Dave and Evan meet with Philipp Heltewig, Co-Founder and CEO of Cognigy.
Philipp is a German-Australian entrepreneur with profound technology experience working for companies in Europe, APAC and North America. He had the foresight to start up a conversational AI technology company in 2016. Long before anyone was talking about conversational AI.
Since its founding in 2016, Heltewig has been the CEO of Cognigy. The company focuses on the automation of customer and employee services through Conversational AI. Cloud or premises-based, text or verbal, as well as employees or customer interactions.
In this podcast you will learn more about how and why Cognigy stands out in a sea of conversational AI companies.
Dave and Evan meet with Philipp Heltewig, Co-Founder and CEO of Cognigy.
Philipp is a German-Australian entrepreneur with profound technology experience working for companies in Europe, APAC and North America. He had the foresight to start up a conversational AI technology company in 2016. Long before anyone was talking about conversational AI.
Since its founding in 2016, Heltewig has been the CEO of Cognigy. The company focuses on the automation of customer and employee services through Conversational AI. Cloud or premises-based, text or verbal, as well as employees or customer interactions.
In this podcast you will learn more about how and why Cognigy stands out in a sea of conversational AI companies.
Tonight Evan and I will be talking to Philip Pennell, wig CEO and co founder of cognitive G. But first, Evan, are you in a physical environment or in a meta environment?
Evan Kirstel:I am really excited about feda, the new social networking platform for cheese lovers. Fetta is going to be huge, as well as meat meat. It's going to be exciting.
Dave Michels:So for those of you that are listening to this podcast, you should all know that Evan and I record this in a incredibly detailed Metaverse simulation. It's a shame you can't see. We're in the giant fish tank. There's all kinds of exotic things happening around us. And unfortunately just doesn't translate well to an audio only experience. But but you just have to trust us.
Evan Kirstel:And we have torsos. I saw the new Microsoft Teams version of the metaverse and for some reason no one has torsos. What what's going on there?
Dave Michels:Well, yeah, I've always been a leg man myself. And it's kind of a you know, what's the point? But in
Evan Kirstel:all seriousness, does the metaverse ever make you ever want to work again, it didn't really inspire me to want to actually work.
Dave Michels:I believe the metaverse is real, I believe it's going to be huge. But I believe all this stuff that we're seeing now from Facebook, and Microsoft is just utter nonsense. Not only is it not real, but it's not even interesting. But it will be interesting at some point. The thing that we're going to flip is when we do things in the metaverse that we can't really do in a physical world. And so that will indeed happen. That's the way disruptive technologies always happen. It's always addressing some sort of new specialized need that makes it work. And that's great. But the idea of going to a meeting and dealing with cartoon characters just doesn't seem like an upgrade to me.
Evan Kirstel:Well, what about sending a fax in the metaverse, it says that going to be a thing you can do you know I'm an anti Vaxxer. Well, I'm looking forward to our next meta podcast. But in the meantime, we have a great guest in the real world. Well, let's
Dave Michels:get to it.
god:Talking Heads is a semi monthly podcast with interviews of the top movers and shakers and enterprise communications and collaboration. Your hosts are Dave Michaels and Evan Kersal. Both of which offer extraordinary services including research, analysis and social media marketing. You can find them on Twitter, LinkedIn, or at talking points.com. That's points with a Z and Devin Kirsten calm. That's KR STL.
Dave Michels:Today we have with us Phillip held to wig which is the American pronunciation of an international globetrotter. Welcome, Phillip.
Philipp Heltewig:Thank you. Great to be here.
Dave Michels:You are the CEO and co founder of cognitive G AI Did I pronounce that right?
Philipp Heltewig:That's right.
Dave Michels:All right. So first question is how does someone anyone really become a founder of a conversational AI company back in 2016? What were you thinking?
Philipp Heltewig:Well, I was thinking that I should have more fun than that what I was doing at the time, and I apologize already to my former employer, it was a lot of fun. But I worked in web content management, digital marketing, and I worked in that company for more than 10 years, we turned it into a unicorn and sold it in 2016. But quite frankly, and never really woke up on a Sunday morning going like web content management man, there's there's nothing I'd rather do than that, right. And I'm a big sci fi fan. And so my co founders and when one of my co founders approached me in 2016, and said, Hey, I have this idea. Should we talk about it? I was intrigued because he was talking about voice AI, right. And it was the time when the Google homes and the Amazons of the world came on the stage, everyone was talking about voice AI. And there was this promise that something like the Star Trek computer would be omnipresent very soon. And that was just a very interesting space to get into. And we thought, okay, um, we had some great success with what we've been doing so far. But why don't we jump into something that's really cutting edge and where we can actually wake up on a Sunday morning and go like, you know what, I'm not going for a walk. Now. I'm going to work on some voice AI because I want to bring about that sci fi future. And now that's what we've been doing ever since.
Evan Kirstel:Fantastic. And clearly, you weren't the only one with that big idea. There are lots of conversational AI companies around Amazon comes to mind Apple, Google, how do you differentiate yourself from those tech giants?
Philipp Heltewig:Yeah, it's interesting, right? Because what's happened here is that conversational AI is a whole new category. But there are so many things that are conversational AI, so you have the Google assistants of the worlds that's conversational AI, you have Siri that's conversational AI. You have your Mercedes Benz car speaking to you on and you can speak to it. That's conversational AI, you have toys you
Dave Michels:can speak for cities is really bad. I've got the Sprinter van. And I never talked to it but it's always like what did you say? I drives me crazy. Exactly. And
Philipp Heltewig:in a German accent, right, but what what did you say? I did not understand this. anyways, then the things like chatbots, right. And also with chatbots. Like there's a whole variety of bots, you have marketing bots on Facebook and Instagram, you have bots that live on internal communication channels like Microsoft Teams, etc. So what I'm trying to say is sure, the conversational AI market is huge. But there's so many different flavors of conversational AI, right? It's kind of like saying, Okay, so we're getting into the programming market, and you go, like a programming can be used for a lot of different things, right. So the space that we're in, which is enterprise conversational AI is actually it's a big space, but it doesn't have that many players in IT. And especially when it comes to end to end platforms, a large enterprises can pick up and build these kind of automations with inside and outside their enterprise. So you spoke about the big ones. Apple is a company that does not play in our space, from the big ones that do play in our space, we would have Google, Microsoft and IBM. But what they have, essentially, bits and pieces of the enterprise conversational AI stack and not an end to end solution. So we're the solution that pulls all of these things together. So we differentiate in that way that we are holistic solution, rather than just a part of the solution. And we focus strictly on enterprise applications. I want to
Dave Michels:get back to that end to end stuff a little bit down the road here. But before we get there, when you talk about conversational AI, or I guess, when I talk about it, people get confused. Is that a text thing like a chatbot? Or is that a voice assistant? What kind of conversations are you referring to?
Philipp Heltewig:So it's actually both, right, because the modality as in, we're using chats on various channels, we're using our voice doesn't really matter in the end is really all the same. The voice bots have just one extra piece in front of them, which is a mechanism that converts the voice into text call speech to text, and then a mechanism that converts the outputs from the bot, which are textual outputs back into a into a spoken voice, right. So in the end, it really depends on what you want to achieve. Most of the company internal bots that we have, are chat bots that live on channels that Microsoft Teams, Slack, zoom, etc. Whereas most of the bots in customer service will actually not most, but let's say it's 5050 50% of the bots are on the phone lines. So built directly into contact center software, and the other 50% on the chat lines, right. And the vision that we're selling, is that you really want to be able to help your customer on the channel that they choose if they want to call in fine. If they want to text you via SMS or WhatsApp or Facebook Messenger fine, they should also be able to do that, right, the service level that you offer on those channels should be the same.
Evan Kirstel:So aren't all conversational AI companies doing text and voice? I don't
Philipp Heltewig:know all conversational AI companies. But I don't think so like if you look at let's take Microsoft, Microsoft offering until two days ago was purely text based, right. And that just introduced voice into the mix. And there are actually a lot of vendors that are only doing texts out there. Because there's a lot easier to do as well integration onto the phone lines integration into contact centers, is a very difficult technical piece, right? You need to deal with protocols that are rather legacy in comparison to some of the protocols you're dealing with. Now on the web, right, where everything is API based, it's easy rest API's, etc. authentication mechanisms follow modern standards. So if you want to integrate into some older versions of Genesis or a via etc, this becomes a whole different beast. So I think it is very natural for conversational AI companies to start with text, and then to progress into voice. But some also never progress into voice. I spoke with a company yesterday from Berlin, and they are doing chatbots for marketing. And they will never go into voice, right? They only live on Facebook, on Instagram, on Twitter, etc, etc. They have no need to exist on the phone lines.
Dave Michels:Hmm. Well, okay, I gotta ask you a tough question here. I've seen a lot of demos of conversational AI. And they're always pretty good. I have to say they're really impressive. But I've also had a lot of really bad terrible experiences with when I call into contact centers, and get these awful, awful AI conversational bots that don't understand anything. And don't help anything. So why are so many implementations so bad? What's going on here?
Philipp Heltewig:Well, I think it's a mix of two things. Firstly, you said you saw a lot of great demos, of course, who would willingly give a bad demo, right? So who knows what that means. I mean, sometimes we're getting sent videos of demos by prospects that go like, Well, why can't you do that? And the videos, they are just fake, right? You look at them and you go like, okay, sorry, that this type of general AI technology that can go like, hey, so why do you want to open a bank account and then you go like, ah, because my niece is going to school from next year on Oh, where's she going to school? And yes, I'm sorry that the board does not say that. I mean, it doesn't have to say that either. So there is some smoke and mirrors stuff out there. But then On the other hand, even if you have a capable platform, conversation design is a discipline that hasn't existed even five years ago. Right? How do I design these conversations so that they actually usable so that they actually easy to use? Right? And I think this is where a lot of companies still fail. So it's a mixture of Do you have the right tool? And do you have the right skills to actually build a useful conversation on a given a given example? Let's say you're booking flights via voice, and the system goes, Okay, what do you want to fly to San Francisco? Okay, where do you want to fly from? And then you say, actually, how's the weather in San Francisco? Most systems will probably say, sorry, I don't understand your answer. All they will say, Okay, you want to fly from San Francisco, because there's some only detected San Francisco and your answer to the question, where do you want to fly from? Right, right? Well, that's not how it should be. It's a digression. In the conversation, you answer the weather question, if you can, or you say, Well, I'm sorry, I don't have an answer for that. Let's continue on the booking process, right. But that's conversation design. How do you handle that? And this is especially important in bots, because people want to also play with the bots. Everyone does that, right? Like you're calling into a bank to, I don't know, get a bank statement, copy or whatever. And then people all of a sudden, go go, when's Donald Trump's birthday? Well, I don't know when you called your bank the last time and when a human answered, you said, Okay, I need a bank statement. Okay. Which account is that for all? By the way, before we get started, when's Donald Trump's birthday? You just don't do that all they'll hang up. But people like to play around with a bots like that, I guess. Whereas the focus should actually be on? How can we help the customer achieve what they want to achieve quicker? And so sometimes we might have to say, hey, sorry, I really can't help with that. But I can help you with a bank statement. So let's get back to that.
Dave Michels:I always asked the bot for the meaning of life. Well, on it's 42. Right. So that's, that's easy. Good answer.
Evan Kirstel:Well, it sounds like what you're saying implementation is key. And when it comes to a lot of other technologies like networking and system admin, there various certifications for that, is there some kind of certification program for conversational AI that design engineers can use?
Philipp Heltewig:I firstly, want to talk about what you said initially, which is implementation is key, I would say implementation is always key, right? You could have a fantastic web content management system, and you can build a crappy website on it. Right? So implementation is key, I think this is the thing with everything, like you have a great CRM system, if it's not configured correctly, then it doesn't help you that it's a great system. Implementation is key. So implementation is always key now to your point, for web content management, for CRM, for ERP, etc, there are various certifications, and they are product specific certifications, but they also conceptual certifications, right? So you need to understand the I don't know, underlying protocols, etc, etc. To build something now what we've done recently, we've released something called common G Academy, which also has two types of training classes in it. One is cognitive specific trainings. And because it's a, it's an open system, you can just go to our website and register for a trial and and play around with it. So you don't need to be a paying customer to play around with it. So you can go to cognitive Academy, do these classes and get hands on with a tool. But more interestingly, I think we also have a training class on conversation design, right? And we got one of the world's best conversation designers, Sacha Walter was been working with enterprises around the world on conversation design to record that with and for us. And we made that available for free as well, because we believe that, sure, it would be great if customers bought our product. But it's much more important that what Dave just said stops happening, right? Because if you have bad experiences with a bot, you are going to project that onto all software as well, you're going to go like Well, yeah, no bots, that doesn't really work. So we would like to lift the industry as a whole do our part, at least, in making sure that future generations of bots are built with certain conversation design principles in mind, so that the user experience is actually very positive. So that hopefully, at one point, we're not having an experience that's like, Oh, my God, it's a bot. But actually the opposite. Wherever human answers you go, like, Oh, my God, as a human, I wish you were a bot. Right? So let's see if we'll get there. But at least we can make the existing experiences a lot better. That seems
Dave Michels:like a lot of pressure. You described her as one of the best conversational designers. I mean, that's a lot of pressure to have like a casual conversation with a at a cocktail party or some like that would be like that's a lot of pressure.
Philipp Heltewig:Well, maybe I think it's actually a heap. But I think he would probably not describe him as that but I'm happy to meet someone who's better at conversation as I'm at so far, it hasn't happened.
Dave Michels:But when you talk about these classes that you just described, are they specific to cognitive G or can anyone benefit from those that are working with conversational AI technologies?
Philipp Heltewig:So in part, they're specific to our tool because we use our tool of course in the tutorials, etc. And there's certain things that We do different to other tools, etc. But then there's also of course, conceptual things in the trainings that you could apply with other platforms as well, like natural language understanding and how to train natural language understanding models, is very similar in cognate, G and Google and Microsoft, the intent based NLU training. And then again, the conversation design course is completely tool unspecific, right. So you can apply that with any tool that you're using.
Dave Michels:So there's all those bad experiences, I have kind of go hand in hand with a lot of the sales pitches on using this technology. And the sales pitch is often about saving money, you can't afford to have live agents answer these calls. So let's have a bot do it instead. And so what happens is, it's a cost savings measure. And it's frustrating. And like, all cost savings measures are frustrating, you know, we now close to three, etc. And so I want to ask you, you know, in your experience, or use it, are there certain situations or sometimes are there examples that you can share, where the bots are actually beneficial or preferred? Kind of a positive thing as opposed to a negative or cost savings thing?
Philipp Heltewig:Yeah, I think anyone who, especially in 2021, talks purely about cost savings when it comes to bots, doesn't really know what they're talking about. And
Dave Michels:told me you asked that question. I'll go ahead. I'm sorry.
Evan Kirstel:Frankly, I would always prefer to talk to a bot versus Dave directly. But in that relationship, we'll
Philipp Heltewig:let you guys figure that out afterwards. No, because he has a thing, right. In Gemini, and also in the United States, I think there's a labor shortage, right everywhere. And I want to see the contact center that goes like, you know, what, we have all the contact center agents we need. And we have another 1000 waiting just to, you know, untrained just to sit here and answer all calls. This is just not the case, right? There's a labor shortage everywhere. And it's actually getting so bad over here, like we're working with one of the largest airlines in Europe. And their average wait time on their contact centers is up to two hours. So you calling in. And the advertised waiting time on the website is two hours. So you need to wait for two hours before you can speak with a human on the phone. Now, I'm pretty sure that if they put in voice bots, and they handle 50% of their cases automatically, they're not going to translate that into cost savings by firing 50% of their staff. They're not going to fire anyone, right? So for them, it's really about saving customer satisfaction, right? Getting back on track, by actually being able to answer customer queries quicker or even instantly. And the concept is very simple. Would I prefer to speak to a bot? Or would I prefer to speak to a human, I would always prefer to speak to a human. I think everyone would. And maybe that'll never change bots. If I have the option to speak with a bot now and try or speak with a human in two hours, or even 30 minutes or even 15 minutes. I think I would rather try and speak with a bot now because what do I have to lose? Right? So let's say I want to cancel my ticket, my flight ticket, I'm coming in some automated voice says okay, there's an average wait time of around two hours. But you can press one to speak to our automated system. And we can try to help you in that way. You will still remain in the queue here. In case the automated system cannot help you would you like to try? I mean, why not? What what do you have to lose? Otherwise, you're listening to some music for two hours that you don't really want to listen to? Right? So you go in bought calls like hey, how can I help you? Yeah, I want to cancel my ticket. Okay, perfect. What's your booking number? Blah, blah, blah. Okay, what's your last name? Held to it? Okay, perfect. So that's your flight going from Dusseldorf to New York tomorrow at 1030? Is that correct? Yes. Okay. So it takes you through the process. Now we are seeing, depending on the use cases, we're seeing resolution rates between 20 and 90%. Now, why is it such a wide range? It's because some use cases are easy, like flight cancellation is easy already. You just need to detect, okay, they want to cancel a flight. You need the booking number, you need the last name, and they need a confirmation that they really want to cancel it. That's it. Now, in other industries, there might be queries that are much harder to answer, right, let's say, I don't know, food production or FMCG. We have a customer there and their customers are calling in with a large variety of things. I bought one of your frozen pizzas on the box that shows eight slices of salami, I only had six. So a bot will never be able to answer that right? Because such a variety of things. But like if you have certain industries, like let's say aviation, there's only so many things people asking about run, book of flight, canceled flights, change of flights, and luggage to a ticket, maybe Corona regulations. So there's only maybe a handful of things that people ask. And the interesting thing is, we've come to say that if you help one customer you help every customer. Why is that? Because every one customer that you take out of the queue contracts the wait time for everyone else. Right? So it's not really do customers prefer to talk to bots or humans? I think everyone first talk to humans. But not if I have to wait for two hours or talk to a human, then I would rather try and work with that bot to get done what I want to get done. I don't have anything to lose, right? If it doesn't work. Okay, I'm still going to be waiting for the remaining hour and 45 minutes.
Evan Kirstel:So Dave and I focus on enterprise communications, unified communications, contact center video, etc. Is that your world to? Or do you have ambitions to serve a broader market?
Philipp Heltewig:So of course, there are many use cases in customer facing context center, right, that is where a lot of the volume is, these days, that is where a lot of the pressures that Emmett enterprises are feeling. But if you look at the split of what we're doing, roughly 20% of our projects are in customer facing contact center and 80% Accompany internal projects. So those are bots, for HR for finance, for legal, etc, but the volume is reversed, right? If you look at the volume, then it's probably even 90% in customer service and 10% of for company internal bots, because of course a let's take the airline example again, the customer facing bot will have much higher traffic and much higher volumes than a company internal HR bot right. Now, the company internal bots are interesting, because they live on you see platforms like Microsoft Teams, Slack, etc. And especially Microsoft Teams is becoming really huge in the space. And then the whole integration with the Microsoft stack that sits beneath that is something that the bot must also be able to handle, right? Because if I'm using a bot on marks of teams on the bot asks me, Hey, should I put something into your calendar, it needs access to my calendar. So there's rights management, etc, etc. Now what we have is we have a platform that allows you to build both kinds of bots, right customer facing bots that integrate into the Avaya Genesis as Cisco's of the world, but also company internal facing bots that integrate into the UC platforms, or marketing bots that integrate into the social channels. So it's, it's one holistic platform, and we're seeing more and more large enterprises actually standardize and centralize on one platform. Because most software vendors have some kind of conversational AI offering, right maybe in your CRM, maybe your contact center might have one. Firstly, none of these is really conversational AI best in class. Right. And secondly, in the end, you end up with like 20 different systems for conversational AI. So we see a move amongst large enterprise to standardize that just like also happened in in other markets, right? Maybe content management and to use that, again, on digital asset management. We see the same with conversational AI, where large enterprises create a type of center of excellence, that then does all kinds of conversational AI projects in and around the enterprise.
Dave Michels:That's really interesting, because I talked to so many, you know, contact center providers that have their pre built conversational AI solution. And I just assumed everyone would just use that. But I guess you offer a generic or enterprise solution that would work with a contact center and other applications as well. So they could just specialize in one conversational design skill set and use it over and over and over. Is that what you're describing? That's correct.
Philipp Heltewig:But in the end, it's also about the capabilities of those platforms, right? I mean, we have never been in a pitch against the bolts that come with contact center software's, right, because they're not even close to being in the same league. Like, yeah, you can spin them out quickly. But then you have the experiences that you're describing, right? Because it's not about having a bot that you can have a little chit chat with, it's about like, let's take this airline example. You need to in real time interact with different backend systems, right, you need to authenticate the person you interact with the booking system, etc, etc. This all needs to happen in real time on the call. And for most of these inbuilt bots, let's just say that it's very difficult, right, and especially when it goes across the borders of that system itself. So let's take ServiceNow ServiceNow has a bot, but it can mostly interact with ServiceNow. Even the same boat also needs to interact with Salesforce and a CRM, and another CRM, let's say HubSpot or a Salesforce for service cloud, then it becomes difficult, right? So that's where enterprises I think, have tried these inbuilt solutions, and are now looking at specialist vendors. Which is also why when you look at reports, like the recent IDC report that came out on conversational AI, the only contact center vendor you find in that is actually Avaya. And that is because Avaya has an OEM relationship with us.
Dave Michels:Oh, nice. So you're saying all those contexts that are conversational chatbots are all talk, let's put
Philipp Heltewig:in different way they adjust talk, but they can't really help you and do much so yes, they can talk but they can't really help customers very well.
Evan Kirstel:So you are the Switzerland of bots as it were maybe that's a terrible marketing analogy, but so what contexts that are vendors can you work with or do you work with today?
Philipp Heltewig:Across our customer base, we are working with pretty much any contact center vendor there is right we support integration directly via the SIP protocol. So We know that customers are doing integrations with Avaya, Genesis, Cisco fear. There's various different ones, messagebird, Twilio, Telmex, etc, etc. So any vendor that supports SIP transfers we can integrate with. The reason why we don't really know is because the majority of our customers is using the platform as an on premise or on their own cloud deployment. And we only get usage metrics back as in how many conversations were handled through the platform. We're not getting information about the projects themselves for data privacy reasons, right. So, for example, we're working with a large manufacturing climbed out of Germany. And I know they have more than 200 projects on the platform. But we really only know about the details for five of these projects. So we know that integrating with Genesis and Avaya, but they might also be integrating with 10, other contact center software that we don't even know about.
Dave Michels:And then beyond context centers, are there other kinds of platforms that you regularly integrate with,
Philipp Heltewig:we look at two types of integrations, we look at integrations on the front end. So where does the customer sit? Where does the communication come through. And that can be contact center, it can be messengers, it can be UC platforms, etc. So this is the channel that the customer is using, we're integrating with a large variety, then again, also, if we're not integrating with the one that you're using, you can integrate that into the cognitive platform yourself if you want. On the other hand, we're looking into the integration to back end systems, right. And those can be the Salesforce as the service now of the world. We're pre integrating with more than 100 different platforms, right? Salesforce and ServiceNow, probably well together with Microsoft Graph are probably leading in this space for us. Right? Those are the integrations that are being used the most, we have a marketplace in the tool that lets you integrate with those, right and some of those and that there's a bit the confusing thing here at some of those like a Salesforce, there is no Salesforce for their Salesforce CRM and Salesforce Service Cloud and Salesforce Service Cloud can be a back end system we integrate with, but it can also be a front end system we integrate with, right, because you can spawn a chat on service cloud, yet the communication is answered by cognitively in the background. So yeah, we have an open system to extend the integrations in both directions. And that that is very important. And that that's one very important piece about carbon G as a whole, it comes with a large number of pre integrated systems, but you can still integrate with any other system that you want and hook that into covenant G.
Evan Kirstel:Awesome. So I associate conversational bots with inbound calls, like calling into customer service, etc. But I assume you're also doing outbound stuff. Is that the case? And is that like telemarketing? Or what are the use cases there?
Philipp Heltewig:Yeah, that's a very interesting question. We are doing outbound stuff, actually, in certain regions quite a lot like in the United States. We're doing outbound quite a lot. We're doing that for insurances. And for recorded like receivables, were at I don't know what you call it like factoring companies rather you're you're selling depth to and then they're calling out to retrieve that kind of money. So we're doing that quite a bit. And I think there's also some telemarketing now this type of telemarketing robo calls, as you call them. That's not legal in Europe, for example, I don't think I've ever received a robo call here. But even here, we have outbound cases. And I'll tell you about one very interesting one, which also shows you that I mean, we are thinking of conversational AI, like we as a society, I think can have conversational AI at the moment in a very limited way. Like we're thinking about service, right customer service and employee service. But it's so much more than that, like our company vision is that conversational AI will be everywhere, eventually, like it will be a part of this podcast, right? It it can contribute to that it will be a part of us interviewing someone for a new job, but it will be a part of a salespersons job, you know, driving to a customer getting a conversational AI to brief the sales person about the customer. And one interesting outbound use case that I personally find quite sci fi ish. But that is very exciting is a case we have here in Germany, where we are working with a manufacturer for medical devices, and they have a heart implant. And this heart implant measures your heart rate or electrical signals that are traveling through your heart. And if you have a heart attack, this device can call an ambulance for your so it's presumably connected to your phone or to another device that can then trigger a phone call or another server. Now, if you're not having a heart attack, but if you're having some symptoms, right, there's some irregularities that's being detected by this device. The device can actually ping cognitive and cognitive you will then place an outbound call to the patient's phone number and quiz the patient about symptoms. So rather hello mr. Michaels, how are you doing today? I'm doing good. Okay. We just wanted to check it's
Evan Kirstel:not used even this example. It's a little little close to home. But yes, I love the use case.
Philipp Heltewig:And if throughout the questioning, then it comes out okay, that this could be something then there is a hand over to human cardiologist right who can then take over from that. But let's think about what's happening here. Right so We have a cyborg like device built into your body. And it might sound a little bit crazy when I say like that, right? But that's what it is. Like if you told someone 50 years ago, like you're carrying a device in your body that monitors your heart, and it can call someone.
Evan Kirstel:Apple Watch, too. So yeah, it's not too fantastical.
Philipp Heltewig:Exactly. So now, if this thing can call an ambulance, something's happening. But if it's not so sure, it can call another AI, which calls you on the phone and talks to you in human language and quizzes you about symptoms, and then based on that makes a decision whether you should speak to a human professional or not, right. 10 years ago, even that would have been something you see on Star Trek, right. And nowadays that is already in the realm of the possible it's not just possible, it's being done. Right. So there's automation even happening in those kinds of areas that we wouldn't traditionally think about when we think about conversational AI. Right. And those are the kinds of cases where I get up in the morning and I go like, you know, that's really cool. Very, firstly, this can be very helpful for people it can actually save people's lives. Right. And yeah, I think it's those kinds of cases that that really get us going,
Dave Michels:Oh, it's Evans, pacemaker calling, again, just ignore it to the voicemail.
Evan Kirstel:Dance case, you'd have to send a fax. But in any case, I do have one point of confusion. We spoke earlier about Amazon and Google, Can you clear up the components of what makes conversational AI? Can you work with those systems in a complimentary way? Or are you sort of an alternative to Google Alexa, etc?
Philipp Heltewig:Yeah, no, we can work with them complimentary. And we integrate heavily into those systems as well. Now, if you look at the different components of the conversational AI stack, you have the channels. So customers need to somehow communicate with your bot, right, like I described earlier, like a contact center or a messenger or something like that. Then you have the natural language understanding piece, which is the piece that that takes her input and make sense of it. So want to cancel my flight, could be a flight cancellation intent, or, I don't know, I have pain in my left arm could be a pain intent, or whatever it is. So that's the natural language understanding piece. And then you have the it's sometimes it's called fulfillment or business process piece. Now that we know what the intent of the last thing the person said, which is called an utterance. Now that we know what the intent of that utterance was, we've extracted the information that is in that utterance. Now, what do we do, and this is actually a very interesting piece. And to me, this is the most interesting piece, because here you have things like conversation, context tracking, etc. I'll talk a little bit about that in a second. And then the fourth piece is the backend system integration. So you need to have all of these pieces in order to have a fully functioning conversational AI system that can actually deliver benefits to the customer. Now, if you look at the likes of, say, Google dialogue flow, that's Google's conversational AI system. With Google dialogue flow, yes, they have an NLU system. That's it? Well, it has some channel connectivity as well, but it essentially detects the meaning in your sentence. And then you need to write code to then do something with it meaning, so you need to write this whole conversation design the whole conversation management piece yourself. Now they have released a new version, which now contains a very basic piece of conversation management as well. But it's nowhere close to what a true enterprise would actually require. Right? So I'm not saying we're the only ones doing that, right. There are competitors that are doing these kinds of things. And that are doing them at a similar level to what we're doing. But the big ones are, are not really, at that level. Like let's take actually Amazon's really good example. Amazon has a system called Lex, which is their NLU stack, right? So the assigning meaning to an utterance, they also have different channels. So they have the one that you mentioned earlier, I can't say her name because she's sitting here right next to me, and she'll stop blabbering. But they also have contact center, right? They have Amazon connect, etc. Now what we can do, we can be the fulfillment piece behind amazon Lex, for example, right? So you use Amazon connect, that connects to Lex for the bot capabilities, and Lex sends the information to cognate G. And then cognate G does whatever we do with the conversation, context mapping, etc, etc. So yes, we actually have several customers that have already existing contracts with Google or Amazon or Microsoft, and then decide, You know what, but we want a professional conversational AI system on top of that, that helps us make the most of that existing investment that we have. And that is something we do so we I learned a new word the other day, it's composable, right? Everything has to be composable these days. And we are a composable conversational AI platform even without intending to speak to that new marketing word. But what that means for us is you don't have to use carbon G's NLU system, you can also just use Lex, but then you still get all the other things that come to brings, right or maybe you want to use dialogue flow, and then you still got all the other things that communtiy brings. So yes, we do integrate with these systems as well.
Dave Michels:Wow, that's really, really interesting. So you mentioned that IDC marketscape that just came out last month, I guess it came out on October the It's interesting, I wrote about it in my insider report 15 providers in that report, and I think you are arguably in the best position on that chart. Congratulations.
Philipp Heltewig:Thank you very much.
Dave Michels:Can you explain a couple things? I noticed like Microsoft is on there, and so is nuanced. But I thought Microsoft bought new ones. And they have two different technologies.
Philipp Heltewig:I don't think they have bought new ones yet. I think they have announced that they want to buy new ones, but it's not through regulatory checks yet. So I think for now, they exist as two different companies. I don't think they've made the integration plans public yet either. So So what's the
Dave Michels:secret to getting in such a prime position on a chart like this in the leaders were the top of that leaders were? What do you think they understood about you that others should understand?
Philipp Heltewig:So I think if you look at the Googles that the IBM's, etc, of the world, they are not focused on conversational AI, right, they're doing many things. So I find I mean, look at Salesforce, right, when Salesforce entered the market, they were focused on CRM, and there was a lot of other players in the market already read like an Oracle, like Microsoft, etc. And then Salesforce just overtook everyone. Now, Salesforce is also one of the big ones nowadays. But in the past, they were not they were winning, because they were focused, like really focused on one thing, and they wanted to excel in that. And we are really focused on conversational AI and conversational AI for enterprise. And we demoed what we have to IDC. And I think they were impressed by the capabilities. And especially we did not demo by showing slides we demoed by going in, and essentially doing a demo where we said okay, and if you want to know more about this, or that tell us we can pivot straightaway in the demo. And I think that that is the thing that they understood. There's a lot of smoke and mirrors out there in the market, like we spoke about at the very beginning of this podcast, right? With Kotlin G, every single thing we say we can do, we can do, right. And this has been our philosophy from the start. And I think IDC saw that in the demo, and thus rated our capabilities as market leading globally. And obviously, we're super happy about that.
Evan Kirstel:Fantastic. So your technology must be very difficult to sell. I only understood about 50% of what you said, That's pretty
Dave Michels:high for Evan, you got to double digit that's pretty good.
Evan Kirstel:What is your sales motion and your go to market when it comes to sales and new customer acquisition?
Philipp Heltewig:See, it's interesting, your question is connected to the previous two questions. Our best prospect is a customer who's been dabbling around with Google's IBM, so Microsoft technology for two years, they have the ROI models already. They know how conversational AI can help them, they just unable to realize that with the technology they're using, then they go to market, evaluate different platforms eventually pick us and then actually get to that ROI that they had calculated for their conversational AI initiative in the first place. So we call them down funnel prospects, prospects that already know what conversational AI is, how it can help them they know what their customers are calling about. So quite educated customer, right, that is the best prospect. But then of course, we also have customers where they're doing their first conversational AI project with Cognetti. But then those are different discussions. That's when we actually have to cut also through all the marketing crap that's out there and explain, Hey, okay, this is what you can actually get out of it, we're not going to tell you, this is what you can get out of, and we were describing, like a Hollywood movie, when that's not true. So that is the way we do that. It's a consultative selling approach. We go in no bullshit, right? I hope we can say that here on the on the podcast. salutely. Yeah, so a no bullshit approach. We tell it how it is and what you can do with it. And we very quickly get to a small pilot, sometimes in the first meeting with a customer, we already built something in the platform. And then we do a bit a little bit of a more involved pilot of maybe a couple of weeks, sometimes even if it's a really large company a couple of months where we integrate into the back end systems, and we prove the value. And then we move on from there. And we actually have close to 100% conversion rates on the pilots to paying customers afterwards. But it's really proof the value and don't bullshit, right? Don't talk about things that you can't do. And that is, I guess, a trick, at least for us.
Evan Kirstel:Fantastic. And just looking at your LinkedIn profile, you appear German, come to our house Deutschland, to this Deutsche order. Yeah, it's been passed. Okay. Fast kind accent. Yeah.
Philipp Heltewig:I'm actually, I've lived in Australia for a very long time. So I'm proud to say that I'm German and Australian, I have both passports.
Evan Kirstel:That is a very unique thing to have. So well done. Thank you. I think I think you and Dave also went to Harvard. Is that right? That's
Dave Michels:right. I saw that you had accelerated education there. I did the I think it was like a two or three week program at Harvard. I did it before you though I would much wiser than you. I didn't like 15 years earlier.
Philipp Heltewig:Yeah. So you had a much longer time to benefit from their program and build on it. Well, it'll
Dave Michels:be I'm running my own podcast now.
Philipp Heltewig:I know. So maybe one day, I can do that, too.
Dave Michels:All right. Well, I want to thank you very much for joining us today. It's been a great conversation and can't wait to hear all the thought that this podcast generates.
Philipp Heltewig:Perfect. Yeah. Thanks for having me. And it was great talking and let's speak again in the future.
Evan Kirstel:Well, that was a great conversation with Philip. I learned a lot actually from you as well. So look forward to re listening to that episode. And we have a big guest from a big company coming up next. Who is it? Who is it? It starts with a V. V for Verizon, so we're gonna learn about what Verizon is up to.
Dave Michels:Oh, the evil empire? Yes. Yeah, Alex Doyle. You know, I've known Alex for a long time I met him at broadsoft He was good till but he's had a really long and prosperous run at Verizon then running the UCAS and see how stuff should be a really interesting conversation. Look forward to the conversation man gotta get out of the phone if your phone no man knows me