like the worst chatbot builder because they don't have analytics, they don't have this, they don't have that. And I was like, we don't because like, that's not our goal. Our goal was to like, basically be a collaboration tool and be the absolute best design tool. that then feeds into whatever your production platform is, right? Like we feed into your dialogue for your Watson. So like, why would we do analytics? And it was kind of funny on that review site, it said they lack analytics, they lack integrations, they lack all this kind of stuff. But I was like, hold up, we have commenting sticky notes, real time collaboration, versioning, like we have all the design tool features, right? Like, you know, this, the CEO of Figma is one of our investors, same with Envision, like we very much put ourselves in the design tool category. So, it’s kind of funny like how you slice. Where you categorize your product it's either going to be the worst product or the best product so we're like the best tool in the world by like a hundred X for Conversation designers and like for just like a basic chatbot builder for like a small business or something We're probably the worst and so it's like it's kind of funny

‍

Jesus Vargas: Yeah, especially when you start out somewhere and then you move on to something else, people want to hold on to the product that they remember that it was built for them and now it's no longer is and it's a disconnect that it's challenging and yeah, support is...

‍

Braden Ream: We've got lots of revenue and companies doing super well. I think we've raised just under $30 million now and we're doing really good. And then I have some investors who invested really early who are like, wow, the voice market's really popping off. And I'm like, that's not even what we do anymore.

‍

Jesus Vargas: No.

‍

Braden Ream: Even the people who get my updates every single month still think it's like, well, actually, like what? Like what Alexa skills are driving all this growth? I'm like, it's not Alexa

‍

Jesus Vargas: Right.

‍

Braden Ream: skills, right? You know, it's, it is kind of funny because at the end of the day, we're all living our own lives and you know, people generally don't pay attention to, or you know, as much as they should to things that are going on. Like we typically, new investors who come to us now will have a better understanding of what we do because they're looking at voiceover fresh eyes versus existing investors who might like, you know, have a business from three years ago in their mind, which is when we pitched them. So I totally get it. It's just, you know, it's kind of funny.

‍

Jesus Vargas: Will Alexa Skills and Google Assistant will have a comeback?

‍

Braden Ream: Well, Google Actions actually just got deprecated. So I think they're out for the count. ButAlexa skills, I still have a lot of faith, to be honest, in the Alexa ecosystem. Like, I view, I view assistants very much as like browsers, essentially. And so I think we're leaving like the Netscape era where you interacted with this browser, like an Alexa, and all of the apps that were there had to be built within its walled garden. That was very much like the Netscape approach to the internet. I think now what you're seeing is companies who use voice flow are building their own assistants. You're going to have every company is going to have its own conversational assistant that can talk to you across multiple channels. It can carry context. It learns more about you over time. You can use your voice. You can chat. You can do whatever. And every company will have that. And then I think Alexa and these other personal browsers, like these personal assistants, will then store your... your preferences and they will then start interfacing on your behalf with these other assistants. Right? And so I think that's the new model that we're now heading towards. We just left the Netscape era, that was Alexa and Google, and now we're heading towards a more open web approach to conversational assistance, which I think will be really cool.

‍

Jesus Vargas: Braden, thanks so much for joining us today.

‍

Braden Ream: Yeah, it was a pleasure speaking with you. Thanks for having me on.