The next generation of search technology for multimedia content

Eric (00:05.314)
And if you want to know how some of the most innovative European startup founders are scaling their business, then this is the show for you today. My guest is Marine Smilinic. And for those who do don't know you, Marine, uh, who are you and why should people listen to you?

Marin Smiljanic (00:20.594)
Hey, Eric. Yeah, thank you so much for having me on today. And, you know, a big shout out to all your listeners and viewers. So, like you said, I'm Mar and I'm the co founder and CEO of Omnisearch, which is building the next generation of search technology. So essentially, you know, whether you're trying to find information inside anything from videos to

images to audio content, documents, literally anything, even the stuff that's not just plain text or HTML or other such formats. So we basically allow you to find information in an efficient, super efficient sort of way within any type of content that you have, either you or your organization.

Eric (01:03.698)
Okay, now, Marine, for somebody like me, I use Google. I think Google's pretty fast. You know, maybe it's slow for some people, but I find that, hey, it does the job. Or if I'm looking for something on my MacBook, I just go to Finder and I do a search for it. So, and I know you said that you can find anything other than just, you know, text-based documents, but why would myself or somebody like me want to consider using Omnisearch.ai?

Marin Smiljanic (01:28.782)
So this is actually where I kind of try to paint this, the picture that we're not actually doing the essentially suicidal thing of taking on Google heads-on. So the main difference I would say is that you've got with OmniSearch, it is primarily geared towards companies and to business users. And so it's not.

Eric (01:39.771)
Hehehe

Marin Smiljanic (01:51.882)
the kind of B2C web search engine that indexes everything, like all the open content. It essentially tries to index everything that's behind, like either a paywall or within a company's intranet or stuff like this sort. So it's more of an enterprise solution and more of an API that allows you to build your own products with search.

Eric (02:11.59)
Got it, got it. So can you share some success stories where OmniSearch really improved the efficiency for some of the brands that you're working with?

Marin Smiljanic (02:19.486)
Yeah, totally. I think that there is a couple that we can always talk about, companies such as Omnipro or Underline Science. So generally, what we've been most successful at up until this point is companies that just have a big library of multimedia content. And they want to expose this in some efficient way to their own end users. And so what they'll usually do is they'll be using these.

industry standard solutions, so to speak, like Elastic or Algolia or such vendors. And usually what the problem is that they'll run into is that can only search textual content. It is primarily a text search solution. So they can search titles, they can search metadata, stuff of this sort, descriptions, but they can't really search the content itself. With us, if they index everything with us and they power the search functionality through Omnisearch, then...

You know, you can type something in, their own end users can type something in and navigate straight to a point in video where that question is answered. Or something that appears in a video, something that is that shows up somewhere on slides. So it's a far richer way to search content and retrieve information for richer and far more efficient.

Eric (03:35.41)
Okay. I noticed that Thinkific, I think it's one of your partners. I saw them on your website. I know what that platform is. For those who don't know, creators that are hosting sorts of like online classes, maybe digital products or something of that sort, they can use Thinkific to run their online classes or digital products. So if I were to have this like video series on Thinkific, would I still need to use some sort of naming convention for when I'm naming the video or tagging it with different tags?

Marin Smiljanic (03:38.867)
Mm-hmm.

Eric (04:03.97)
in order to be able to use your tool. Or I don't have to do that.

Marin Smiljanic (04:07.99)
I'm actually happy that you brought up Thinkific because that brings up a whole lot of warm memories. And I think you guys were involved with that ecosystem too at some point. I think that might be the first time I heard about Bonjoro. So yes, way back in time, yes. So this was actually our first way that we got traction. So for Thinkific specifically, no, you don't need to have any kind of a convention there. We just...

Eric (04:21.282)
Yeah, that's right. We're going back in time here. Yeah.

Marin Smiljanic (04:35.202)
built an integration with the Thinkific API. So we can index this completely out of the box. Like it's a one minute install and you'll get this stuff right off the bat.

Eric (04:47.538)
Okay, okay, no, but from a user perspective though, you're saying that if I'm in the ThinkEvvot platform and I want to use your tool, OmniSearch, I just go in their integration library and download it right there? Okay.

Marin Smiljanic (04:58.558)
Absolutely, absolutely. You've got a pretty, and you know this, I mean, thinkific has a pretty good app store experience just like Shopify does. And essentially you just click there, you install it, you create your account and you start the indexing process.

Eric (05:14.458)
All right, all right. Okay, sounds pretty easy, sounds pretty easy. Now, when did you get this idea to launch the company and how old were you?

Marin Smiljanic (05:23.906)
How old was I? I think probably 28, so a couple of years, three years ago, something like this. And I think the idea, so a little bit of my background. Before jumping into the startup world, I spent three years at Amazon, so on AWS and the Alexa teams. And the problem that we encountered there is, you know,

Eric (05:30.222)
Okay.

Marin Smiljanic (05:46.326)
both of these teams were fairly technically complex. There was a lot of facts that you needed to know, a lot of algorithmic stuff that you needed to wrap your head around. And Amazon had a very extensive library of internal video content. These would be taped video lectures of people explaining these fairly complex concepts on the whiteboards, talking a lot about it, hour long videos and hundreds of them, if not thousands. And...

The problem was that if you're, if you were looking for information inside that you couldn't get past, you know, the title or the metadata searches. And so that was kind of the spark that lit it. If only I could ask a question and get that little snippet of information, as well as the point in the video where that is actually corroborated. So that got it there.

Eric (06:34.054)
Got it. Yeah. So I love it. You saw the problem in your own current job and you're like, I can go solve this on my own. So were you building this at night? This was like a side project during the weekends?

Marin Smiljanic (06:39.406)
Ciao.

Marin Smiljanic (06:47.622)
No, actually, so believe it or not, there was also another startup in between. So I figured that, you know, at a certain point, and I had been at Amazon for about three years, I figured it was a good time to start doing something else, and to give it a shot. Because at that point, it was looking like the so I want to one of the fun facts is that I actually got my promotion right before that and got one paycheck at the higher at the higher salary range.

Eric (07:16.818)
Yeah

Marin Smiljanic (07:17.722)
and then quit. But it was in, you know, I loved my boss and everything. But, you know, it was just a good time to, you know, take that leap of faith and, you know, just continue doing products and building products.

Eric (07:32.014)
Yeah, I've read a lot of things about meetings at Amazon or AWS and I don't know if they're true or not, but when you went to a meeting, was there any sort of a structure to it that was very different than other companies that you imagine?

Marin Smiljanic (07:43.218)
I think it depends. Like the meetings that are very well run at Amazon, those are very different. Like I'm sure that you're probably alluding to, you know, the fact that you need to come prepared with documents and then the first, I don't know, like 15 minutes or 20 minutes of the meeting, these people just reading the documents so that they're all on the same page and then they can continue the discussion. If they're well run,

That has been my experience, yes. And those are pretty efficient meetings. There were also a lot of meetings that are, you know, just ritual meetings. You need to hold them because the higher up says so, but like, you know, hey, how's everyone doing? What's going on?

Eric (08:26.35)
Yeah, how was your weekend? Okay, so let's get back to OmniSearch here. What were some of the things that you did to get traction, like getting your first users?

Marin Smiljanic (08:34.05)
Thinkific was definitely the first one. That's another fun fact. That was the first way in which we got Bain customers. And this goes back to a Vancouver connection. So we were talking to an investor who was also an investor in Thinkific. And so he told us that Thinkific was at that point just launching the app store and opening up their API for developers to be able to develop apps for the ecosystem.

And so we got then connected to the people who were in charge of that program. They gave us the APIs and the API docs, and we developed the app and launched it. So for a fairly long time, most of our income was actually from the Thinkific ecosystem.

Eric (09:23.246)
Here's a good one for you, Marie. Imagine a potential client saying, hey, I'm really sensitive about data privacy. Is Omnichurch still a good option for us?

Marin Smiljanic (09:34.57)
Absolutely. And like you, you would be surprised how prevalent this is nowadays. Like how many, now that we're actually, you know, talking on a regular basis to bigger enterprises, how many of them, you know, just hate this concept of shipping all their data to open AI or something like this, you know, so it's very prevalent.

With all these AI startups today, I use the quotation marks intentionally to just hack a whole lot of APIs together, and that's the product. We have a self-contained system, so it can be deployed to anywhere. We like to deploy it as a SaaS if there is companies that aren't as sensitive to privacy. But for the ones that are,

you can always deploy it on the servers. And so that I think really differentiates us that and the way that it is performant. So we don't kill your entire hardware infrastructure with very heavy models rather. It's all fairly light and fairly quick.

Eric (10:39.186)
Okay, where you at today and what are you excited about in the next 12 months for OmniSearch?

Marin Smiljanic (10:43.67)
So I think that one of the nicest things is that we finally have two verticals that were really successful in meaning online education and the media industry. And so these two, actually I have to say that I'm a little proud of making that prediction right even in the very first days of the company that these two were gonna be the first big successes. And so we've finally started closing some really good enterprise customers.

We're currently piloting with a Fortune 500 company in the, or at least S&P 500. I don't even know the distinctions, but in the US, so we're definitely going up market in that sense. And that's been a very nice development in the past couple of, I would say in the past year. And I'm excited about, I think that I'm excited about us, you know, doing a whole lot of good R&D and good engineering work.

Eric (11:34.042)
Okay. And what is...

Marin Smiljanic (11:41.814)
to bring the models even further down in size and make them more efficient, more accurate, and just building out the team.

Eric (11:50.294)
Yeah. What's been the biggest struggle for you since leaving AWS and running your own startup.

Marin Smiljanic (11:55.774)
I think that the main thing is like you have your skill set and your focus when you're doing big tech engineering work rather than, you know, engineering work in general is that your focus is fairly narrow. Like your context is, you know, your team and your sub part of the product. And that's the main thing that you're focused on. And you're focused on that, you know, eight hours a day, five days a week, maybe six, whatever.

I think that when you get into the entrepreneurial route, you will definitely need to master a whole lot of skills. You need to do sales, you need to do marketing, you need to deal with investors, you need to actually make sure the product gets built, you need to write some code yourself, you need to make sure you're on top of everything else that's going on in the industry, so forth. So it's a far wider variety of things you actually need to be good at.

Eric (12:43.014)
I'm going to go.

Eric (12:53.314)
Yeah. How did you get a grasp on, I guess the weaknesses that you have running the startup and, uh, were there some things that you were completely blindsided by? Or did you know beforehand that, Hey, if this thing is successful, I'm going to have to hire this person to do that part of the, of the job, because I'm just not very good at it, or I don't want to do that.

Marin Smiljanic (13:13.954)
I think generally, yeah, you always, like whenever you are building a startup, an essential part is bringing the right team together. So it cannot, I've seen a bunch of my friends where it was like one founder being like the guy and everyone else being sidekicks. It usually doesn't work. Like not even Steve Jobs did this. He also had a strong team. So generally it's always important to be on the lookout for the top talent and to see, you know,

Out of the people that I actually know in my network who would be a good fit for a given stage, you always got to be careful about, you know, what the talent is. And also what stage you're at, because, you know, you can't always pay top money and maybe, you know, that there's people that would be good, but at a slightly later stage. You know, you see this with engineering all the time. So there's people that are really good engineers, but maybe they work better on bigger systems or

once you have some sort of a structure in place, whereas in the very beginnings, you need hackers, generalists, so it is.

Eric (14:20.73)
Yeah, yeah. It's been one of the main key things that you've had your team focus on as it relates to making sure we're delivering a good product.

Marin Smiljanic (14:31.798)
I think that they actually, you know what? Our test coverage is very good. I think that, you know, that that's been essential because, and I've done a fair bit of infrastructure development, you know, even before I was at Amazon, I was at a company called MemSQL or nowadays Single Store. So that's a database company that ended up being a unicorn. And also I was on S3 and now our thing is fairly heavy, like infrastructure heavy.

In all of these places, you've got to have good quality assurance and you need to make sure that once something gets shipped, it is well covered and that once you run the tests, you actually know that this works, that it doesn't have egregious bugs or even smaller bugs. So I think that was an important investment and I think that was a total team effort. And I think other than that, it's mostly just listening to customers and shipping what the industry actually needs.

rather than, you know, hit kill Mary features.

Eric (15:31.886)
Yeah, let's talk about that a little bit more. Do you have somebody on the team that that's their main responsibility is customer is primarily you. Okay, tell us about that. What are you doing?

Marin Smiljanic (15:37.631)
It's primarily me. Yeah.

Marin Smiljanic (15:43.494)
I think basically the main thing is I talk to customers, that's I think my most important job at this point. I gather some sort of requirement lists from them. And then, you know, we do a fair bit of triaging and saying, okay, you know, what's the effort required for each of these and how important and how generalizable do we actually think they are? So like, if we ship it for this customer, how many other customers will we be able to ship?

to, like, what time can we address with that? And from that point on, yeah, we've got a fairly good engine, like, fairly big engineering team at this point, which is six people. Not big, still small, but definitely better than when it was just me and my co-founder coding in essentially a garage. And so then they prioritize among themselves, and we ship.

Eric (16:36.762)
Yeah. How does your team look right now? Is it entirely remote or is there a few like in one location?

Marin Smiljanic (16:41.782)
So the engineering team is in Zagreb, yes. And then parts of the business team are in the US. So I think, you know, I'm always partial to hiring salespeople in the US.

Eric (16:54.478)
Okay. All right. All right. Uh, so last question before we get into the, uh, the fun facts questions here. Um, what are you excited about in the next 12 months? I think you, you spoke a little bit about that, but is there like any, anything on the roadmap or any sort of expansion, uh, you know, strategies that you've got going on?

Marin Smiljanic (16:58.391)
Mm-hmm.

Marin Smiljanic (17:08.887)
Mm-hmm.

Marin Smiljanic (17:13.97)
Mm-hmm. So I think that the main things from a product perspective are going to be to really make the video searches good and efficient and like make them so that they're easily deployable to data sets of any size. This is a holy grail. A whole lot of people claim that they can do it. I don't think that anyone has ever managed to do it at scale because processing video is just very, you know, compute intensive.

And so we're investing a lot into models to, you know, for people, detecting people, detecting logos, detecting objects, that sort of thing. And so in the next year, this is going to be crucial to get those two, you know, into a stage where, you know, if you've got exabytes of data, you could be able to find the information within that data set.

Eric (18:06.842)
What do you think some of the tools out there that are on the market, like for podcasters, cause I've used Riverside and Riverside has this new functionality where it automatically pulls clips that you can post on social media. I'm not sure what it's using, but there's other tools out there like Opus, uh, claims to do the same thing. You upload a video and it'll pull like maybe 10 different clips that you can use for YouTube shorts or Instagram. Are they using just something that's off the shelf that they plug it, they're plugging in, or are they building these internally?

Marin Smiljanic (18:11.965)
Mm-hmm.

Marin Smiljanic (18:27.204)
Mm-hmm.

Marin Smiljanic (18:35.29)
That I wouldn't, I doubt that they're pulling, that they're building all of that internally. I would imagine like the, sorry about that, that they'd dump it into a transcript and then dump the transcript into GPT or something like this and say, get me some clips. Cause with a podcast, again, like you're not working a whole lot on the video part. You're primarily working on the spoken part.

Eric (19:03.346)
Right, right, right. Okay, all right, now it's time for the fast questions here. Rapid fire, okay, you ready? Okay, instead of stock options in bonuses for new hires, you now give them what?

Marin Smiljanic (19:08.03)
Rapid fire, yeah. Yeah.

Marin Smiljanic (19:22.778)
Pinball machines and Xboxes.

Eric (19:26.731)
I love it man. That's a good one. That's a good one. Blank is a contest or game that I have won.

Marin Smiljanic (19:34.05)
that I have? I have one? Risk or monopoly?

Eric (19:37.166)
Yes, you have one. Okay. Yeah. I like monopoly too. That's good. All right. Uh, craziest thing you ever did to make money or save money money.

Marin Smiljanic (19:42.943)
Yeah.

Marin Smiljanic (19:53.423)
I would say buying very cheap food. I was about the time I was doing the first startup. I think that I would just take some take some cereals and then one meal somewhere outside. Like some ramen. They always say ramen profitability.

Eric (20:11.598)
Yeah. No, I, ramen is good. Me throw an egg in there. You got like, you got like a gourmet meal right there. Okay. Last question for you. Okay. Speaking of food blank is a food that my partner or somebody close to me enjoys, but I really hate it.

Marin Smiljanic (20:16.239)
You got a good meal.

Marin Smiljanic (20:37.234)
Anything that has to do with cauliflower, man, I don't dig cauliflower. I don't dig cauliflower. They try to sell it to me like, you know, something that is both healthy and delish. I was never sold on cauliflower.

Eric (20:47.875)
Yeah, yeah.

Okay, here's the key. Here's the key with cauliflower. Here, here's the secret maple syrup and you can get plenty of that in Canada. Yeah. Sprinkle after you bake it or fry the cauliflower, just sprinkle a little bit of some drops of maple syrup all around. And so it's got that little sweetness factor. That's the key right there. Marine. Yeah.

Marin Smiljanic (20:54.558)
No, you're getting...

Marin Smiljanic (21:01.01)
Oh really? You can get plenty of that. So do you fry it? What do you do? Cough cough

Marin Smiljanic (21:15.978)
Alright, let me give it a shot. Okay, fine. This is, this, this motivates me to reevaluate. I'm not, you know, close, close-minded.

Eric (21:21.974)
Yeah. Hey, this is not just a podcast. This is also a tips on, uh, making your diet a little bit better, right? Give it a shot, Marine. Thanks for coming on the show. People listening. I'm going to put links to Marine's LinkedIn profile as well as the Omni search Omni search website.

Marin Smiljanic (21:30.142)
Good, good, good. I'll give it a shot.

Marin Smiljanic (21:40.667)
Omni shirt, that's what our shirts are called.

Eric (21:42.814)
Yeah, yeah, yeah. All right. Thanks so much for coming on the show, Maureen.

Marin Smiljanic (21:46.582)
Thank you, thanks for having me.

Eric (21:48.142)
Yeah. For those listening, you enjoy this. This show is not free. In fact, I'd appreciate it. If you enjoyed it, hit that subscribe button, whether you're watching it on YouTube or listening to it on Spotify or Apple, and until next week, this is Eric signing off and I'll be back next week with another exciting European startup founder or marketer cheers.

Eric (22:11.298)
Okay, I stop recording.

Marin Smiljanic (22:13.645)
Alright.

Eric (22:15.042)
Yeah. No, man, it's not often that I meet, I interview startup founders that are as, I don't know what's the right words, but that seem as happy and excited and motivated, you know, as you, you know.

Marin Smiljanic (22:19.658)
Sorry about that.

Marin Smiljanic (22:30.363)
And things good to know.

Eric (22:32.046)
Yeah, yeah. When you go back to Vancouver, how do you, what do you do? Do you go like spend a few months there and then come back to Croatia?

Marin Smiljanic (22:37.33)
I think actually now when it comes to North America, I spend more time in San Francisco than I do anywhere else. So I mean, Vancouver, I spent like seven years at. So I'm definitely very familiar with both the startup scene and the city itself, but now I think it's mostly just San Francisco.

Eric (22:57.382)
Do you have a Canadian citizenship or no? Okay, okay, cool.

Marin Smiljanic (22:59.658)
I'm a Canadian citizen, yeah. Yeah, I just got that a year and a couple of months, a year and three months ago.

Eric (23:09.33)
All right, man. All right. In a year and a half, I'll get to apply for my Romanian citizenship. So now...

Marin Smiljanic (23:14.15)
Oh yeah? Good. You should definitely do it because then you're in good shape for all of Europe. Because EU.

Eric (23:20.066)
Exactly. Yeah. I know right now, my wife and kids, when we're traveling, they're always looking at the passport line and just like, whatever one is shorter, they'll go through, you know, the European one. Houston.

Marin Smiljanic (23:29.782)
That's always a good hack. Yeah. And where are you from in Texas? Houston, okay, fine. Houston was one of my first US trips. My first US trip was Chicago. My second one was Houston when I was in the first year of high school.

Eric (23:47.682)
Okay, what brought you there?

Marin Smiljanic (23:49.79)
we had a programming competition.

Eric (23:52.642)
Okay.

Marin Smiljanic (23:54.43)
It was really good. I still remember it. Very warm, but it was really good. I enjoyed it. Oh yeah.

Eric (23:58.414)
Oh man, that's one of the reasons I left. It's so freaking hot and humid the majority of the year that you can't really do anything outside and I just don't want to live there because of the weather.

Marin Smiljanic (24:07.55)
Yeah. Yeah, I can imagine. My aunt lives in Austin. She teaches at the university. So I've done my fair share of Texas visits in.

Eric (24:21.07)
Yeah, yeah. It's not like the West coast up there where you got that breeze in San Fran or even Vancouver, it was just nice.

Marin Smiljanic (24:26.794)
Like I think that San Fran is definitely, like climate-wise, it's on a completely different level than even the West Coast cities, because there's a lot less rain than you get either in Vancouver or Seattle. Because I always joke that everything I do is on that West Coast, like Vancouver, Seattle, San Francisco, so I'm always there. And San Francisco's definitely a way better climate than anything else.

Eric (24:48.088)
Yeah.

Eric (24:54.522)
Hey, so my wife and I are planning to go to Croatia probably in a couple of years, but try to spend at least two or three weeks there. Um, are there, what are like your top two or three cities that you suggest that we spend a few nights in?

Marin Smiljanic (25:01.356)
Yeah.

Marin Smiljanic (25:09.287)
If you're actually going to spend three weeks, I would try picking one of the islands. I think that's... so first and foremost, I would actually suggest you go during the summer. And it is actually very nice even in the continental parts. So you can go during the winter. Especially Christmas time is actually pretty nice in Zagreb and in a whole lot of the other cities.

But kind of the name of the game is Go to the Seaside. That's still the main attraction. So you've got the islands like Viskvar, Korcula, whatever. I can send you those in an email, because I don't know if you'll write them correctly, because Croatian pronunciation is tough. But.

Eric (25:55.91)
Hehehehe

Marin Smiljanic (26:00.258)
Dubrovnik depends, like, it's the most famous place, I think, because, like, the Game of Thrones lore and that sort of thing, but... it is very touristy in that sense, so, like, it can get very crowded. So... But I would say, like, islands are always good to go to.

Eric (26:14.67)
Yeah. OK.

Eric (26:20.162)
And then from the country you get on a ferry, you're talking like one hour or longer.

Marin Smiljanic (26:24.206)
Depends, depends. So like ferries are from Split, which is the main city on the coast. And it'll vary, it varies between like an hour long trip to I think the most distant ones which are going to be like Vis and Corchola out of the bigger islands. Would be like two and a half hours of ferry.

Eric (26:46.35)
Okay. That's cool. Yeah. Whenever we get a chance to send me the names. Definitely. I obviously want to go check out what the locals think is the best thing to check out.

Marin Smiljanic (26:51.15)
I'll send you the names.

Marin Smiljanic (26:57.322)
Yeah, I think that's always smart. Okay, I'll send it over.

Eric (27:00.814)
All right. Thank you so much. I'll keep you posted for when this is going to go live. And this has been a pleasure, Marine.

Marin Smiljanic (27:04.086)
wonderful. Yeah, absolutely for me too. Thank you. Bye bye.

Eric (27:07.738)
All right, cheers.