The AI Advantage: How Businesses Can Benefit From Artificial Intelligence (Episo...

Transcript

Jeff Bullas

00:00:03 - 00:01:53

Hi, everyone and welcome to the Jeff Bullas Show. Today I have with me, Artem Koren. Now, Artem is a Chief Product Officer and Co-Founder of Sembly AI, a software as a service platform that uses proprietary AI. And for those of you who haven't heard about AI, it's artificial intelligence as an acronym, algorithm to transcribe and analyze meetings. And here we are in a meeting, it's actually a virtual meeting. So we'll talk about how we can use Zoom and AI in the next part of what we're gonna chat about. Artem has 16+ years of product management experience and has applied AI learning to his work since 2009.

I hadn't heard about AI much before that. I just heard about social media back then. So, but obviously, Artem is a trailblazer and we look forward to hearing more about that. Artem grew up and lived most of his adult life in New York City. He's originally from Belarus. He graduated from Columbia University with a degree in Computer Science. He later attended the NYU Stern School of Business where he earned his MBA.

Artem's passion about using AI to enrich humanity and create positive change. He combines his experience as a systems engineer, product manager, IT executive, management consultant, entrepreneur to create innovative products, bringing to life new ways that technology can help us reach our goals. Sembly AI, which he co-founded with Gil Makleff in 2019, is at the forefront of advancing natural language understanding, paving the way for a symbiotic partnership between humans and technology. Artem is passionate about using AI to enrich humanity and create positive change.

Welcome to the show Artem, it's an absolute pleasure to have you dialing in from Netherlands and thank you very much.

Artem Koren

00:01:54 - 00:02:05

Hi, Jeff, excited to be here from the Netherlands, indeed from the cloudy and usually rainy city of Amsterdam, which I love. So I'm glad to be here.

Jeff Bullas

00:02:06 - 00:03:04

Yeah. Well, I do love it but most of the cities in Europe because they're walkable. You get out of your apartment, you walk down the road, you go to a cafe, you don't need to get in the car. Quite often you don't even need to call Uber. So I really love the European experience and I was there twice last year and in Europe and I love it. So I look forward to hearing it. So what I want to know about first that I'm curious about Artem, is number one, you obviously were interested in science and, you know, basically technology from, you know, at university. How did you get? So, let's go back a little bit to how did you get interested in technology and, you know, digital engineering and programming and all that? Where did that all start?

Artem Koren

00:03:06 - 00:06:56

So, my interest in technology dates all the way back to maybe fifth or sixth grade when I had my first exposure to personal computers and basic computer games at first. But on those computers, we had something called Q basics and my friends and I would try to write these programs to protect the screen and kind of hide what was going on behind the screen. Obviously, you love your kids, you want to hide things from your parents and this was done in Cuba and these were very, very rudimentary little script programs, but that's where it started. And I got interested enough. And I was also encouraged by my father to explore it further that I learned to program in sea. This was 7th, 8th grade or so. And I think that's where things kicked off because when you're a computer kid in those days, and this is the heyday of the early 90’s, you had all kinds of interesting privileges. So for example, your teachers would engage you to help them with their computer stuff. And then in high school you could even get pulled out of class because there's a network problem and there's only so many individuals in the entire high school who could even understand what the problem is, let alone try to solve it. And so it snowballed from there. And I also was part of that crowd that was into the early operating system. Linux stuff early, hacking. Not, not, nothing to sit in the area is just for fun.

And I even remember the first time I saw Google as a search engine before anyone has ever heard of Google as a search engine because really it was only the techie geeky kids that knew about it. So that's where it kicked off. And I got a job in programming while I was in high school. And when I went to the interview, the interviewer spoke to me, you know, answered old, older questions I got hired. And after that, the interviewer realized that I was still in high school. So it was an interesting journey up into tech. I did computer science in undergrad. I did study some AI at Columbia University. I have to say I was disappointed by what was thought to be AI at the time. Oh, yeah, but came back in a big way about 6 or 7 years later, when I had a chance encounter with a developer from young Lakos lab at NYU at the time. This was a person I knew from, we went to the same high school. He was a few years older and he put me in touch with people in the lab. And I think that's where my first exposure to the real taste of modern AI began and that was around 2009, 2010.

Jeff Bullas

00:06:56 - 00:07:08

Right. Ok. So at that time, yes, you were underwhelmed by AI initially, it was I suppose, rude and crude and very undeveloped.

Artem Koren

00:07:09 - 00:12:13

The idea of what AI was, and this is interesting because the definition, the concept of AI has shifted significantly over the last 15, 20 years, at least. When I studied at AI in undergrad, it was the AI of deep blue, the computer that beat Casper in that very famous IBM match in the late nineties. At that time, the thought of AI was a certain family of algorithms that could search all of the possibilities very quickly and pick a possibility. They like the best these algorithms would do that. The problem with that was, and this was obvious to me from day one, is that humans don't think that way.

That's not how we solve problems. We don't pick through every possibility and figure out if that possibility makes us happier or not. We have a different process, whatever that process is. And so when I saw that I was, I thought, this is certainly not artificial intelligence. This is a problem solving technique at best and it gets okay, results in a certain very narrow domain of problems, very narrow. Yeah, but it's nothing like reasoning and certainly nothing like human reasoning. So that was the early days of AI, these deep search based solutions that OK, they could play chess reasonably well. Um with a whole huge rule set on top of this kind of deep search, but they couldn't do much of anything else. So, for example, at the time, it was very, very difficult. I would say it truly didn't exist at a usable level to have a program that could tell you if there was a, a cap in a, in a photo, it just didn't, didn't, didn't exist. The state of the art was, you know, maybe something like 50-60% accuracy, something like that, very, very crude. Because there wasn't, there weren't wide application of neural net based algorithms at the time and the revolution started.

Let's say the modern wave of revolution started around the late 2000s, like Jan Lugo was one of the leaders in the field who re-engaged with an old idea of convolutional neural nets, which are these systems that in software approximate in a sense, human neurons. So there's individual nodes that process information and then they have something called weights that adjust as they learn. And it turns out that this framework, this architecture is very flexible, very malleable and ultimately can be trained to perform on a very broad set of tasks. So a very uh direct application, for example, was handwritten digit recognition for zip codes in mail addresses. And then it went on and on from there to recognize cats and recognize other things. And then eventually other kinds of architectures for AI were invented. But that's the two, that the late 2000s is roughly where it was reinvigorated. The idea itself, however, is not a brand new idea. It's an idea that dates back to the 1950-60’s. The idea of a virtual neuron, a digital neuron was born all the way back then. Already, there was solid concepts around it. But an interesting thing happened between the 60’s and the 2000’s, which is a paper came out, I think it was around 1980 or so that basically had an assertion showing that these algorithms can't be provably accurate. What's something like that. And it was essentially a scientific proof in quotes of the fact that neural nets can't work. And because of that paper, we lost maybe 20 or more years of AI research in history. But, but they, but some, some computer scientists were curious enough to, to reopen that Pandora's box, play with it and find that it actually does work against, you know, some of the theoretical findings that were there. And so now we see the full brunk of this technology starting to impact society.

Jeff Bullas

00:12:14 - 00:13:58

So, you know, let's talk about the most obvious form of AI, generative AI which is being brought to the public consciousness on November 3rd last year, even though it's been around for a few years because it got given a user interface that was easy to use. And what blows me away is it's the fastest consumer platform to reach 100 million users in human history.

And I, the other thing that fascinates me about anything like is that any technology that's given a friendly user interface is where it starts to explode. That is, it's useful. So I remember discovering Netscape, the first one of the first search engines back in 1995, right? And it took me 20 pages to get through to find something useful. But it provided a user interface for anyone to use. And I think ChatGPT has done that for AI and we are now seeing the application. In other words, the research has escaped the box and has been brought out into the real world and it's used by everyone. So, what we can do is we can play and learn in real time. And that's what's exciting for me. I'm interested in your thoughts on ChatGPT. And what it's done for AI.

Artem Koren

00:14:00 - 00:16:38

Yeah, I think it's a situation where it's very slowly, the quality of the A I results was improving and the technology that ChatGPT is based on, which is something called transformer has been around for a few years. It's not brand new. But over these years, these transformers have gotten slightly better, slightly better, slightly better. And I in my team, we already saw chat like results as early as a year before GPT formally came out. So we had an early preview of where this was going. Of course, no one really knew exactly when it would reach this level of quality. But what GPT did or the reason that it was, it got so popular is that eventually it reached this level of accuracy and correctness and a low level of hallucinations, which is when it makes things up, that are not true, still does that which is not, not too often that it became just, it became so useful. So it was, I think for, for many of us in the AI technology space, it wasn't such a drastic flip or it wasn't such a drastic discovery because it's, you know, x percent better than the other transformer that we worked with. But for the broader population, there was some kind of a magic watermark that was surpassed with ChatGPT to the fact that now when people engaged with it, the results they get are frequently enough, correct enough to be really, really useful. And then suddenly that's, you know, that kind of magic, the magic moment where it exploded and became so so popular. And now, you know, there's, there's clones of Chad G BT, there's many other large language models in the market as well. And it's a field that's blooming and, you know, in the short few months that has hit this, this has hit the public.

Jeff Bullas

00:16:38 - 00:17:19

Yep. So let's wind back to the inspiration to start a business based upon AI other words to write to, to use the AI platform that helps people facilitate and get distilled meetings, gets transcripts of meetings, gets insights about meetings to summarize meetings online, especially using like, you know, Zoom or Microsoft meetings. So where did the inspiration come from to start a business that you're doing today with Sembly.ai?

Artem Koren

00:17:21 - 00:21:13

My co-founder and I spent a lot of time in management consulting space and as a manager, as in management consulting, your life is basically meetings, that's what drives results, you're connecting different teams, you're status teams, you're moving milestones forward, you're moving resources to accomplish some major task in a large company that requires a lot of coordination. And so there's knowledge being shared, tasks being defined, artifacts being presented, there's artifacts being scoped for future presentation. So meetings are the vehicle of progress for large companies getting management consulting efforts done. And as I came out of that world and we were discussing with Gil, you know, some of the latest and greatest that's going on in technology and in business, we thought that it's interesting that technology has penetrated every facet of work life, word processing, spreadsheets, email communication, right? Everything you're doing is digital. Even how you have meetings is becoming digital. This was before COVID. So Zoom was still not completely dominating everything, you know, after COVID, suddenly everybody's doing online meetings, but there was enough online meetings happening already. So even how you do meetings is digital. But what you discuss in meetings, the actual content of a meeting is completely opaque to the digital landscape. So technology sets up an environment, a digital environment in which a meeting is conducted. Then this white noise of human activity happens, white noise from a person perspective of like from a technology perspective. And then, and then the meeting is over and something needs to happen and it's completely analog and human all the next steps. And we thought that if you could have technology, listen and understand the conversations inside a professional meeting, the value of that would be very, very great. And there is so, so much you can do by having this AI participant in a meeting that would sit in here and then can do something about it. And the key idea was that this would be quote unquote a fly in the wall. So this and it wouldn't require any special commands or any awareness of the fact that there's an AI agent in your meeting. And that was crucial because the technologies at the time were, you know, the Alexis of the world and you would have like, you have to call out Alexa's name and give it the command that it knows. And if it doesn't understand your command, you're screwed. And I knew that wasn't gonna work, right? We knew that if you had to force people to change something about how they naturally go through their meeting, you've lost, there's no way. So it had to be completely natural, completely unaware of the fact that there's an no special awareness of the fact that there's an AI participant on your meeting, you conduct your meeting as this and the AI participant is smart enough to hear it, understand what it is and then do something about it and that's, that's where the idea was born for Sembly.

Jeff Bullas

00:21:13 - 00:21:22

That's very cool. So, what I heard as one of the major aims was, is basically to make the technology invisible.

Artem Koren

00:21:25 - 00:22:11

Right. Yeah, technology invisible, I would say, you know, that's kind of one way of looking at it. The other side of it is make technology, human compatible is a good way of thinking about it. Technology becomes a natural participant along with other people on the team. Not a separate kind of thing, not a different material or not a different type of thing than other people in the conversation. So there is, you know, Joe, Bob, Sarah and then there is the AI participant and there's nothing different you need to do or different, you have to behave because the A I is there.

Jeff Bullas

00:22:12 - 00:23:38

Yeah, it makes total sense. I remember I attended a meeting. I think it was at the beginning of AI and I think I launched a podcast and recorded the meeting and I went into a real meeting in the city. I drove in, parked the car, walked to the office. I finished the meeting and I don't think anyone took any notes. It was a meeting. So, right, sometimes you take notes sometimes, you know, it's just like a meet and greet. I got the end of it and I had this aha moment I said, which is what you've done. You've actually enabled technology to summarize the meeting for you. Because you think about it, if you take notes, write it all down, then you gotta distill that, then you gotta write it up, then you gotta send an email, then you gotta, you know, try to make sense of all the noise which we've talked about as well. Totally get it. I had an aha moment about three years ago. What you guys have done, which is pretty amazing is you are using an AI agent to amplify humanity because you don't have to spend all the time distilling it, it does it for you, but it's still human in that. It's the human insights and expertise revealed within the meeting that the technology just distills and summarizes for you. Is that correct? That's what Sembly does?

Artem Koren

00:23:39 - 00:28:01

So that's where we started. And we created something called the glance view, which is it figures, it effectively reverse engineers the agenda for the meeting just by listening to it. So if you, if an agenda is not specified, just by hearing the meeting, it could come up with a list of agenda items that were discussed that understands topical discussion and it finds the key items that belong to those agenda items. So for example, if there's actions, issues, risks requirements, uh noteworthy items, events, all kinds of things that are mentioned, it will say, for this agenda, these items were discussed, but that's today, that's really the beginning.

So yeah, and there's, there's meeting notes functionality around that where you can customize a set of notes that you can share with your team. And we call that the meeting minutes composer. There's also un attended the, there's also proxy attendance which is the agent can come to a meeting instead of you. So if you're running, you know, if you have an appointment, you have a conflict, you have another meeting, but you really want to know what's gonna, what that meeting is discussing. You can send somebody to attend. You don't have to be there. It will come, it will greet, it will say hi, this Sembly, your meeting is being recorded.

You can even leave a little message with Sembly. Hey guys, I'm off to the dentist. Sorry, I couldn't make it. Please make sure to mention this topic and Sembly will listen, record, digest it to a glance view and then deliver it for you in your web app or mobile app for you to consume when you're ready. So that's a really beautiful function as well. Obviously, there's infinite recall because Sembly remembers all the meetings it goes to. So you can search across all your meetings and that was, you know, we've had that in the product for, for quite some time now. But the new stuff that we have in the product, specifically my commitments in Sembly and is the next level stuff. And that's really exciting.

So my commitments in Sembly tracking the to do items that are voiced in your meeting. So if you say you're meeting with your team and you go, okay, you know, I'm going to put together this marketing presentation for next week. Sembly will note that as your to do item, it automatically understands that and it will add that to your to do list. You can even have it automatically populate your to do app. If you use Microsoft To Do or Google tasks, it will, you can integrate the Sembly interface to that, it will automatically populate your to do app. So you have an automatically filling to do list from Sembly AI. So that's one really cool unique functionality that Sembly already has today. And then the other one and this is the one that's really taking our customers by storm. This is probably the most loved feature in Sembly today. It's something called Sembly or Ask your Sembly, it's effectively ChatGPT for meetings. So you can ask it, suggest agenda for the next meeting. Based on this one, I will give you, you could ask it, write me a follow-up email based on the meeting. You could ask it, you know, what were the main points made by each of the people in the meeting? What was the tone of the meeting? Tell me, you know, write a poem based on this meeting. I mean, there's no end to the level of flexibility that you can engage in it for really saves a lot of time and provides some brilliant insight. I use it constantly, for example, I used it yesterday and it's fun because I just have these real, like everyday use cases. Like we have, we're following up on our stock to audit and there was a meeting we had with our auditor a while ago and there's a certain period of audit that they go through and I couldn't remember exactly what the period was. Was it two months? Was it three months? So I went to Sembly and I literally said, what did Chris say was the period duration of audit? It thought for 10 seconds said ba ba ba ba Chris said it was three months.

It's really, really helpful, really brilliant.

Jeff Bullas

00:28:01 - 00:28:39

That's really, that's awesome. In fact, one of my team members said, could I use a, you know, Sembly to actually attend the meeting for me? And that's my next question, which I said, yes, it does do that. But what if you said to Sembly, what are the questions I should be asking at this meeting? I can't make out what are on this topic in this area? And it sort of like maybe you create an agenda before going what questions should I be asking as CEO or founder of my team that I should know, can I simply do that yet?

Artem Koren

00:28:40 - 00:29:59

So before the meeting, you can talk to Sembly. And so let's say you had a previous meeting and this is a follow up meeting, you can ask Sembly and about the previous meeting for sure. And it will do a great job. That's the scary part. It really is really good and you can ask it based on, you know, based on this meeting, what are the key questions I should ask in the follow up with the CEO and it will tell you and then you can have your next meeting and somebody effectively prepares the agenda for you or prepares the key items that you should be discussing. Let's say you had a meeting with your supervisor or manager, your executive before the meeting. It's like, what were the key concerns of, you know, what were the key concerns of Natalia in this meeting? And it will give you and then you wanna address those concerns in the follow up meeting for sure. That's, that's there. And it really is the cool thing is what really makes me sad is that it works really, really well. We put a lot of really good brain power into making sure that the way that it answers these questions is really relevant and useful and it's doing a really good job.

Jeff Bullas

00:30:00 - 00:30:09

Now that raises another question for me, which is can Sembly currently interpret tone or sentiment?

Artem Koren

00:30:12 - 00:31:18

It can if we already have in the product where it shows you the top 10 highlights and top 10 low lights, you could also ask about the tone. It tends to not, it tries not to be very negative generally and not critical. But we found ways to, you know, get it to tell you like a good thing. It's like you can tell it, you know, like, you know, this is like this, like you can say like we're, you know, this is an interview and I'm trying to improve for the next time. Can you give me some feedback about blah, blah, blah? It will tell you some feedback, it will tell you about the tone and it will tell you about the sentiment in the conversation. It doesn't currently use any mimicry like meaning that it doesn't read your facial expressions and it doesn't read your audio tonality. So those elements are still absent. So it's purely based on the words you use, let's say but based on that, it can already discern a lot.

Jeff Bullas

00:31:19 - 00:32:01

Because there are a lot of good things in a meeting, what I've discovered is that some people are very blunt. In other words, they'll ask a direct question or, and assume something instead of that skilled human communicator will go something like this. I go, so what I sense what you're saying is maybe ABC in other words instead of, oh, you said this, that's direct. Instead of being tentative in communication by listening to what they're saying and then feeding it back in a tentative subtle communication which doesn't threaten people. That would be interesting.

Artem Koren

00:32:02 - 00:33:51

It can definitely assess communication style. I mean, it can, you can go as far as ask it, you know, assess the top three communication styles used in this meeting and for each style sort, each participant in the meeting under the style they've used and it will give you that. It will give you that. Because we, yeah, and that's not, yeah, that's certainly, that's certainly possible because we already have things in the system. Like we have a list of suggestive prompts just to give people an idea of the breadth of the kinds of things they can ask to kind of like a starter. So we'll say like, okay, you can ask it to give you a summary of the meeting in Dutch just to, and just to let people know that they can actually ask for answers in any language. They can even talk to it in any language. So you don't have to talk to it in English. So our platform understands eight languages right now, eight languages. So just like English, French, Spanish, Italian, Portuguese, we even have Dutch, we understand Japanese as well. But regardless of the language that the meeting was in, you can talk about it in any language you want. So the meeting was in Italian, you can ask a question about it in English and ask for an answer back in Chinese and it will all work out.

So that's really cool. So we get this, this breadth of different questions. But one of the questions says, identify the issues that were raised in the meeting note who raised the issue and rated by criticality, high meeting low. And it will give you the issues for each person in the meeting with the criticality rate. So we can do that as well.

Jeff Bullas

00:33:51 - 00:34:04

That's very cool. So if you introduce this technology to the team and you knew how to use it very well, you could use it to make yourself look really, really, really bloody good at what you're doing.

Artem Koren

00:34:05 - 00:35:14

Absolutely yes. And a lot of people use it that way, which is, you know, we talk a lot about productivity and time savings and more effective outcomes for meetings and all of that is true. And we're seeing our customers get huge, huge value gains from using Sembly. However, what we often don't talk about is the human element and people you know, want to improve their performance and their perceived performance and Sembly and technologies like Sembly are a bit of a superpower for that. It's both from learning, like what could I have done better? How was I perceived getting additional insight and being able to adjust based on that insight, but also you know, improving your actual results. So you, for example, let's say, you know, you have to share the meeting notes. Well, you can put the meeting notes in and say, like can you, you know, suggest like how relevant are these notes to this meeting or is there anything in these notes that I missed relative to this meeting? I could do all that.

Jeff Bullas

00:35:14 - 00:35:21

Cool. So is simply recording and being used for this.

Artem Koren

00:35:22 - 00:35:30

Yes, so I kicked it off a little bit but yes, I started with Sembly on this call. Yeah.

Jeff Bullas

00:35:31 - 00:36:30

So can we get some insights from Sembly that you can send through to us that we can share? That would be very, very cool. Yeah, definitely. So I saw the agent at the bottom going, okay Attending, I went oh maybe he's using Sembly to do this, which actually is really cool. OK, so let me ask some other questions. Everyone's a little, there's a couple of different camps with AI. In other words, some are shit scared because it's gonna take their jobs. Some are excited because they see themselves as being amplified and have bigger business opportunities. There's gonna be new AI digital business models. That's what excites me as well. Saves me time. That's brilliant. It provides better insight. So I'd be interested in what you see as the future of work with technologies such as ChatGPT.

Artem Koren

00:36:31 - 00:41:03

I think that the change is so massive that it's really going to touch absolutely everything we do. I like to say that technologies like GPT they're more of a material than the technology, meaning that there's gonna be bits of this material in almost every business and almost every technical solution you're gonna find very similar to how the internet is today where you're going to be, you're going to have a difficult time finding a solution that doesn't assume internet in some way and isn't somehow online connected. I think a similar thing is going to be true about AI where there's going to be products that use a bit of AI to just enhance and then there's gonna be on the other side of the spectrum of products that are entirely AI and like Chat P T, for example, right? The whole product is AI, but most products, most technologies and most businesses will touch AI in some way.

No, there isn't any danger of humans being replaced by AI any time soon, I'm not saying far into the future that might not happen for certain types of professions. But we also have to keep in mind that the presence of AI will transform businesses, transform professions and so new kinds of professions will also come to the fore. But today, you know, you're absolutely right when you said that AI is now a superpower, it's something that people can use to multiply their talents, multiply their abilities in a really, in a really effective way. And so whether you're a designer and you can use AI to generate some mockups and ideas for you and then refine them from there or your copywriter and you can use AI to create a draft for you and then you refine from there. Those seem to be the primary creative use cases today. But as we go forward, you're going to see AI step into more and more higher level functions across the environment. So for instance, Sembly, we're building towards what we call work management. So I mentioned that we have this AI to do functionality in the product and I call that like the baby version of work management going forward, Sembly will have, will do a really, really good job of understanding what are all of the things that could or should be done as a result of the conversation you just had and it, it will understand, you know, the full context and the full detail of those activities. And it's going to first of all be able to update the workflow applications of our users. So whether that be a to-do system, a task system, a project manager system or a wiki where you have your project ideas, those can be updated and then that work is consumed by individuals, teams and departments.

But what's really, you know, next level and frontier is something called autonomous agents where you think about ChatGPT, but with a goal and the ability to coordinate other ChatGPTs towards the accomplishment of that goal. So that already exists in a very early format. It's very rudimentary today, but a year from now, autonomous agents are going to be really powerful and very prominent. It's going to be I like to say that if large language models or ChatGPT is like the internet, then autonomous agents are gonna be like social media.

So it's the huge thing that builds on top of this AI capability that's gonna be autonomous agents. And the reason that we're excited about is because today, when we can identify work items to be done out of meetings, to put into your task system, your to do system project system. When an autonomous agent is available for certain kinds of activities, autonomous agents are going to be themselves an additional category of ask consumers and produce work based on that.

Jeff Bullas

00:41:05 - 00:41:33

So what I'm hearing is that potentially Sembly could take a meeting, create a to-do list and then share it on collaborative project management platforms as well as a chat platform such as Trello for project management tasks. And on top of that is communication to Slack. Is that the sort of things we're talking about?

Artem Koren

00:41:34 - 00:42:15

Yeah, absolutely. Exactly. So wherever you know, you kind of keep track of the work your team is doing, we can put the items, the work items that need to be done in those areas. And this bridges a lot of this gap if something happens in the meeting. But ultimately, you want the, what, why did you have the meeting? Well, there's, there's a few reasons to share information, to sync up et cetera. But one of the big reasons that the meeting is because there's some things that need to be done and Sembly can very accurately affect that aspect of the meetings that teams kind of.

Jeff Bullas

00:42:15 - 00:42:28

Yeah, that is, that sounds awesome. There's a bit of a standing joke. I used to, you know, as a joke, I had a long time ago that meetings there where you'd take minutes and waste hours.

Artem Koren

00:42:29 - 00:42:38

I like that. I think never been true in our post COVID all Zoom online world.

Jeff Bullas

00:42:39 - 00:42:51

So maybe there should be a new saying that Sembly allows you to take minutes and create a shit load of work for other people.

Artem Koren

00:42:52 - 00:43:21

Well, I don't like to think about us creating work for people. I think that, you know, if anything it's formalizing the work that does need the AI who does it ultimately. So and our slogan is attend less, produce more.

Jeff Bullas

00:43:21 - 00:43:56

I like it. I love it. That's fantastic. So just to sum it up, we've talked a lot about how companies can use Sembly and meeting technology with AI that Sembly does to supercharge their teams. Can you sum up maybe the top few points on how that can be done. In other words, how can Sembly be used to make meetings much more efficient and supercharge teams? I'd be interested in the distillation of what we talked about to sum things up.

Artem Koren

00:43:56 - 00:47:33

Sure. The first big value point for Sembly is that meeting recall. So now all your meetings that you've had, you can search back and recall information from a prior meeting that you might want to have clarity on or maybe you weren't there for the entire meeting and you wanna look up what happened. The second bit is proxy attendance, which means that somebody can go to a meeting instead of you, what this means is that you get to attend fewer meetings, not all meetings, you are a decision maker in not all meetings, you have input for some meetings, many meetings, you may need to attend just because you want to catch up on what's going on or be informed. Well, you no longer need to be sitting there in the chair and staring at the screen for that meeting because somebody will go record and process the meeting for you and then give you a digestive version of it. You're gonna be up to speed in, in 30 seconds and you just save yourself an hour of sitting in front of Zoom. So the value there is fewer people need to attend meetings generally and you get back in more productive time. Next is you, we track key items for meetings, like actions, issues, risks and so on. And so all of the important points of your meetings are captured. You can focus on the meeting content as you're going through the meeting and not worry so much about taking notes because you have simply taking the notes for you. And then after the meeting, those key items are available for you, you can make sure that you're following all of the important stuff you to do if there's things that you said, I'm gonna do this in that meeting. You might have had 100 meetings that week. You might have, you know, made a few commitments during those meetings. You come in on Monday, I can tell you for myself, I have a very hard time remembering what all those commitments were, but Sembly puts them in a nice list for me and even puts them in my to-do app. So I just open my to do up and Sembly has populated them in a separate list. So I know all the commitments that I've made. It's super useful, super helpful to remind me of the things that I need to be working on. And then finally, of course, their Sembly which is uh super valuable. It's something that can clarify elements from a past meeting like an important date or an important dollar sign. I had that experience with the sales team that quoted me something that I had to look up from Sembly, just ask Sembly what the quote was from a previous meeting and Sembly told me, but also something to generate uh follow on content for you that you don't have to write. So assembling will write an email for you. You know, write me a thank you email, write me a follow up email, focusing on the following agenda, topics that we discussed, uh propose, you know, write me a follow up email and propose a time to meet next time. All of those things simply will write for you and then you can just copy and paste that into your email, adjust slightly and end up.

So really, really great, again, across different languages. So, not just English, if it's a European language, we probably support it plus Japanese and then you can talk the Sembly in almost any language with Sembly. And so a lot, a lot of interesting value and a lot more exciting things to come as well.

Jeff Bullas

00:47:33 - 00:47:59

I think it's potential is almost unlimited because as I was hearing your summary of what the Sembly can do is that a lot of what we've talked about is a lot of internal communication. But what if you could take Sembly run an interview and then turn that into a training that is distilled as well for external communication and training.

Artem Koren

00:48:02 - 00:48:06

Sorry, Jeff, I lost you for a bit there. Can you repeat that?

Jeff Bullas

00:48:06 - 00:48:38

So what we're talking about is a lot of internal communication and distillation and time saving. But in listening to what it does internally, I think there's maybe a possibility and it might even be on your road map is could you use it for interviews such as this and then turn it into a distilled summary that could then be turned into a range of marketing and training opportunities for external use.

Artem Koren

00:48:39 - 00:50:17

Of course, of course, our product is currently used by recruiters. It's used by marketing teams. Certainly, I would say really any conversation in the professional domain can benefit from an AI participant attending that conversation. There's things you're going to gain by doing that Sembly specifically are, are we lean into working team conversations that happen to be our focus area? But we have a lot of different kinds of users and customers from individuals to small teams to large teams to very large customers. Indeed. So then that's, and that's maybe, you know, that's even on one hand, that's really great because there's so much different kinds of value that technology can provide. But on the other hand, it actually is a little bit of our Achilles Heel is that it's hard to pin down just one aspect of value for this platform. Because there's, I'm sure there's a lot of use cases that we haven't even thought of yet.

All, you know, let's say doctor patient conversations or in the educational realm or right, there's a lot, a lot of different contexts that this can be used in. And so I think that this tech will continue to proliferate and we're gonna see some special specializations going forward. Now we're focused more on the working environment and productivity aspect of things and there may be products that focus on other directions.

Jeff Bullas

00:50:18 - 00:51:02

Yeah, because like I said, I think the challenge is it can do a lot of things because it's all about human communication. So, like you said, could this be used? And of course, it could be us for doctor patient communication, could it be used for recruiting you're obviously using for recruiting communication? So, yeah, what a fascinating conversation and I think this is just amazing and exciting and really, really cool. So, anything else you'd like to say to our audience before we finish up,

Artem Koren

00:51:03 - 00:52:07

I would encourage everyone to try. It's one of those things that you don't know, you can't live without until you've tried it. And so I encourage all the listeners to give it a whirl. We have, you know, a free trial that you can use and check out Sembly and ask it, invite it to your meeting and ask it some tough questions and see how it does. The website is www.Sembly.ai, that's S E M B L Y. And if you want to connect with me, I'm Artem Koren on LinkedIn as a Chief Product Officer of Sembly AI. So encourage everyone to try and see how you can use it in your day to day work and how it can improve your life. Speed you up and free you up to do the work that is really productive and it's not just time consuming.

Jeff Bullas

00:52:07 - 00:52:24

Yeah. So you and your price, I had a look at your pricing, your price points are quite attractive. Tell us a bit about your price points. You've got personal, professional teams and enterprise. So tell us a little bit about how the price points for your product works.

Artem Koren

00:52:25 - 00:53:39

Sure. Very standard SaaS kind of a price breakdown. So we have a free personal plan. It's probably going to be fairly limited for most users. It's really intended for you to get to know the product, but you can use it if you're a very light user, you can probably get value from it as well. Then there's the professional plan that's only 10 bucks a month and that's for individual users who want to experience the full power of Sembly. But just use it on by themselves without a team, then there's the team plan. So there you can have a workspace and you can have an agent shared across the workspace and you can collaborate across this workspace. You'd have a single agent, attend your team meetings and then you'll have intelligent access to those meetings. That is $20 a month per user. And then we have an enterprise plan as well that's negotiated on the basis of however, many customers that enterprise wants to be on board. So that's the range of plans today. There's also a credit system where you can purchase credits for things like uploads and, and additional Sembly and questions and things like that.

Jeff Bullas

00:53:40 - 00:54:00

Cool. Very cool. So it's very attractive for people to start using straight away on a range of different plans which is monthly subscription. So one of the question I did have was maybe getting the weeds a little bit but do you have APIs into Trello and Slack, for example, at this stage?

Artem Koren

00:54:01 - 00:54:57

Yeah. We have a Slack integration natively. We have a Microsoft To-do and the Google tasks integration natively. We are going to be offering them many more integrations over the next few weeks on our radar, things like Notion and several others. But we also have what we call custom integrations inside the product, which means that for example, if you have your own like application or if you user, for example, you can beam information from Sembly into some third party application using that. So there's a lot of flexibility in it. And over we're going to be adding a lot more native integration options over the course of the next few months.

Jeff Bullas

00:54:57 - 00:55:11

Very cool. And, and what I just keep having questions as we chat and what is your, I suppose ideal customer or is it just a kaleidoscope of customers?

Artem Koren

00:55:12 - 00:56:10

There really is a wide range. I would say that teams and teams of teams who really value productivity are our best customers and that really opens, you know, a wide, wide range of users. And I guess by productivity, I mean, those whose time is valuable and who want to invest that time into specialized and productive work and not just sitting in an online meeting. That's, that's who our product is for. So, it helps the individuals and makes, you know, their job more fun. And it lets them do more with less, but it also is great for organizations who can get the max productivity out of their workforce and improve the quality of work life for their staff.

Jeff Bullas

00:56:10 - 00:56:26

So, are you considering another question which came out of what we're talking about before? Are you looking at because you talk about specialization? In other words, you are providing as a tool. Are you looking at white labeling in the future?

Artem Koren

00:56:27 - 00:57:00

We have had some requests around white labeling. We haven't had a proposition that we thought was compelling enough for that yet. But there, that is something that could be possible. Given the right kind of partner for that, it would have to be, you know, we're looking at case by case. But, yeah. Yeah, there's the white labeling has come up in the past. It just hasn't, you know, we haven't had, we haven't had a partner that we clicked with enough to do that.

Jeff Bullas

00:57:01 - 00:57:49

Yeah, it was just a thought that crossed my mind because in terms of, because what we're seeing is a lot of general, I see two parts to a number one is a very general ChatGPT does for everyone. Then I think what we've got is we've got the big models that are industrial AI with all the content and all the technology and all the servers. But on top of that, then it's gonna be, I think what's gonna merge is very much specialty AI and you're just doing it for meetings, for example. But within meetings its AI verticals as well such as you mentioned, doctors and how big is that market? Huge. So, but it's, it's just really interesting to hear what you're doing. And my brain's basically exploding but it's fun.

Artem Koren

00:57:50 - 00:59:27

Good. I think that's a good sign. You know, I did a, I was invited to speak at the APMI Sweden event recently in Stockholm. And I asked the audience to raise their hand if they have been in a meeting with an AI participant and maybe 5% raise their hand. Very few people. I said, ok, so I said by the end of this year, 30 to 50% of you will have raised your hand. By the end of the following year, over 80- 90% of you will have raised your hand. Our future is with AI teammates.

That's the future in every vertical, in doctors, lawyers, IT, marketing, sales. AI teammates are absolutely inevitable in our future. It's already happening. So, for example, we had a call with a VC investor yesterday who joined the call with guess what? His Sembly agent. And he's like, yeah, I met with someone he had his Sembly on, I said, oh what's that? I want it for myself. And now, you know, he is Sembly in his calls. So the future is AI participants, AI copilots in all of our professional conversations because the value is so compelling, the value is so compelling. And that's just gonna be the bar.

Jeff Bullas

00:59:28 - 01:00:10

Amazing. I've loved our conversation. Artem, it's been absolute pleasure and it's very exciting. And on top of that, almost overwhelming because we, there are so many possibilities that rise out of just AI for meetings which I knew it was going to be very interesting. But it's more than interesting. It's incredibly exciting and well done. And so just to give you an idea of where you're going, I think you've got about 40+ employees now. Is that right? And can you reveal how many customers you have?

Artem Koren

01:00:11 - 01:00:17

We have over 35,000 registered users on the platform.

Jeff Bullas

01:00:18 - 01:00:37

And are you still raising venture capital? So because quite often in technology, there are competitors and say, not only a technology arms race but a market share, arms race as well. So are you still lending into raising money so you can actually win the race?

Artem Koren

01:00:38 - 01:01:20

Yeah. So we're talking to VCs, most have reached out to us by their initiative. So we're in those conversations. And there's a few that we're singling out that we'd like to talk to as well. Yeah, so we're, we want to make sure that we position ourselves in the domain of AI that we want to succeed in and have quick market share gains in that area for sure. And also obviously be able to offer uh the most competitive technical offering that we can, which, I think we've done a good job.

Jeff Bullas

01:01:20 - 01:01:31

So far, you know, I think you have been amazing. Congratulations and the team and thank you for sharing some of the future of AI.

Artem Koren

01:01:32 - 01:01:39

Thank you for having me. It was a great conversation. If you're ever in Amsterdam, I'm happy to meet up.

Jeff Bullas

01:01:39 - 01:01:43

Well, that might be next year or even earlier. So, thank you very much.

Artem Koren

01:01:43 - 01:01:45

All right. Have a nice day. Bye.