Episode 16: Beating Google at its Own Game: Search for the 21st Century

 

Chris: Welcome to The Tech Lowdown show where each episode will be discussing opportunities in the tech space with entrepreneurs from the US and around the world. I’m your host Chris Jones. I remember arriving at business school many years ago. And meeting people whose brains were just wired to handle and process lots of complex information. While I consider myself to be intelligent these people had what I like to call mental horsepower.

Today’s guest falls into that same category in my mind. Hamish Ogilvy is founder of Sajari. Sajari is a real-time cloud based search and recommendation engine. It’s designed to work with very large queries like documents people profiles stuff like that. I can promise you that you’ve never seen a faster search product in your life. Frankly it’s amazing and I want to understand how Sajari is turning how search, something we all use every day, on his proverbial head. Hamish received a PhD in physics and lasers and still found time to captain the basketball team at Macquarie University in Sydney. A new father to a beautiful baby girl – Hamish welcome to the show, what’s that lowdown?

 

Hamish: Chris what’s up man it’s been a while man I hope you’re well.

 

Chris:  I’m doing well man thank you it’s great to have you on the show. I’m going to jump right in you’re a rare combination of an athlete, geek. High level rugby and basketball player and you’ve got three patents to your name. Tell us a bit about your background and we’re just interested I technology and data came from.

 

Hamish: Yeah it’s interesting I said I think I was always interested in technology growing up. And always interested in sports as well so I kind of ran down both those paths. And ended up in University doing physics wanted to go and learn a lot about lasers which were interesting at the time. I think I don’t think many people realize but the whole internet is powered with lasers. You know all the over communication fiber running around the world. So laser pretty critical to the world that we have today. I was pretty interested in that and then that just led on to data and other worlds. And to where I am today.

 

Chris: interesting so before we jump into Sajari. You once told me a great story about your first little side project which is still ongoing. Go petition tell us about that business how it got started what you guys were doing.

 

Hamish: yeah that was really interesting I was at university in the time and living in a shared house. Didn’t have a lot of money and we had a one of our house mates that live with. Actually he spends a lot of money gambling and it was hard to get the rent of him. And we were sitting there sitting there one day and thinking you know we shouldn’t have slot machines in Basel.  I think it’s a bad idea we should petition about it and then nobody had really done that in the online space. At least in Australia but it was starting to happen in overseas. But we thought we could do it better, so a couple of us got together and decided to start that. And many, many years on it’s grown over I think there’s over 20 million registered members now. So it got pretty big and been involved in some massive pieces of change around the world. I think from my perspective though that was just a great experience to learn how to build an online business from zero up to millions of views yeah.

 

Chris:  Yeah you guys really grabbed a tiger by the tail on that one. Most people when they think of search immediately think Google. And maybe secondary how they think about Bing or maybe the old Yahoo search. I was lucky enough to see an early version of Sajari in action a few years ago and we were looking at searching within the US Patent archives and frankly I was blown away. Tell us about Sajari and how you identify search as an opportunity.

 

Hamish: Yes I think the interesting thing that we see is that the world’s drowning in information I think as you know I think there’s 40 or 50 times more information every couple of years. Than what there was previously and there’s some of those crazy stats that you know in 2003. There was more data created than all of the other years combined. So the way that information is growing is you know absolutely crazy but then from a human perspective. How we actually interact with that information has to get become easier. Because we’re being forced to digest more and more information. And so search is one of those critical information bridges between humans and data. and so we kind of saw that as more than just keywords so search is really thought of like Google.

[0:05:05.1]

And being we’re going in typing some keywords and you go and find information. And what we wanted to actually get to was to think about if you had a lot more information and you had more idea about what you want. Could you run a search that was more would give you results as if you had read every single document in the collection? And so that was the initial goal for what we’re doing which is one of the reasons we tested it on the Paton archives. Because it’s a big highly complex data set of information. but we see that opportunity everywhere to be honest I mean anywhere there’s information that humans trying to read it. And make sense of it, then we see ourselves in that space.

 

Chris: Alright that makes a lot of sense. So I know you guys have built search from the ground up. You didn’t start with hey this is what Google is doing this is what Yahoo’s doing or anyone else. So yes in layman’s terms tell us how Sajari built its search engines fundamentally different from how it’s been done in the past. And what the benefits are for web and app owners as well as end users.

 

Hamish: yeah so I think when you think back to web the origins of most search that’s available today. And I mean forget Google they’ve got a lot of smart people inside and some of the other big companies as well turn their own thing. But the technology that’s available for website owners or for apps to use is actually a lot more simplistic in terms of the way that it stores data. And that comes about because most of that technology was actually first developed back in the late 90s. And in the late 90s it cost you a lot more for a CPU. Cost you a lot more for disk and memory. And so they made trade-offs which made a lot of sense at the time to store information in very static fixed ways. Which made it reasonably quick to query. But the downside of that in today’s world is that information is moving a lot faster. And so one of the things that we wanted to do is to be able to adapt to that in real time much more like a database would. rather than writing up you know these big long indexes that never change. So that’s kind of the first difference and then the second difference was that we wanted to make it very flexible. In terms of what machine learning can be used to drive the ranking as well. So we’ve spent a lot of time and on machine learning for different verticals to actually become a lot smarter in the way that the search works. And the idea is that it will be able to teach itself to become smarter over time. So that’s our other core goal.

 

Chris: Got it and the benefits for say end users.

 

Hamish: yeah sort of the benefits of the end users is that they get something that actually works is high performance. But it improves itself over time. And not only does it improve itself over time but it will tell you in different ways that it’s improved. And that’s been really interesting particularly for some of the website owners using for site search. Because they don’t realize that they have issues with content. And then suddenly you know that they’ll have content drop in their search rankings we tell them these are the things that have dropped. Here are the areas that have gone up. And then from that they can tell that they have content issues with the way they’re talking to their customers. So we’re seeing that there’s a lot of benefits in terms of performance which people want. And they don’t want to spend a lot of time on it so the other aspect there is that most companies that do really well with search. They actually have several full-time engineers actually working on the problem. And that’s fine for companies who can afford several expensive people but most companies can’t so that’s kind of where we come in.

 

Chris: Got it so you’re a bit more turnkey for these companies and they don’t have to worry about having engineers on-site. And dealing with some of the intricacies of optimizing search performance.

 

Hamish: That’s right yep it’s normally a very manual process for most.

 

Chris: so what are some of the competing technologies and is your advantage by building from the ground up it is significant enough for you to scale that business over time? What’s the key to scaling this business?

 

Hamish: Yeah so we’ve seen really interesting things over the past few years. Has been some great products like elastic search has done really well just putting a nice API around we’ve seen. Which is the technologies from back in the 90s. It’s gone incredibly quickly, so they’ve shown how if you can create a really simple API that the developers are happy to use then. You have a real advantage in the market we’ve seen the same thing with Algalia has grown like a Light speed.

 

[0:09:58.6]

But they’re more focused on the structured data web more leading towards the unstructured data sets. But we’ve seen in this market that one of the most important things is speed. Which we have and another important thing is the simplicity to the end-user. And I think that in the past a lot of the products have been sitting in that too complex space. And that force companies to have engineers to mind them. And so what we’re trying to do is to simplify that complexity but give you high performance. And we think that’s a huge, huge, huge opportunity to stay on yeah.

 

Chris: Yeah that actually makes a lot of sense tell me about the machine learning aspect. You mentioned that and what I’m wondering is are you building your own machine learning algorithms. And is that how you’re doing that or are you leveraging some existing technologies out there to use that for the machine learning piece.

 

Hamish: Yes we do both actually so we have our own internal stuff and we also use various libraries and other things externally other techniques. The aim of will be getting to those to be very pluggable. So what we actually want to have is in the background to be able to plug in different pieces of machine learning. And we already have processing pipelines that will go back over the historical data look at performance. And then see that if we had have used some different learning in the past with that have positively impacted the performance. And if it would then we know that that’s technology that we want to add for the future. So we’re really data heavy business and really constantly looking at performance opportunities. And so the machine learning is a big part of that. Eventually we don’t use or what we use some newer networks to a big degree. But we’re trying to set things up in a way that will be very easy to plug in. Newer networks are some of the other new algorithms that are coming up down the track so that’s the goal.

 

Chris: Do you see a place for Sajari in the voice app space?

 

Hamish: You know that’s actually really interesting we have one project now where we’re search is being used for in a chat sense. So which is more like the voice conversational style usage. and it’s been phenomenally interesting for us because the queries are totally different. If you go to a search box you know and you want to you want to buy a bookcase you just say bookcase. If you go into a conversational or a voice search interface you’ll say I want to buy a bookcase. And so the amount of information in the context that you have from a voice based interface is far greater than what you would get from a search bar. And that’s really suited to what we do more information in the query is Better. So yeah it’s definitely something that we’re looking at we do have active experiments in that in the space.

 

Chris: That one is personally intriguing to me I’m a big fan of what Amazon is doing with Alexa. What Microsoft is doing and Google is doing with their, what Google is doing with their home product. And I’m seeing lots of voice-based opportunities particularly out of here where lots of IOT devices are connecting to voice based platforms. And so that added context that you guys it’s important for you guys search engine seems to be a really natural fit. And that could be particularly if you think about things like cars automobile based applications voice based applications. Can see some real interesting synergies there.

 

Hamish: Yea I guys totally agree.

 

Chris: So what’s the opportunity in the future for search and how are these changes with machine learning and AI and predictive analytics impacting what the future searches going to look like for us?

Hamish: yeah well I think some people will talk about the future plugging search into your brain. And you think of a query and get an answer, so I guess it depends it depends out for them it down the track you want to go. But I think we’re seeing that convergence of you know the information is going to get closer and closer and closer to what you want as a desire. Whether you think it or how you communicate it whether it’s voice and voice is a really interesting one. Because voice is one step back from having to type something into a keyboard. So again it’s getting closer and closer to that just think about something and you are able to get it. And so I think we’re going to see an advancement where search is just going to get increasingly more intelligent. And it’s going to get increasingly closer and easier for humans to use. So you get to a point where you almost be getting results before you actually want them if you know what I mean.

[0:15:06.]

Chris: so very much predictive.

 

Hamish: Very much predictive it yeah.

 

Chris: and in terms of your core architecture are you guys the only ones that are going down the particular architectural path that you have laid out? or there others who are going down a similar are you part of a movement or you guys out on your own?

 

Hamish: I think from the machine learning perspectives definitely there’s others that are going down the same path. I think there’s advantages and disadvantages to you know the other approaches. And we’ve taken a very specific path so we know will be very pluggable and very real time. Which we think is important for where the future is going. Which is different to everybody else and we think about a lot of advantages in terms of laying in this machine learning and predictive analytics.

 

Chris: for a company like Sajari for technology that you built. what’s a exit candidate look like if you wanted to sell this two years from now next week, five years from now. What are the types of companies that would be attracted to this type of technology?

 

Hamish: That’s a good question I mean with we’ve had people companies that have big problems that we can solve very, very well. And I think that they would love to have us in as know-how at the moment we have no intention or no desires to sell. We actually really we think this could be a massive business and our goals to actually build that up. But how the use cases are way too many so the plenty of people who would be interested I’m sure.

 

Chris: Absolutely wonderful so I want to jump into what I like to call the real lowdown segment of the show. And here the aim is to try and get beneath the surface a bit the stuff down don’t find on the glossy brochures. And find out how things are really going with the business. You guys are built and unquestionably amazing technology, but what if it challenges that the business faces today.

 

Hamish: Yeah so I think when we started out we didn’t we didn’t fully comprehend how complicated it would be to rebuild a search engine from scratch. And I think I the amount of manpower that’s actually needed to do that and do it well is unbelievable. and so that’s taken us longer than what we thought and when you take longer the what you think then that obviously comes with challenges in terms of funding. And do you try to self-fund the look for external funding. And if you do then do your metrics actually warrant a decent round of funding. So I mean those challenges and particularly coming in Australia a few years ago that there wasn’t the same degree of funding that there is today. So for us I think the biggest challenges have been how do you do something really complicated with a lack of resources. If you know what I mean.

 

Chris: Yes, yes have you guys gone out and raised money already, do how much have you raised, what do you need to raise now to get to the next level?

 

Hamish: Yeah so we raised 600K a couple of years ago which is enough to get us to where we are today. And we’re basically breakeven now so not under a lot of pressure to rise except to the fact that we really want to move a lot faster than we have been. So we will be raising capital soon and that’s part of the short-term plans. In terms of how much we need to get where we want to go I mean that’s a that’s an open-ended question at the moment. Yeah I can’t give you an answer on that am not sure.

 

Chris: Totally makes sense been there done that skied that jump that. What’s the ecosystem like for your industry in Sydney? Like I know when I was there and I’m very much in app marketing space working with app owners it was challenging. Because there weren’t a lot of international or global apps being developed out of us out of Sydney or out of Australia in general. What about for you in your industry, what’s that ecosystem like?

 

Hamish: Yes this this hits us from two thoughts is the fact that we have customers in five countries now and we try to support that from Sydney. And that’s difficult when you have users in Europe and the US and we go to sleep so how do we deal with that.

 

Chris: You’re not sleeping you got a new born.

 

Hamish: That’s true that’s true, but yeah we will be opening the US office this year. Because we need to have a better spread across the globe in terms of our support and other things. From an ecosystem perspective Sydney’s a really interesting place I think there’s a ramping of technology here. But in terms of search and you know people who build things like databases.

[0:20:03.6]

There’s not a lot that’s a little bit that some groups out of Canberra that have done some very interesting stuff. I’m hoping to meet some of those guys soon we just in the last few days and that a bunch of people from Melbourne. RMIT has a huge information retrieval section down there at the University so there’s plenty of study going on down. And part of those connections have been people who’ve worked at eBay and Google and other places have come out of down in Melbourne. Obviously that’s not Sydney but it’s reasonably close but I think the people in our industry and mainly moreover in the US and in certain parts of Europe. So that is challenging and we would like to be considered some of those people because I think you do you improve things a lot faster when you have like-minded people around. And good meet ups and other things to help fuel industry of the you’re in.

 

Chris: yeah that makes sense would you guys consider relocating the headquarters elsewhere do you want to keep that in Sydney?

 

Hamish: yeah we’re considering now and the other in the US office we’ll probably run to in parallel and then see. See how we how we go with that whether the US becomes to the headquarters or not we’ll work that out as we go.

 

Chris: And as you looked around for money in Sydney. And you’re looking at valuations are you seeing some challenges from a valuation perspective in Sydney versus what you might be shopping for in the US?

 

Chris: Not really I mean I think you see wherever you are to investors who would love to beat evaluation down for whatever reason. I think you get that anywhere I think it’s been interesting for us we’ve met some investors who really don’t get what we’re trying to do. And for those you know if you don’t understand what somebody’s doing and what they’re trying to achieve. Then obviously the valuation is going to seem too high. But then we’ve had others and totally get what we’re doing I understand where we want to take it and they’ve seen other businesses in parallel. Or worked on other businesses that are similar that have done what we want to do. And they can see the potential there and they don’t blanket an evaluation we put in front. So I think it’s a mix and I think that’s a challenge anywhere you go at the moment that just happens to be. A couple of billion dollars of venture capital in Australia that was basically nothing a few years ago. Even 2 billion that could probably more but yeah it’s definitely a lot healthier than what it was a few years back when you’re living down here.

 

Chris: yeah definitely it’s wonderful to hear finally what personal sacrifices have you had to make for the business and in your opinion has it been worth it?

 

Hamish: yeah that’s a really good question I mean you as a business side of you would know this as well. But it takes a toll from a never switching off perspective I think that’s one thing so from the family you  wake up on a Saturday morning and you still have things you need to get done. You wake up on a Sunday morning and you still have things to get done and Sunday evening you’re still thinking about what the rest of the staff and everything you need to do on the Monday. And it’s one of those things you never walk away and I think from a personal aspect that the biggest thing to me. has been I really missed that working in a job knowing that somebody else would deal with all the admin around me you know. Everything and I could go away on a Friday night and have a few beers that everybody else. And wake up on a Saturday morning and do whatever I wanted to do. And even if you do that now you still have a sort of cloud hanging over the back of your head. Thinking that there’s other things that I really should be doing so that’s probably one of the big sacrifices. And it doesn’t active family. And then I think the other one is that from the financial aspect. If you if I was sustain a job I would be earning a significantly more than what I am now and if you do that over a period of years then obviously that that stacks up. So the trade-off is you know if you if you do that and you fail then obviously you shot yourself in the foot financially. But if you do that and succeed then you have the other side where you can potentially do incredibly well.

 

Chris: High risk, high reward.

 

Hamish: High risk High reward, it’s a calculated gamble and when you see you know a big opportunity and you know that you have the ability to do it. It’s not it’s not so bad and I don’t really worry about that we’re kind of past the point where we’re like successful failure. It’s more about how big we’re going to get. So that’s not too bad but yeah in the short term it would be nice to have has more cash.

[0:25:01.2]

Chris: A little extra coin. I know exactly what you’re saying Hamish I want to thank you tremendously for your time. Can you please tell our listeners how they can find out more about you and Sajari.

 

Hamish:  yeah so come check us out Sajari.com s-a-a-R-I.com and yeah give us a ping if you have any questions or crazy applications you think you think we might be able to help you with.

 

Chris: awesome folks as always if you like the show please download all of our episodes and leave us a 5-star rating on iTunes. You can find show notes at Tech lowdownshow.com and follow me on twitter @CJones2002.

Leave a Reply

Your email address will not be published. Required fields are marked *