Artem Koren Podcast Transcript

149
Headshot of Chief Product Officer Artem Koren

Artem Koren Podcast Transcript

Artem Koren joins host Brian Thomas on The Digital Executive Podcast.

Welcome to Coruzant Technologies, home of the Digital Executive podcast.

Brian Thomas: Welcome to the Digital Executive. Today’s guest is Artem Koren. Artem Koren grew up and lived most of his adult life in New York City. He graduated from Columbia University with a degree in computer science. He later attended the NY Stern School of Business, where he earned his MBA.

Artem is passionate about using AI to enrich humanity and create positive change. He combines his experience as a systems engineer, product manager, IT executive, management consultant. And entrepreneurs create innovative products, bringing to life new ways that technology can help us reach our goals.

Sembly AI, which he co-founded with Gil Macleff in 2019, is at the forefront of advancing natural language understanding, paving the way for a symbiotic partnership between humans and technology.

Well, good afternoon, Artem. Welcome to the show.

Artem Koren: Thank you, Brian. Happy to be here.

Brian Thomas: Thank you so much. I appreciate you jumping on. I know we traverse the world in different time zones doing this podcast, but it’s really fun when I do go international. So, thank you again for hailing out of the great country of the Netherlands right now. And I know that time difference can be challenging, but we love this stuff. So, Artem, let’s jump right into your questions.

You’ve got quite the career in technology as a product manager. You’re a senior executive and entrepreneur, and now you’re the co-founder and Chief Product Officer at Sembly AI. Could you share with our audience the secret to your career growth and what inspires you?

Artem Koren: Big question. I don’t think there’s a secret per se. There’s no magic. I could mention a few things that I do that maybe are not. That I’ve observed maybe are not so common. Maybe that could be helpful. So, my background is originally in technology and technology management and I’m naturally very curious.

So, I like to learn new things. I like to get deep into things. I kind of have this. An engineering mindset about me in the sense that I like to understand how things work at the detailed level. And that helps a lot, because once you have the details, the underpinning that you can invest forward, and that always helps when you’re making a higher-level projections decisions planning.

You can use that detailed information to have a good sense of what will work and what won’t work. And so, my curiosity really helped in that. And I think that’s been a common thread across my entire career. Another thing that I do is I don’t actually pursue like, advancement or high-level roles.

I, I, that’s never been a goal. You know, I’ve worked with some people who are very focused on getting that title, getting that next promotion. That’s never been a goal for me. I’ve never done something to be promoted to something. I’ve only done things that I thought were the best things that I could do in the role that I’m at.

And that’s always. Worked well in my career, because what happens is people notice people, especially managers and your superiors, they figure out that they can rely on you to deliver very high-quality results to specifically what they need. You can get things done that they want for them. And that’s serves as a.

Of your career very naturally and so rather than try to scrape up this career ladder, you’re kind of using the flow of your activity to propel you. And so, you’re letting your work and your results propel you up rather than. Climbing up that ladder playing politics or anything like that. I don’t like playing politics.

I don’t like positioning. I like to do well and in quality organizations that tends to get rewarded. And so, I think that’s been responsible for a lot of the responsibility that have been handed over time and eventually also my ability to work with my co-founder, Gil McLeod to put together assembly.

Brian Thomas: Thank you so much. I appreciate that. I love. The fact that, you know, I think we’re kind of, it’s ingrained in us. I don’t know if it’s just taught in school and the career ladder whole thing, but I like your perspective on that. And you’re right when you’re not focused on some ulterior motive. And you can focus that energy.

Into just doing a great quality job, producing some great solutions that makes a huge difference. So, thank you for sharing. I certainly appreciate the unique story on that. And Artem, you’ve developed a productivity platform that is clearly dominant in your space for AI virtual assistants, right? Can you tell us what makes your platform unique and the 1st choice among your competition?

Artem Koren: A number of things and several of those things are not. Yeah. Necessarily obvious or some things that hit you in the face. Some things are so, for instance, our platform is the only 1 of its kind that does a really good job at mixed language meetings. So maybe a quick background. What’s Emily? I is it’s an meeting assistant also a teammate, which is how we actually prefer to be known.

And assembly attends your work meetings, whether they’re in Microsoft teams, Google, me, zoom. You can also upload prerecorded meetings and record microphone meetings as well. and it shows up. It’s a passive participant that doesn’t try to tell you what to do or interact with you just to say, I’m here.

I’m listening to it and recording it. And after the meeting, you’ll have the full transcription of your conversation. You’ll also have the video of the conversation. You have some of the best-in-class meeting notes for the conversation you just had. And also, human great tasks, and these tasks can be automatically integrated with your to do apps with your task management apps or anything like that.

Then we also have this a chat bot. So, you can talk to them to the bot about the meeting you had. And it’s something that becomes an extra appendage that you didn’t know you needed, but then you can’t live without just the other day. You know, we had a customer interaction where the customer asked for something very specific in a consent notification email, and she spelled it out.

She’s like, this is what the email should say about a week later when my team is doing testing and vetting with me the content. I literally went into that meeting, ask the AI bot. Hey, what did this person say? Should be in that email? Boom, boom gave me the bullets I vetted against what the team was showing me made a few small corrections and good to go.

This is new magic. This is something that you never do before, because you have to rely on memory on the notes you took and when you’re taking notes, instead of participating in the conversation, you’re diverting your energy. So, those are some of the things that. That the teammate does today, there’s much more to it.

Large organizations have certain like security privacy policies that we support, certain sharing mechanics that we support, but a very fundamentally that’s the value. Value of the product so we’re very strong on supporting corporate needs. Specifically, we have individual users. We have small teams, but we’re actually very, very good at the corporate side, which is, you know, we have special deployment modalities.

We, we can do. We’re very strong on our security and privacy aspects. There’s a lot of things we do that. Others cannot you know, we can support different deployment geographies. Custom network access custom security rotations inside the private cloud that we set up for our customers. So that’s 1 of our that’s 1 of our strength.

And of course, there’s always the quality of results. And that’s see, it’s that’s 1 of those things where it’s like ice cream. You know, everybody loves ice cream. There’s a lot of ice cream shops out there, but there’s always that 1 or 2 ice cream shops in your city that have that line and that everyone who visits the city has to go there.

You know, Kansas City, there’s certain barbecue spots. Everybody has to go to that spot, even though there’s like, 100 and so that. That aspect of quality it’s 1,000 little details. And that’s something that we pride ourselves on assembly because the result, there’s a lot of different ways.

There’s a million different ways to get many notes and to identify tasks, but we do it in such a way that when you see it, you’re like, wow, this is exactly what I wanted to see. Like, I just had this kind of a meeting. These are the notes I wanted from that meeting. I just talked about these kinds of tasks and we talked about a lot of random things that kind of sound like tasks.

But when you show us, like, when you show us the task, this is the task that I wanted to see. And so that engenders this trust in our product. You know, we hear a lot with our competitors that. They’re really, they’re okay for transcription. They do fine. And, you know, we support over 40 languages. Some of our competitors support many languages as well.

But when it comes to actually generating useful results that make a dent in your day-to-day productivity, we’re very, very strong and sharp on that.

Brian Thomas: Love that. That’s amazing. And I’ve seen a lot. And I’ve seen a lot just in this last year. And the quality is pretty good, but there’s always things that it just can’t do.

And so very excited to delve more into what Sembly AI does. So, I appreciate you to share on that, Artem. Like I just mentioned, 2023 has been a huge year for AI so many productivity tools emerging with the release of chat GPT in their API. Can you provide your perspective on shaping the future of work with AI?

Artem Koren: So, what we used to think about is the future of work and we started in 2019. So, we’re, we were doing a, I like to say, before I was cool. anD we talked a lot about the future of work and in the many ways what we thought of the future of work, then is the present of work today. So, work has already evolved in a big way.

And it’s going to continue in this direction specifically with regards to I, I think. LLMs, the large language models, which are things like ChatGPT, Lama, there’s others have been the topic du jour for the past year. We’ll continue to be an important topic, but those LLMs on their own are really just the substrate, their material.

And in order for you to generate productivity results or do something useful, you need to do something with that material and companies are just starting today to figure out what this material is, how can it be used and how can it be shaped into something that’s really. That’s really effective. I like to liken it to the invention of carbon fiber and the ability to create new kinds of airplanes as a result.

So, I think the future of work and the impact of in the future of work is going to be in understanding how to use this material now. In the past year, it’s really been 1 off. You ask a question, you get an answer. And that answer has been more and more impressive as we’ve been on throughout the year.

And now we have GPT 4 and other models that are doing tremendous things. But I think this is the very early days. And I actually agree with Sam Altman that the future is agents because agents take you out of that Q like 1 question, 1 answer paradigm and bring you into a goal orientation. So, so instead of You know, I know I want to travel somewhere, so let me ask, okay, like, what about Asia or what about, you know, Bangkok or what about Tokyo instead?

I’ll say, okay, you know, my goal is to go on vacation at this time of year, like, with my family. I want this type of environment. You know, I want to bring my kids. We have this much time. I want to spend this much money. Give me like, 3 vacation plans. That would be great. And the agent will go. Okay, you know, there’s a couple more questions I have for you to complete this task.

Can you tell me, for example, like, what do you kids like to do or are you a family that wants more leisure time or you want to mix? And you’ll provide those answers. So it’s already asking you questions to accomplish your goal. And then it will actually finally accomplish the goal of building this whole travel plan out for you and potentially even allow you to book tickets right from that same interface to the plan that it created just end to end complete service.

So this goal orientation that agents can provide using modern AI that I think is where the magic is going. We’re certainly looking at that direction in a big way. And we have some really, really exciting things coming next year with regards to that.

Brian Thomas: Thank you. And it’s refreshing to hear that there’s others like you entrepreneurs that are really digging in and trying to produce something that’s just going to be a game changer for us.

Everybody says, well, open AI is going to be the dominant force in the industry for AI. And I really think that it’s going to be the people like assembly AI, you know, yourself and your co-founder. Making a difference doing things your own way, not cutting the same path that you know, that some of the big giants are doing.

So, I appreciate that. And the technology that you’re producing. So thank you. And Artem, of course, last question of the day. We are a tech platform, tech publication. We’d love to get into tech. Obviously, I’ve been a technologist for a lot of years, but we’d like to ask if you could dive in just a little bit.

On some of that new and emerging tech that you’re leveraging in your tech stack. Maybe there’s something you can share with our audience today.

Artem Koren: It’s hard to pick 1, because we’re so on edge with a lot of the things we do. So, you know, there’s with the rise of you now have to think about supporting platform technologies that.

Allow you to flexibly work with those so effectively these large language model hubs effectively that that you need to build. So, there’s a few companies that are doing that. There’s some companies that are rolling that internally, but that’s a new kind of attack where you’re basically building this very smart intermediary or gateway system to work with underneath.

So that’s 1 kind of technology, you know, in some sense, it’s. If LLMs are kind of like the internet or like, and then like open ai, like Cisco, then you know, these gateways are these like kind of middleware routers now, you know, you, you have to kind of address it at a higher level. So that’s one new kind of tech, and I think you’re gonna see more and more of these kinds of foundational platform technologies to, to do that.

As far as themselves it’s such an interesting beast because it doesn’t it’s not functional. There’s not 1 point function that it’s attempting to do. And so certainly we’re looking across the board. So Google had a huge announcement recently about Gemini. That’s very exciting. You know, llama is doing great things, but now you have that’s on the market.

That’s a. French leading now leading that’s only that’s pretty compact. It’s only has 7Billion parameters, but it’s managing to outperform 70Billion parameter model. So that’s really, really exciting. So this miniaturization of a lot of technology and while retaining accuracy is also really exciting.

I think specifically. What’s exciting about all of that for me is the competition across the elements in, in the business and coming up with new and innovative ways to adjust the training mechanisms to get these alarms to be smaller, faster and smarter at the same time. And I’ll mention 1 last thing.

So there’s this concept called, like a model of experts, which is essentially this idea that it doesn’t have to be 1 model rule of all kind of an approach. You can have a constellation of expert models who very intelligently collaborate kind of like a team, like a think tank, but instantaneously and these expert models given some kind of a query can have a little huddle among themselves and give you.

With some layer of processing kind of a result that’s many times better than any 1 of those individual models could provide. And this model of experts approach, I’ve seen it already applied to Mr. I’m sure it’s going to be applied more and more. There’s some hints that all the latest models from are using those kinds of approaches.

And I think that’s going to be very exciting to see where that. Leads.

Brian Thomas: Thank you. That’s awesome. I appreciate you diving into some of that. That’s the, the type of stuff that I love, and I know a lot of our folks in our audience love to hear what’s going on around the world. Right. So, I appreciate that.

Artem, I just want to let you know it was such a pleasure having you on today and I look forward to speaking with you real soon.

Artem Koren: Thank you, Brian. It was great to be here. Great questions and it was fun.

Brian Thomas: Bye for now.

Artem Koren Podcast Transcript. Listen to the audio on the guest’s podcast page.

Subscribe

* indicates required