#256 Artem Koren's Vision: AI, the Future Workplace and Beyond

0:00:01 - Mehmet
Hello, and we'll come back to a new episode of the CTO show with Mehmet. Today. I'm very pleased joining me Artem Koren. Artem, the way I like to do it is I keep it to my guests to introduce themselves, so the floor is yours. 

0:00:13 - Artem
Excellent, great to be here, mehmet. My name is Artem Koren. I'm the co-founder chief product officer of Sembly AI. That was founded in 2019 with my co-founder, Gil McLeod. My background is in technology, product management and management consulting, so I've been on both sides of the fence in terms of larger organizations and as well as startups and what Sembly AI does. It's an AI teammate, also sometimes known as an AI meeting assistant, that attends meetings as a colleague would and then gives you a lot of really great insights after the meeting. 

0:00:53 - Mehmet
Great. Thank you very much again, artem, for being here today. So what I always wonder about, because there is a story behind every startup. So what was the story behind Sembly AI? What was the opportunity that you have seen, the problem that you've seen you need to solve with your co-founder, and how did things work with you? 

0:01:18 - Artem
Sure, so being that, my background is in management consulting, and so is Gil's, so Gil was the CEO of UMT Consulting Group that was eventually acquired by Ernest and Young. We spend a lot of our time in meetings as a consultant. In fact, meetings is what drive the consulting practice. It's meeting after meeting, after meeting, and by getting people together and making moves forward in different aspects of an engagement, that's how you achieve results. 

And one thing that we noticed when we were getting together this was 2018, 2019, is that there's a lot of technology to facilitate the actual happenstance of the meeting, meaning how you connect and talk to people, so video conferencing, audio, noise cancellation, that kind of stuff but there really was no technology that could take advantage of all the valuable information that happens inside the meeting, so nothing was really understanding what was happening. You have the meeting. After the meeting is done, the meeting evaporates away, and it's whatever each person carries away from that meeting that will hopefully make some actions happen, and different organizations, different teams and different individuals are better at that and worse of that, as it happens. And so what we thought was if you could introduce an AI into the meeting that could understand what was going on, the opportunity for value creation there would be tremendous, and that's the idea with which we started Sembly AI. 

0:02:47 - Mehmet
OK, great that's. 

You know, like I can resonate a lot of things to myself because you know I was also working on, you know, technology consultancy for a long time and, yeah, usually things goes ugly, I would say, after the meeting, so we forget who said what and you know what are the next steps and so on. So, yeah, like I can relate a lot to this. Now, what I noticed, like you know, you believe that you know it's not just a tool, it's something that can enrich you know us as humans and create positive change. So, if you want to, you know, a little bit elaborate on this point, like how AI really can enrich us, how it can create a positive change. And the reason I'm asking this item is, you know, in the news we see a lot of you know these voices that say AI going to take my job, ai would do this, ai would do that, and we see a lot of negative thoughts. But I know, like, actually, it can be on the opposite side and it's nice to hear that from someone who's expert in the domain, like yourself. 

0:04:01 - Artem
A lot of people are concerned, and validly so. There's voices, very knowledgeable, very educated voices, that give very different perspectives today. Some are a little bit more alarmed, some are very optimistic, and it's a we're in a bit of a chaos in the scientific definition of that term, in the sense that we're going through a phase transition in technology and the path forward is not so clear. We're certainly experiencing a transformation in the kinds of technologies that will become available in the future that we haven't really thought about before in a very paradigm shifting way, and certainly certainly that creates a lot of concern. What I can say is that first, just a few conceptual highlights, let's say, is that I think of AI as a new kind of material that allows for new kinds of technologies to come to the fore. 

0:05:15 - Mehmet
So, AI. 

0:05:16 - Artem
It's not in and of itself a tool and it's not in and of itself a tech. It's really a kind of a material. So, just like web-based development became a material that enabled online applications, high bandwidth mobile networks enabled new kinds of applications, so it's a material in that way, like carbon fiber that enabled planes to travel longer, and so using this material, you can build all new kinds of experiences that you couldn't before. Imagine if we didn't have plastic before and suddenly we had plastic right. So suddenly all of these new form factors are possible. All these new environments where technology can participate and products can participate are possible. 

With AI is the same thing, and so one thing that's important to understand is that there's elements of AI that can make existing technologies smarter, better, faster, and that's already happening. That's been happening for more than a decade, really, in all sorts of different areas creative, technical, data, analytic, I mean, you name it. So there's that aspect. Then there's aspects of AI technology that can make new kinds of tools available, but still in the classic sense. So you know, like essentially a better toaster, let's say right, so like a toaster that can talk to you, or something like that. But then there's this third layer of AI that can make absolutely brand new experience as possible that were never before, this kind of product, that didn't exist before, and a good example of that is ChatGPT. 

So, ChatGPT is unlike any kind of bot or chat that you've had ever before. It delivers the kinds of answers, the kinds of knowledge, the kinds of integrated interactivity that's never really been possible, so hyper-customized. Okay. So that's a very long prelude to the question of you know, is this all gonna take our job? I think the consensus is no. It's going to certainly change a lot of jobs and it's going to create a lot of new jobs. So it's a pyramid expanding material. It's not gonna shave parts of the pyramid completely, but it will change. 

Just like over the past couple of decades, what it means to be a designer has changed. What it means to be a developer has changed. There's new programming languages, new paradigms, new facilities like DevOps and all sorts of frameworks, platforms like GitLab and GitHub, which didn't exist 20 years ago even though they could have been used so just in that way. And so now there's like a DevOps role on the team or whole teams of DevOps. That's a concept that didn't exist 20 years ago, but it doesn't mean that we no longer need programmers or we no longer need cloud admins or system admins or engineers. So I think what I foresee is that in some one-off cases, maybe jobs will go away, but I think those are gonna be very exception. 

I think most jobs will be impacted in some way by availability of new kinds of tools and products that are AI enabled, and they will shift those jobs, meaning that the skill sets and the kinds of activities that those jobs will require will shift and will allow people to be more productive and in more powerful ways, and then it will create a slew of new jobs, a slew of new. 

You already have this. You have this new idea of a prompt engineer. This is a new job that got born over the past one year. Right, and that's just touching the surface, and I think there's gonna be many, many new kinds of both engineering and non-engineering jobs that will be created going forward. And there's also like a silver lining with AI that I think will be very important, that I think a lot of people don't think about, which is for the past century or maybe even a little bit more. It's really the scientists, engineers and mathematicians or the STEM fields that had essentially a monopoly of the progress of civilization, so you have to be a scientist, engineer or mathematician or related chemist right To make significant progress. 

To do what? To create new materials, to create new technologies and fabrics, both material and the material. To create new, enable new kinds of products, and then to actually create those products and make them work. So the airplane industry, the whole internet industry, everything nuclear, solar I mean everything, everything we use and enjoy today in our modern society, or at least those countries that are privileged to come from the STEMs. And I think, something that's very seldom talked about today, which makes me think either I'm crazy or I just haven't heard it yet. 

But is that what AI makes possible? It really changes what kind of a mindset and what kind of a almost a personality you need to create productive change in society. It's no longer STEM with AI, it's the person who can most effectively communicate to the AI what is needed and how it can be applied that will be the most powerful creator of new things, because now AI abstracts away a lot of this STEM detail and it can generate applications, sites, compounds, everything. You just have to be able to talk to it. And so STEMs are historically have been maybe unfairly, but I think in some cases fairly branded as not very social of all of the people in the world. 

I'm sure that's not true for everyone, maybe not true for the two of us in this podcast perhaps, but in a lot of cases I think that's it's a stereotype, but I think in some cases it's a fair stereotype. I mean, famously, newton never left his house right and he's brilliant right, but never left his house. So this new century that's coming up is a century for those who are effective communicators, meaning that they can translate the idea and in their mind with communication to the AI, so that the AI can create it for them in the real world. And I think that's a very fundamental shift that will empower a lot of people who haven't been empowered before. 

0:12:34 - Mehmet
Yeah, great insight, artem and I think a couple of days back even, we've discussed this on the show a lot of times and I used to ask this question are we going to still have jobs? And I heard Elon Musk saying that he believed that you know, we'll reach a phase where we don't need to work. Maybe there will be kind of a universal income system, something similar to that. Of course, we need to wait and see what will happen in the future. 

Now you mentioned chat GPT and you started it, of course, like open AI, they were working for a long time and you know, usually they used to have their APIs. So, and I know like and we will deep dive a little bit in what technologies you use at Simlyi but, as you mentioned, for everyone now, they are concerned that chat GP or, let's say, open AI, might you know, go after every single use case, so they will not leave place to start apps actually to play. Do you have this? I don't like to call it fear, but I mean like concern that you know the major large language model providers, they will put themselves into every single use case, so there will be no space for other startups to play. 

0:13:59 - Artem
In this area. I'm more of a pragmatist and I believe that if there's a technology, that, or a product that solves problems really, really well, that's a good thing. Whether that leaves space for others or not, that's kind of irrelevant. If I have I don't know the best quality hamburger or whatever right Like, should I be sad that others can't make a hamburger as good as I can? I think I mean not the perfect analogy by any stretch, but I'm not. That's not a concern for me, right? I think that open AI has created, has arrived at the potential of a new kind of a platform that can enable all kinds of new products and experience all on its own. I think that's great. I think the long term, people benefit by having the best possible technology do the things that they need to do. Now, is it true that open AI will just be like the only software in the world? No, I definitely don't think so, for a few reasons. So there's kind of two angles to look at this. One is the closed source open source angle. So, just like OSes, there's a few commercial OSes and then there's a lot of open source OSes over the past couple of decades. So we've shown that open source can work. Maybe it's not the dominant or not number one although in servers today it is but there's certainly a place for commercial closed offerings and then competing open offerings and there's different dynamics of open source offerings that make them very attractive in certain markets, certain news case and certain price points. So I think that's one angle of competition. There's going to be, I think, a handful of really great commercial foundational LLMs and then a bunch of open source ones that are also really great for other reasons. So that's one. 

The other thing is that there is something important to be said about practicality. You can have the best airplane engine in the world, but if your I don't know in-flight meal is really terrible, people will go to a place like people who have a strong preference for in-flight good food will go to an airline that has maybe 1% worse engine but 10 times better food. So that whenever you touch reality with any kind of technology, but especially with AI, things start to become a matter of a thousand little details. That's my Michelin restaurant analogy. Why are Michelin restaurants so coveted? Is it because they're the only ones that can make a Kobe steak? They can get a Kobe steak in a lot of places, but they're the ones that when you come in, it smells amazing, the service is incredible, the wait staff is on your. Every women want Everything in sight stilistically makes sense the way the menu is presented, the way the accoutrement around the Kobe steak it's a thousand little details and people will pay a lot of money to have a Kobe steak in the Michelin place than in the regular place. I can buy a Kobe steak and prepare it at home and I think it's the same thing. 

So when you touch reality subtly, many, many details are important and you can't be a universal generalist. In an end, account for the thousand details that makes sense in every situation. And so the application of technology and specifically the application of AI because it's so damn mallible is so, so important, and Chagapiti, I think, is going to have a field day. I think they found something really incredible and they have an advantage this year. We'll see how next year pans out. They have a lot of strong competition by the ankles. We saw Gragh come out just now but I think that it by no means do. I think it's going to be like the only and kind of gubble up the opportunity for new startups. There will be plenty opportunities for startups, entrepreneurs, especially for the smaller ones in the applied areas, when you have to touch reality. 

0:18:41 - Mehmet
Yeah, I agree with you because you know, like, despite, for example, we saw some not related to AI, but some examples you mentioned. The OSS search engine is another use case, you know, like it was not only dominated by Google, for example, like there are some other alternatives to AI. In the whole time, it's I mean, you know, and the thing which I'm seeing, especially if I can I'm not affiliated with them, but what I can see, the way they are doing it, and I can understand, because they have Microsoft backing up them. Microsoft, they are very famous of having this ecosystem and they allow a lot of other players to be in that space, and this is where I'm seeing things going on, which is great. 

Now, coming back to somebody, I am back and this is something I want you to a little bit explain it to us. So, because you deal a lot with natural language processing and with with speech recognition, because and this is a fundamental part of AI and unfortunately, or fortunately, you don't know Now when you talk to anyone and you say, hey, I have an AI product, their mind goes to charge you PT. You know, and you know I want to highlight again that this is not only charge you PT. There are a lot of other domains under AI, like vision and so on. So, if you can, you know, a little bit deep dive, I would say, in the technology to use with simply AI. 

0:20:04 - Artem
So Sembly AI turned out to be a really annoyingly difficult stack of technologies that you have to put all together for the experience to be very compelling. And so we've been at it, you know, for almost five years now. And there's there's a number of things. Well, there's certain core things, right. So one is the ability to capture the meeting, and that's a tricky animal by itself because just recording a meeting is often not enough, because, for many reasons, one is all sorts of privacy, practicality reasons, like how do you tell it's being recorded? Do you have to press the button every time? What if you forget to press the button? It's a very kind of legacy way of thinking about capturing content. This is like record the audio way. I think it's very antiquate. You're also missing a huge amount of metadata as far as what's happening in the context of a meeting, who's talking, who's not talking, what the name of the people are, the video, potentially. So there's a lot of stuff you're missing when you're just hitting record button. So just the fact how you capture, take into account practicality, like I don't want to press the record button every time, I want everyone to be aware that they're being recorded. I want to capture all the meta. All of that stuff gets very nuanced, and so how you capture is one part of our tech. So we went about it by creating a virtual attendee who actually joins a meeting just like a human does. So assembly is your teammate that comes to your meeting when you invite it, or him or her, however you prefer to think about something, and they assembly joins as a participant would. You can mute it, you can kick it off the meeting, you could invite it back whatever you want. So that's one aspect of assembly that's pretty unique and allows us to support multiple platforms. We show up in Microsoft Teams and Google Meet and Zoom just like another attendee, and so this whole mechanic of I call it attendance ops, this whole mechanic of well, how do you get AI in your meeting, becomes just a very natural extension of, just like you would any other human being, you can add it to an email invite, you can add it to, you can sync it with your calendar, or you can tell, hey, I have a meeting going on right now, come on, you can do all those things, okay, so that's, that's one piece of tech. 

Once you've captured the audio and video and the meta, you have to convert it into a text that computers can understand. Here is a lot of more. There's more interest now. So multiple languages. So somebody, for example, supports over 35. 

The interesting thing is, in a lot of meetings today, especially global teams, there's mixes of languages happening. So there's like a little bit of English, a little bit of Spanish, a little bit of English, a little bit of French, and that's hard for traditional speech recognition because you kind of have to tell it what the language is that it's trying to transcribe. So that's something that assembly had to figure out as well, and we support mixed language meetings so you can speak a little bit of English and a little bit of French, and assembly will be okay with that. The next thing is, of course, the high quality transcription itself, and that's an area that's been a great leap over the past year or two, and I think, both in terms of quality and in terms of cost performance. And so I am I'm fairly confident that over the next year the cost performance of the most common languages will probably go down by factors and it's already gone down by factors, which means that very quickly you can get a very high quality transcription, and part of that reason is that is also the large language model or transformer based technologies that are getting very, very good at figuring out what you're likely trying to say versus handling one word at a time, and we had this very interesting example with our partner. So we're partnered with Phillips Phillips speech. They manufacture both conference room and personal devices with audio recording and many of those devices now are powered by, are kind of partnered with assembly and they have a very high quality logo on the box so you can get your results from those devices into assembly very easily. 

And initially, when we and this has been a great ongoing partnership for over two years now but when we started, phillips speech comes from a history of dictation and when they were testing assembly, you know that the way they did it was they basically stood around like a big room around the microphone and they would like yell out random words and then like for a few minutes, you know from different locations, right, and they'll say cat and then they'll say window and then they'll say like, and then they go back and they look how assembly did and assembly didn't do very well. It's like it's basically gave like semi gibberish. You know like it's basically. And so I talked to one of the leaders at Phillips around this and he's like, look, I don't know what's going on. We tested it and it's just giving us weird you know weird answers. And I'm like, well, tell me how you test it again so we can go to the window. You know carrots, whatever, right. 

So assembly is about the conversation. It performs well in a team meeting environment and they're running word thrown it. So assembly really understands, understands in a conversational context, and the more of that conversational context that has, the better job it will do at understanding what you're trying to say. And that's the power of the more of the more novel asr or a tab, automatic speech recognition engines. Is this contextual power? It's very different from the dictation approach of you know, like dragon, naturally, or whatever. You have a nuance where each word was well trained and then you're trying to figure out the word. That's no thing, that's done. You cannot achieve a certain level of quality in speech by doing it that way. So that was a big shift for them. They would never thought about that before. 

And then, of course, when they tried whole sentences, their results were a lot better. And then when they tried whole conversations where people are actually talking to each other in a reasonable way, their results were very good. So so that's so. That's another part of transcribing. So once you've transcribed, you've handled the language thing, the multi language thing. You have to figure out who is speaking. That's also very hard. 

When we started Most services Google, amazon you had to tell them the number of speakers on the call. We don't know the number of speakers were not there, that you know. Assembly is there, so very rudimentary, and even when you tell them the number, the results were a far cry from something that was really usable. So that it's called diarization, where you try to figure out, attribute the speech to the different people on the call and ideally also figure out who those people were. So that's just person, one person to someone. 

Assembly started we mostly had speaker one speaker to, because we don't know who you are, but we knew like different people. Now you go to assembly and you'll see, you know named people talking. So that's, that's another big deal. So that's the. So that's the second piece of our technology, which is the ASR kind of converting audio into into text, and once that's done then it's fed to a family of services that go to work on the actual content and they can produce extremely high quality meeting notes, produce the best quality task identification that's available in the street today a bunch of meeting insights like issues, risks, decisions and so on, and it can send all the insights, like tasks and meeting notes, downstream to applications. 

So once the comments come in, they're formed into your meeting result a few minutes after the meeting is over, usually, and then those insights can be automatically or manually pushed into your other applications. For example, like I have tasks that automatically go into my to do is after my meeting. I don't have to do anything. It's just assembly who figures out one of the tasks for me, figures out that I'm the assignee and also the due date and things like that, and it will push them directly into to into my to do that. So it's fully automated in that way. And then, finally, there's a piece of our product that's called Semblian, which is a very popular idea these days across a series of products, where you can talk to the meeting with an AI chatbot. 

So, you can, you know, ask questions about the meeting Like what's you know, what was our terms, like main purpose for being in this meeting, or something right, or what were my main points, or what was the summary, whatever, or. But you can also ask it to generate content so you can say, you know like generate the agenda for the next meeting, or suggest you know like ways that we can more effectively run future meetings, or write me an email summarizing like what happened in this. So we'll do that for you as well, so it actually like generate things and save you some time and actually writing stuff post meeting. So those are some of the kind of core experiences and core technologies, that assembly process. 

0:30:40 - Mehmet
Yeah, great, and thank you for this detailed explanation, Artem. Like you know, it's just like very complex. I mean, it's not like as people things, it's just, you know, a chat box and then you ask it some questions. Like you, you went over all these ones, so you know. One thing, you know, which I wonder about is you mentioned, you know, time, reducing the time, you know, and helping teams in, you know, getting better insights, but where do you think you really we can take this to? Like what? If I mean another in other way, if I want to ask you, like, what is the ultimate vision? You know what, what more we can achieve with your technology? And they are in general, when it comes, you know, to the traditional way of doing office work, and you know, especially now we have, as you said, people, you know, spread across. So so how this, you know, intersect also with with the whole idea of future of work environment. 

0:31:51 - Artem
The long term vision here is that AI has the power to make organizations come alive, and what I mean by that is how do we know that something is alive or not? So, for example, I poke a rock. It doesn't do anything back. It's not really a lie. It doesn't respond to its environment, and modern companies of size, and global enterprises especially, are more like the rock then like I don't know, pick any animal, like like a hamster, I don't know, or fish even a fish right where you're like you touch a fish, it's a god, it's out of there, even if you're like in the water, it's gone right, it knows, it knows what's going on in the environment and it responds. 

Today, enterprises can do that. There's a in order for companies to be responsive. It's a Herculean effort to organize roles, processes, tools for it to digest all of the different inputs that are happening and these are both, you know, from its customers, also from its competitors, also from new technology bases, also from compliance, also from legal, also from some kind of internal things that you know that need to be reworked or added on, and so on and so on, and these all have to be digested into a structure that then can be decided upon for a strategy. Now, this process in companies is called strategic alignment, strategic assessments, project portfolio management, different things in different places, but ultimately a company is charged with bringing all of those details together and then figuring out you know what's our plan for the next year, sometimes even longer. If a company can do this effectively once a year, that's considered a huge win, and it's very expensive and very difficult to do that. You usually would involve some kind of an expert consultant. See one, you know a big four like a why, or or or, I don't know even my GSOP cube. 

Let's say, if you're more in the PMO space and what AI allows you now to do is imagine if you have an assembly teammate hanging out with every team on all their meetings. So now suddenly you're building this awareness of what each team is up to and each team and each team member has this kind of a digital trail of all their activities. So things they're doing, things, they're talking about things, they're working on the statuses of those things. Now you're building kind of a local AI with each team, but then the teams roll up to a departmental teams, talk across departments. So imagine there's kind of a little node. So you're starting to build this octopus of a, of a technology, this network that can all roll up into the executive suite and give you a near real time information on what's really happening in your company and it can go up to give you a really great understanding of what's actually happening across all different slices of your organization and autonomously percolate up the more salient and the more impactful things or the more risky things, let's say. But then you can also bring that down where there's an idea of, like what direction is high priority and more valuable, or how we want to distribute our resources towards different value directions, and then, at the local nodes, align teams to support those strategies much more quickly than is possible today. And so this kind of creates this alive organization that can, bob and we, even respond to a global environment Very, very quickly. 

This is unprecedented. This doesn't this. This, no company can do this today, still 2023. Not possible. It's only possible when you have this AI awareness that can first of all be the sensorial portion of the organization, meaning like absorb the things that are happening, make sense of them, aggregate them intelligently, surface them intelligently to the management levels and the executive levels and then bring also down the alignment perspective from those layers all the way down to the teams. You need that. You need AI for this. Without AI, it's not possible to do. 

0:36:51 - Mehmet
Yeah, it's a great insight and you know I know it's maybe short answer can can be enough, but from what you're mentioning, are them to me like any enterprise, or even even small business, who neglect, you know, the factor of AI? They will be, they will be out of business. Do you agree with me? 

0:37:11 - Artem
With few exceptions, yeah. So like the only places I think that don't need AI for the most part, let's say our craftsmanship businesses, where the the the value is in the craft. So if I'm a glassblower, if I'm a wood cutter or you know, there's craftsmanship industries, Certain kinds of art. I'm not saying that AI won't participate in those industries, but then there's always going to be a craftsmanship pocket where it's actually the point that there's no technology. That's kind of the point. But other than those, yes, I think AI will be pervasive, pretty much everyone. 

0:37:56 - Mehmet
And you know I'm keep talking about it and saying, of course, anything that related to the areas you mentioned craftsmanship and ours, like guys, you need at least to know what's happening and you know so you can plan how you can adopt AI in your environment. And again, I repeat this one more time, it's not only just GPT. You know you need to need to go through Lee and deeply understanding what they are can do for you. And you know just moments ago what you explained about natural language understanding as our, and so on. So these are like few examples of the big picture. Now shifting little bit gears here, like I'm talking about you know being a startup and a chief product officer at a startup, like how, how this you know is completely different from you know being and because you were in in the big consulting firms also as well. So how this is different experience than being in a more established environment and doing things on on your own pace. So how that experience was different for you. 

0:39:07 - Artem
In some sense completely different. To some extent, you know, when I've thought about my experience in the past, when you're a lead on an engagement and consulting, it's kind of like a little bit of a startup. But context it to an organization that you're consulting Because you're starting usually from blank sheet of paper, which means you need to figure out all the people involved on both sides. You need to understand the goals, the perspectives, the timelines, the constraints and you kind of need to figure out the environment deeply and then you need to produce something effective and useful in that environment. So in some sense, leading a consulting project is a little bit of a startup. But there's a lot of many differences, which means that your budget is already figured out so you don't have to worry about capital. That's a big deal. You're most usually, hopefully you're well-resourced, meaning you've planned the project and you have the resources you need to do it. Your endpoints are very clear. So there's an entrepreneurial kind of a startup aspect to those. 

But a lot of things that are variable in a startup are fixed in place in a large company. When you're in a startup, I think the one universal of a startup and this is gonna be extremely cliche, but I'll say it anyway is the lack of any constants. There are no constants, there are no templates, and it's always. It always tickles me when people give generic startup advice, because I can immediately think of like 100 ways that that would not work. And if there was generic startup advice, then that really worked and all the VCs would be tremendously rich. It would never fail with their investments, so, and the VCs are as expert as it gets when it comes to startups, and yet very few VCs are batting at a high percentage. 

So the difference so one of the key things in a startup is, first of all, you're building something that you don't yet know what it is. You have a like a cloudy idea. It's a lot more quantum. You have like a cloudy idea of this thing that's gonna work, but you don't have like a very clear idea and the idea actually becomes reality by virtue of you making it. As it touches users, developers, competitors, it starts to kind of shape different pieces of it. So the kind of this cloud in your imagination, parts would become more real. So that's one thing is that when you're starting, you don't really know exactly yet what you're building until it touches everyone, and this is, I should add that this is more true for the higher innovation products than more cloud, right, If you're like building a pencil, maybe that's not doesn't apply as much, although you probably need some kind of innovation perspective there as well to be successful. But for high innovation products, those of all the AI, for example, that's very true is that you don't really know what you're doing. You just have a directional idea and it's kind of like these print. You know there's these printers, that these 3D printers, and when the laser touches the suspension it creates a solid. I think that's actually a very great analogy for how an innovative startup produces a product. Every time the laser touches it, every time you have, like, a reality touch point. And that reality touch point can be in different ways it could. Sometimes it's a budget constraint. That's a reality touch point. Sometimes it's a technological constraint, right. Sometimes it's a user feedback. So all those reality touch points shape the product into existence. So I think that's one interesting bit of a startup that's very different than from a large working in a large company. 

And the next one, of course, is the capital. Capital is the live blood of any startup. It's not the most important thing, but it's fairly close to oxygen. That's not my quote, but yeah, so you can say, oh yeah, capital that's not, and I think VCs love to say that capital is not that important, so don't worry about that. Yeah, okay, fine, true, but almost like oxygen, so it's not that important. But you need it to breathe and your capital determines a lot. It determines what you can offer when, how good it is to what kind of customers and what localities, and how long you can work to find your product market fit before, before you have to close shop or be successful, capital is a very important part of the equation. Managing that capital both getting it in and also managing it to spend is a very important aspect that you never have to think about when you're in a corporate. 

0:44:46 - Mehmet
You know, I'm very happy because you started by saying being a consultant and I mentioned this, I even wrote a small article about it a couple of weeks back. I said people ask me why you are so big believer that consultants are entrepreneurs. I said, yeah, because I did this job. I was also a sales engineer and for me, it's a consultancy job also as well. They said there you have a problem to solve, you have stakeholders, as you said, to deal with and you have a product to go and try to fit it to that specific problem. And you need to keep going and trying to fit whatever is the product service you're trying to pitch. At the end of the day, you're pitching something to the client. 

And I said, yeah, because there's no one like an SE or like a technology consultant or a consultant in general, that he's an entrepreneur, as you said. The only difference is the product is made for him or her. So all what they need to do is just to fit that piece of the logo inside that empty, empty slot let's call it in the customer's environment that they are working with. And this is why I consider myself entrepreneur by nature, because I did consultancy all my life. Now, one important aspect. You talked a lot about capital and about VCs, but I think you had a view also on alternative ways. Today's capital right. 

0:46:27 - Artem
Right. So, yeah, I don't want to mislead, to say VCs is the only route. In fact, we are not VC backed assembly is not VC backed, so but VCs are a big part of the capital process. They are, I guess, the investor of first resort for a startup of at some point right, like once you get beyond the very, very initial steps of angel and friends of family, right, but it's certainly not the only way to raise. There's now, first of all, there's public raise platforms. They're also like investment offices that are not VC's, that have capital to invest and then there is companies and industrial capital that gets invested. So VCs are not the only game in town, they are the most prominent game in town. 

I think that VCs are professional investors. So what does that mean? What differentiates a professional investor from a non-professional investor? Well, it's the same thing as a professional poker player. What does that mean? 

So a poker player has a very specific scheme, mathematical scheme by which they play the game. So what does that mean? So a poker player has a specific scheme by which they play the game. Even if it's like something like looks really juicy on the board, they're gonna run their math and they're gonna make a math decision, let's say, for the most part, if you're a really professional player, because for a professional player it's the long-term that makes a difference, and the long-term is making the best bet and fold decision, and those decisions are mathematically governed you have to be able to read your chances really, really accurately. On the poker board, vcs are the same, which means they have a very specific set of criteria that tells them whether they make the better dough. And I know that criteria can be varied, but only very, very slightly. Just like a poker player, they'll vary from the math, but very, very slightly. Generally they're gonna stick to the math and if something is very non doesn't agree with the math by a lot, that's a very difficult investment decision for a VC and usually it won't happen. So, for example, vcs will generally invest on the core stats that they're looking for versus some kind of a subjective conviction, and because of that it's much more difficult for startups that don't effectively speak to that complex equation that the VCs run to raise capital with VCs. 

For example, in certain rounds VCs are very driven by your recurring revenue. You have to have like a certain, almost like a watermark of recurring revenue for VC to consider the investment. You can have all kinds of other factors, but generally that would be a factor In earlier rounds the VCs are very driven by do you have something strongly protected, like, do you have the Coke formula, coca-cola, not the other one? So if you don't have, like, a Coca-Cola formula, they're gonna have a tough time like investing in you because they have to manage their risk and they have all kinds of ways to hedge that risk. But for earlier startups it's something that's very clearly differentiated and protective. For later startups it's the demonstration of scalable business. And if yeah, and again, this is not set in stone, and this is where all the VCs get up in arms and start coming back on Artemis Crazy, but that's been my experience and it makes sense because VCs are professional investors, so they make strict bets that follow the formula and they know that in the long-term the formula will lead them to financial success. I get it. Some are a little more flexible, some are a little less, but generally they'll stick to those guns. When you don't fit the formula as a startup and many startups don't that's when things get interesting, because there's no playbook outside of the formula today for raising capital. 

And so what do you do? And there are things you can do. So, for example, there's industrial investors. Like most large companies have funds, like American Express, let's say, comcast or another company. They have funds where they'll invest in companies that are then gentle or, let's say, revenue expanding for their business. Google has an investment arm, microsoft has an investment arm. So if that company thinks that your product can help them to expand their market, expand their revenue, they would consider investing in you, even if you don't meet the more strict VC criteria. So that's another potential alternative for investment. They see something there that a standard VC would not. And then there's also public markets now for raising startup capital. 

And then there's angel investors, and there there's also a lot of opportunity for capital raise, because angels and the public markets, they're more around personal conviction than specific formulas for the most part. And so there, if they think your idea is brilliant, they'll invest, even though maybe you don't have patents, even though maybe you haven't really sold that much, even though that, that that right. But if they think the idea is brilliant, they will often invest in the idea and that could be your investor base for a certain kind of startup and a certain kind of a entrepreneur. And I'm not even talking about all the other stuff like. There's all sorts of auxiliary benefit investors, like ecological, climate, social, you know low, you know low access, low income, different geographies. So there's funds that will invest in you because you're doing like you're providing water to places where water isn't accessible in Africa. 

There might be a not, it might be. There's definitely funds that will invest in you just because you're doing that. So the C's are the main game. But then there's a lot of sources of capital raise. There's government, by the way, government investors and investment in startups for certain kinds of activities. So there's other venues as just VCs. I guess is the most popular most often. 

0:54:10 - Mehmet
Yeah, great insight, artem. Actually, and you know, sometimes because I do this now part of what I do. So people come and the first question I ask any founder or founders? 

have you tried to bootstrap your startup, you know why you need the fund in the first place. All right, so before going to AVC, and yeah, like the alternatives are so much more now and I think, and I believe you know, if they do it the right way, they would be discovered and actually the VCs will come after them. They don't have to go after the VC, especially if, as you said, if they are in a very it doesn't have to be related to tech. Very nice example you gave about you know water for you know clean water for countries like in Africa or so on. So, yeah, like a lot of options over there, and they say, okay, don't forget, like it's like kind of a marriage. 

One of my guests he mentioned this. He said like you're gonna marry you know in a fictional way, of course with that VC because he will be a shareholder or she will be a shareholder, and you know you need to start to discuss the. You know what the market is reacting to what you're doing. So it's a whole other discussion, artem, like almost we come to an end. Is there anything? Any final thoughts you want to leave, to leave us with today? 

0:55:39 - Artem
It's a great conversation. I guess, if you know, a lot of what we say depends on what horizon we're looking at. So you mentioned Elon Musk. He said you know, one day, you know, maybe we'll just need a basic income and technology will do everything, and maybe that's true, but I think that's more of a start-track horizon, like a few centuries away. I think for the next century, people will still need to do stuff and earn. Specifically today, the transformation is afoot and I did. 

I was invited to speak at a conference in Stockholm earlier this year where I did a poll and I said how many of you have had an AI participant in your meetings? And maybe like 5% of people raise their hand Like almost no one. And that was so interesting and so telling to me because I told them look, by the time this year is over, probably at least a third to a half of you we will encounter an AI participant in your meetings. But in a couple of years, it will be more often than not that AI participants will be in meetings, and in a few years, if you have a meeting without an AI participant, something is going, something has gone terribly wrong. And so there's this changes on their way, because it's just so valuable to have AI in your call with you and you get so used to it, it becomes like an extra appendage. 

So like right now, when I have a meeting without my AI teammate there, without assembly there, it's like very unnerving because I know that after the meeting, like I'm not gonna have meeting notes, I'm not gonna be able to like look up later, I'm gonna miss some things. I'm gonna try to remember, like did he say 14% or 16%? And I know assembly gives me all that without all the meetings that it attends with me, and so I've grown to be very reliant on assembly and I think that's gonna be the normal thing. We'll start to grow reliant, just like you rely on your phone. Now, imagine if you don't have a phone like with you, right, suddenly it's like very unnerving. 

You can check every five minutes. It's the same thing. It's the same thing, and so I guess the last, the parting message would be I encourage everyone to try adding AI to your workflow and to invite AI to your meetings and check that experience out. You can do that. Here comes promo. You can do that at our website, wwwassemblyai S-A-M-B-L-Y. That AI and there's a free trial to start and you can check it out. It's all unlimited. We don't charge you by hours or things like that. 

And try the professional, the team plan and see how cool it is and how you get pretty hooked on the idea. And then, if you wanna reach out to me directly, you can find me on LinkedIn Artem Coran K-O-R-E-N. And connect with me there. 

0:58:30 - Mehmet
Great Actually, artem, usually after I tell you ask my guest your final thoughts, I ask where we can find you, but you answered that to me. Thank you very much. So what I would be doing is, of course, your LinkedIn profile, if you don't mind, and, of course, assemblyai the URL will be in the show notes or anyone. And actually I agree with Artem, I advise you to explore the opportunity because and to your point as well, okay, I'm not affiliated with open AI, but because chat, gpt is what available for me. I'm hearing about the other ones. Even I use BART sometimes and, as you said, it became part of my daily life. So now if I need to check something, I go to AI because I get the answer much, much faster and I built some automations for myself also as well around that. So, yes, it's just part of our lives. 

Moving on, artem, really I enjoyed the conversation with you today. Plenty of information, whether about the technology itself, about the future of work and the collaboration and how assemblyai is helping organizations to get on the edge of the latest technology and to find meaning in their meetings and take action space on their meetings. So it's really much appreciated information you shared with us today and this is how I end usually my episode. So, for the ones who are tuning in the first time, thank you for joining and I hope you enjoyed this episode. For the loyal fan, thank you for your encouragement and your feedback that always make me more, I would say, energetic to do more of the podcast. And if you are interested in also joining me for a meaningful discussion, same as I had with Artem, then don't hesitate to reach out to me directly on LinkedIn or by email, and thank you very much for tuning in and see you soon. Bye-bye. 

Transcribed by https://podium.page