How to make AI work for everyone, not just data scientists?

Amir Haramaty, СEO & Co-Founder at AiOla ● Feb 6th, 2024

The full transcript

Oleg

Hi Amir! Thanks for joining Devico Breakfast Bar today. Thanks for finding time. Could you please tell us a bit about yourself and your background?

Amir

Thank you. Thank you, Oleh, for having me. I'm a serial entrepreneur, if you will, even though I like a better definition, which is a serial problem solver. Being around tech for many years, in different areas, always trying to identify a significant problem, come up with a clever solution, build a very strong team, and then deliver value to our clients at scale. And I've been doing that for multiple years in different areas. I had my fair share of failures, but I'm very proud of the numerous successes I've been part of. And every step that I took in my career and my life brought me here now. So, what we're doing here, in aiOla, which I'll be happy to elaborate, is really right now the product or results of all those failures and successes.

Oleg

I've never seen an entrepreneur who was successful without failures. So, it's an obvious step being successful.

Amir

Absolutely. Actually, I've been speaking on a conference in Poland, and they were talking about Israel being startup nation and the number of exits for Israeli companies. And they asked me about how come they don't experience the same thing in Poland. And I say, 'If I had to guess, your education system is better than ours. And I've been working with a lot of great Polish scientists and engineers, but I'm sensing that culturally in Poland – and again, that's my observation only, and I can be wrong – failure is not an option.'

And when I said that, many people nodded their heads because, culturally, the fear of failure is paralyzing and preventing. And I said, 'You know what? I could not agree with you more. Failure should not be an option; it must be mandatory.' So, the importance of failure, I experienced it firsthand, and you need to fall hard, fall quick, and remember the scar and the pain to make sure you don't repeat it again. So, to make mistakes is human. I'm making plenty every day. To repeat mistakes is stupid. So that's what I'm trying to learn.

Oleg

Agree, agree. Couldn't say better. Could you please provide an overview of your company and elaborate on the specific challenges or issues it addresses?

Amir

Absolutely. So, for the last 10 years, I've been deeply involved into AI, long before it became the most used and abused letters in English language at present time, before it was so sexy. And I've been lucky enough to work with one of the largest management consulting firms globally and serve with them three-digit number of engagements with Fortune 1000. So, I had a chance to have a courtside seat, if you will, to see what's working and what's not working when a big promise of AI is meeting traditional industries and traditional enterprises. So, from my perspective, what's been done in AI, artificial intelligence, and ML, machine learning, in the last decade, specifically in the last 36-40 months, is incredible. But sometimes people get confused because, with all the respect to AI and ML, those are tools – incredible tools – but just tools, meaning everything starts and ends with data, the raw material. And a known fact is that the majority of data is still not part of the game because the majority of data is still unstructured and uncaptured, uncaptured and unstructured.

So, first is realizing that there's a need. There's a need for a better way to capture more data, and generate granular data that previously was not captured and structured, and turn this raw material to the wonderful AI platform that waiting for quality raw material that doesn't arrive yet. Number two, for the last several minutes, Oleg, what we've been doing is we're talking, and speech is still the undisputable, most efficient, richest, deepest, quickest means of communication. And if you try to use Siri, if you try to use Alexa with an Israeli accent, good luck.

So, I'm not picking on those guys, but I realize, number one, there's a challenge with not enough data as part of the game; speech is the number one source of communication. Yes, speech is not part of the game because all the existing solutions, the automatic speech recognition, are just not good enough. So, when people see a problem, I normally see an opportunity. And I saw that as an amazing opportunity. And what we realize is, especially when you deal with business processes or enterprises, what we found out that more than 50 percent of the vocabulary use was not a language, was not English, for example, it was actually a jargon of a specific industry, of a specific company, of a specific location, et cetera. And there's no way to solve it. So, that's number one, the first pillar we wanted to address, and I'm going to connect them in a few minutes.

Second, you know, working very closely with this leading global management consulting company was an amazing experience for me. And by nature, I'm a disruptor. I enjoy very much to disrupt. And it took me maybe too long to realize that most people, while I enjoy to disrupt, most people in the world don't like to be disrupted. And that's when we need to eat a little bit of a humble pie and respect existing processes in established organization in traditional industries. And one of the things I learned is a tremendous pressure right now for the boardroom to management organization – 'What are you doing with AI?' And people are rushing in to do something with AI because that's really the bon ton, that's what really people expected to do. But at the same time, we're dealing right now arguably with one of the biggest, perfect storms in modern history – combination of geopolitical and macroeconomical, from what happened in Europe to what happened in the Middle East, to COVID before, and many other areas, which absolutely dictating to organizations to make a clear separation between what's perceived to be nice-to-have versus must-have.

Whatever nice-to-have – forget about it. Oleg, I don't have time for cosmetic surgery, I have to stay alive. So, I realize it brings a tremendous opportunity, specifically in traditional industries. So, rather than try to disrupt, respect existing processes and come and say, 'Hey, what if I'm gonna? You have to do it anyway, right, Oleg? Great! What if I'm gonna help you to do it more efficient, safer, smarter, more collaboratively? It's complementary. I'm not fighting against you. I'm helping you to do better.' And everybody at this time needs to do more with less. Everybody needs to be more efficient. It doesn't matter what you do and how big or small. And everybody talks about, you know, how can I be more ESG compatible and generating and cutting many, many, many trees to generate paper. You know, I was talking to one of the largest pharma companies in the world just last week, and he said, 'Do know, Amir, what we are producing?' And I said, 'Yes, medicine.' And he said, 'No, paper, a lot of paper. Because what we do, there's so much paperwork around pharma that what we do, actually, we chop trees all day long, so we can have enough paper to produce all the reporting, et cetera, et cetera.' So, I saw, it's really a combination of different elements, raw material to big... And generate the data and quality data.

They need to do more with less to respect existing processes. And by doing so, if I can overcome the challenges of the existing technologies off the shelf and provide reliable, accurate data capturing by speech and turning that into raw material, that's going to impact everything. It's going to shorten time for processes. One of the key elements I realize – it doesn't matter what's going on in the world right now, which is crazy, some verticals are immune, meaning, it doesn't matter what's going on in the world, we have to eat. So, food and beverage. It doesn't matter what's going on in the world, we have to have energy, supply chain, pharma, semiconductor, logistics. So, I can easily identify those verticals which have to go. So, I decided to focus on solving that challenge and starting first with must-have verticals – critical processes where we can demonstrate quantifiable ROI, return of investment at scale. So, rather than talking about great technology, it's not hey before you start your day – and this is a real example, by the way – you cannot start production before you complete what's so-called pre-op inspection. Great! What is pre-op inspection? Oh, pre-op inspection is 20 pages that inspectors at 4:30 in the morning, checking every single piece of equipment. And before that's being completed, you cannot start production. Production is idle. Great. And how long does it take you to do that 20 pages? About two hours.

Great. And this paper goes nowhere? Nowhere. So, what if we're going to turn the entire process speech-based, paperless, hands-free? You're going to walk and talk in your own language, in your own environment, and the forms are going to be filled in 95 to 100% accuracy. One, I think it's gonna be way more efficient. Two, it's gonna be safer because I'm gonna spend more time looking at the equipment and not on a clipboard. Third, very importantly, now I'm gonna generate tremendous amount of granular data that previously was uncaptured and unstructured. That becomes my raw material for the evergreen and ever-learning insights and intelligence. Four, it's collaboratively. It's no longer this production line, now this entire factory, this entire region, the entire country, et cetera, et cetera. And you can identify trends before they form. And they got very excited and said, 'Let's try it out.' So, I'll talk about the technology in a second. But once we tested the solution, within two months, it dropped from two hours to one hour. So, without adding a single piece of equipment or a single person, we just created another hour of production per shift or per day, which is huge, especially in this razor-thin margin industry. That translated to almost one million dollar per month of added production just by automating the process.

So, first was automatic impact to efficiency and productivity. Second, we took 100% paper that went nowhere and turned it into 100% green knowledge, which is also ESG solution. But now the second wave of impact came and started to connect the dots beyond what the human eye can see or beyond what the human brain can process. And all of a sudden, the data was screaming, 'You're starting the day with the wrong product. Because this specific product requires a lot of time to clean before you shift to the second product. So, if you put that product as the last product of the day, it will make better sense. So, your downtime is going to be narrowed significantly, et cetera, et cetera.’ And then we realize it's a massive opportunity, and once we solve it… Just to give you a sense: food safety alone, it's a 100 billion-dollar-a-year business, just food safety. And one of the things I realized to do here is to build something, which is going to be not vertically built, but it's a generic horizontal platform that can deal with any language, any accent, any acoustic environment, any jargon in any industry, on any vertical or any process. So, that was the concept that I tried to accomplish here and so far very successfully.

Oleg

Okay, great. It was a very detailed answer. Thank you for that. But how did the idea of aiOla come about? And what motivated you to focus on speech-powered process completion and data capture? Where did it start from?

Amir

Absolutely. So again, at the end of the day, the human brain is still the best tool. But unfortunately, the human brain is limited to what we call low dimensionality, meaning there's so much we can see and so much we can process, and it's fairly limited. So, I still believe in the human brain and the human nature, but I realized that now we have technology that can complement and can actually provide you an auxiliary power that allow you to process way more data faster, deeper, and then capture it, and then process it. But then, so the first step was like, and that was the first few years in the business, I realized the human brain, like I mentioned, is limited. We don't know what we don't know. And at the end of the day, we need to come out with better questions in order to have the right answer. And I realized that when we face a problem, we can ask.

You know, we come up with three, five, ten very creative ideas, but that's it. And I realize this is something that if we can build a hypothesis-generating engine, so whenever there's a challenge, I can generate more than one million hypotheses a minute and come up with ideas that nobody thought about before. That's when I started to get deeper and deeper into the world of AI. Because keep in mind, artificial intelligence has been around since the mid-fifties of the previous century, but the ecosystem around it was not ready, from compute power and unit economics, et cetera, et cetera. Now we have it. And actually, we can talk about it later on when the likes of OpenAI, ChatGPT, and others came around that, created another major jump, but also a major opportunity. Because what they are doing, basically, is in brute force, constantly boiling the ocean, and then you can get your derivative out of it. And we came up with a completely different approach, I'll be more than happy to share later.

Oleg

What inspires you in the field of artificial intelligence and technology, and what innovations do you see on the horizon?

Amir

Again, the innovation, you know, it was a very interesting article that was published, one of many, a few weeks ago at the end of 23, talking about 2023 as the year of AI or GenAI, generative AI, which is true. So, the hype is tremendous. And actually, everything around this area was an opposite trajectory to everything else. That was a difficult year in tech, difficult year in investments worldwide, and yet at the same time, anything associated with AI just took off like crazy. It's a fact. So, that was an incredible year in 2023. But they're talking about what's gonna happen in 24. And the smart people predict that 24 is gonna be another banner year, another incredible year for AI. But it's gonna be a different type. It's gonna be the maturity, which means it's gonna be less gimmicks and crazy stuff. It's gonna be more focused on practical solutions that AI can support to help organizations to do more with less, to let organizations be more successful.

So, it's gonna be very, very focused, very pragmatic, and very practical. So, the industry as a whole is transitioning right now from the gimmicky AI capabilities to narrowing down to very specific solution. We realized, actually, a year ago, that we don't care about the gimmicks; we want to go after the heart. So, basically, at aiOla, that's exactly what we have done in the very beginning. And in a very short month, with our unique approach, we were able to take it to the best of breed in every industry and show value and ROI in every industry. So, It's very exciting, but it's going to be a different type of year because the hoopla and hype is coming down to the harsh reality. How this is going to impact my business How can I generate quantifiable results? How can I take it to a very pragmatic approach? And whatever doesn't fall into those categories is going to fall away.

Oleg

Yeah, I totally agree. It's a logical step of every hype. First of all, everyone puts AI everywhere where there is nothing to do with AI. We did it just put AI. And then people understand that, 'Okay, guys, it's too much.’ AI is everywhere, when in reality it should be like in 20% of things we do. And then it comes to the step when people want to see where is value exactly. Yeah, but, it's logical step, I think, of every hype. So yeah, 2024 might be that year.

Amir

Yeah. And I think it is. And if you put it all together in a greater context, okay, when it's all about efficiency, it's all about productivity. I just saw how the tech industry now firing a lot of people. So, the growth funding for growth startups is, basically, paralyzed. So, there are many, many challenges, and I think it's going to be, you know, I call it a filtering or cleansing year when you'll be able to separate the contenders from the pretenders. So, the hype will be gone, but if you can generate... You know, I've been lucky enough to have wonderful, wonderful conversations with Larry Allison from Oracle just a few weeks ago. So, I had a one-hour session with Larry, and he was talking about it. Because now it's all about, it's an arms race. Everybody, the big guys are trying to find ways to connect it, and you can see what's going on from the chip industry to every part of the supporting component because this is a monumental point. It's not just another technology. It's not just a hype. This is a transition period. In my mind, it's like when the Internet was introduced for the first time, how we changed our life forever. This is going to be the same thing on steroids, if you will. But it's going to be that filtering process or cleansing process of taking the parts that actually can contribute and impact our life in a meaningful, quantifiable, measurable way. And I'm very excited to see where it's going. And I'm very, very proud to play an important role in the process.

Oleg

Definitely. What are your thoughts on 2024 in general, in terms of tech layoffs and the investment climate? Because the rates I expect going down, not me, but people expect going down very soon. That was the main breaker for investments and the main challenge for startups. Since it's going down relatively soon, what's your prediction on 2024 since you speak with people like Larry, other great entrepreneurs? It would be great to hear your thoughts.

Amir

As we are talking, I just declined a call from one of the biggest investors that I'm working with in New York. I wanted to have a chat. I've been lucky enough to work with some of the best in the industry. And I think you hit it on the nail to start with, meaning the cost of money went significantly higher. And forget about anything else. That's the single, that's a simple math. Okay.

Oleh

If an investor can get 5% from treasures, U. S. treasures, what we are talking about – I'm not saying about bonds from like HSBC where you can get 10% – why do they need to risk their money? That's the essence, obviously.

Amir

Exactly. Right. So, first of all, because of that, that's overshadowing, single factor. I think for the last 24 years, that's the most expensive money we had.

Oleh

Yeah.

Amir

In our career time, we haven't seen such a difficult situation. And, you know, keep in mind that some professional investors, some casual investors, some pension funds, some other areas that easily will go elsewhere because there're way better alternatives, safer alternatives, better returns, et cetera, et cetera. So, first, the pool shrank because of that. And this microeconomics, there's nothing we can do. That's a fact. Nothing we can do about it, even if rates will come down on their will. Second is the fact that there's still plenty of money out there. And third, without going into psychological elements of investors that now are basically sitting and waiting for bargains, because they're very good in understanding the landscape and understanding that a lot of great companies are going to struggle. So, no need to rush. We just wait patiently, and things are going to be desperate enough so we can get it in our own bargain deal inside. But at the same time, again, technology never played a bigger role in our lives than it does right now. And it's irreversible, meaning in order to continue to move forward, you need to have the funds to support it.I just think that what happened here two years ago, three years ago, in 2021 and 2022, was abnormal because too much money was spent everywhere.

And if that had to do it all over again, I can easily see that more than half of the investment should not have been taking place. So, what I'm saying right now is the technology is exciting, and the technology is moving forward, and the technology will be funded despite the fact that money is much more expensive. So, is that easy? No, it's difficult. Is that part of the cycle? Absolutely. But if you clean the financial side and just look on progression, when it comes to technology adoption and progression in our lives and everything we do, it cannot go back. You know, I always tell the team, when you move somebody in a desert to an air-conditioned room, there's no way you can bring him back to the fan. Okay, so there's no way we're going to turn off our phones. There's no way we're going to stop using ChatGPT. There's no way. And all of that requires continuous investments, but it's gonna be more focused, better funded, more realistic. And I think this is a time in a cycle. If you look back, historically, in every element of our life, there are always cycles. But art tech as a tech is the highest point ever, and it will continue to grow.

Oleg

In addition to your professional life, do you have any personal interests or hobbies that you are passionate about, and how do they complement your work or provide balance in your life?

Amir

Actually, I have many hobbies, but before the hobbies, one of the things I enjoy here is I always work in technologies that, you know, great state-of-the-art technology that broke many barriers, but unfortunately, very few people were able to use. Because you need to have the best data scientists to use it. And there's not enough data scientists, and data scientists far removed. One of the things I wanted to do different and I've done different here at aiOla is how I connect the promise of AI to the masses. I don't care what you do. So, for example, one of the things we learned is one of my clients is one of the largest logistics package delivery companies in the world that have 100, 000 drivers. How can I get those people to use it, like it, adopt it? So, one of the things that I enjoy the most here is at the end of the day, rather than me teaching them something new, I'll develop the technology that will speak their language literally. So, when they were not able to understand or do something on a computer, use what you're using. Take your phone and talk about, 'That's what I've done. I've just delivered a package to Oleg, and the package was wet and damaged. That's what I've done.'

And be able to capture this in any language, in any accent, in any jargon, and integrate it to whatever. So, one of the key things that I enjoy very much today is taking traditional companies, which are far beyond where they need to be on the digital transformation journey and maturity, and I fast-track them to the most advanced technology by using something that connected to everyone. So, connecting to the masses is something we do here all the time. In addition, one of the things I enjoy is sharing the knowledge with young entrepreneurs, sharing the knowledge with other co-founders, and people that trying to make a difference. And I encourage everyone to be part of the game. Beyond that, I enjoy very much sports. I play basketball for many years. I'm a tall guy. It's hard to see on this podcast, but I'm a tall guy. I play basketball semi-professionally for many years. So, basketball is still high in my list, but I spent many years in the US, many years in Silicon Valley, so I fell in love with American football. And so, yeah, those are the two sports I like.

And also I got into some motor racing, and I started to get involved with NASCAR. I don't know if you're familiar with NASCAR. In the beginning, I thought I'm not sure if it's a sport and always tell them call me when you make a right turn because they're going to the left all the time. But then, when I realized how much technology goes into those cars and how much efficiency, for me, it was an incredible thing to see from one end the technology that goes into the car plus all the science that goes into the driver and the combination of the technology plus the science in order to get results under tremendous environmental conditions in the racing environment. It's fascinating! So, I love that part as well. And I enjoy very much reading. I'm a self-declared info maniac, so I have to read all the time. And I enjoy reading different things, and there's a ton of useless information in the back of my head. But like I do with my technology, I constantly digest data.

Oleg

Where do you, where do you get the ideas what to read? Do you follow some?

Amir

Again, so now it's easier to follow on the recommendations of friends or colleagues.

Oleg

There is so much information and you need to be picky.

Amir

Yeah, absolutely. I need to be picky and not only that – there's a variety. So, I can read some serious literature. I can read some very light spy books and all the way to autobiographies and biographies in general. So, I love the fact. And I normally read in two languages. And it's confusing because one goes from right to left, and the other goes from left to right. It's funny: for the longest time, I took my notes half in English, half in Hebrew. So, you can imagine, like, going from left to right, then right to left, left to right, right to left. I think it's pretty good for your brain and good for a headache. But the key there, I think, this is part of also what I believe. I believe that each one of us has to be multidimensional. So. I'd love to have a discussion talking about sport. I can go very deep into technology. Even though I don't enjoy it much, I can talk about politics. I'm a history buff. I love history and wherever I go I'm always interested in what happened in that country thousands of years ago. And you know, where I grew up, I always tell people that Jesus Christ and I share the same zip code because I grew outside of Nazareth, not Nazareth, Pennsylvania, the real. It's a bit older, a bit older, but the same area. So, I was born into that ancient history. I enjoy very much reading about it. And more than anything else, I know it's a cliche, but I believe that all of us, and I take it personal, have a bigger role to try to leave this place in a better shape that we received it. And it's becoming more and more difficult lately.

Oleg

I have a question. Where do you play American football in Israel?

Amir

I don't play much, unfortunately, American football in Israel. But somebody like me, every Sunday night starting at 8 p.m., Israel time. I have the NFL package, so I'm watching the games from 8 p.m. until 6 a.m. on Monday, and not just me, my entire family is big football fans. We lived in the U.S. long enough to be the fan of the wrong team, which really testing our loyalty. Because our team constantly loses. But don't choose family and football teams.

Oleg

Okay. Fine. Could you mention any professionals or leaders from your network who inspire you in your professional journey?

Amir

Again, I believe that inspiring is the right way to use it because I'm not idolizing. It's inspiring, because I had the benefit, the fortune, or I was fortunate or unfortunate to meet some of the biggest names. I have always thought very highly of to meet them in person and then to be disappointed. Because none of us is perfect. I'll give you one, which is very, very clear, you know, the Elon Musks of the world, that inspires me – Satya Nadella. What he did in Microsoft is incredible. Because I had a chance to work with Microsoft before, and I saw the impact he created. And his personality is greater than this wonderful leader that he is as a person. So, Satya Nadella, for sure, Elon Musk with his creativity and craziness. Karp from Palantir is another guy that I really like because it's unusual, it's different, but he's very honest with himself. He has his beliefs, and he lives accordingly. Even here in Israel, I think the founders beyond the Mobileye, it's incredible what they have done and pushing the barriers and all the way to what Intel did here in Israel when they developed the chips for the first portable laptop, and the rest is history.

Those are the people that I'm really following closely and working with. But, I'm doing this service by mentioning a few because I've been lucky enough to bump and work with some of the smartest people in the world. Give you a very, very personal name that you haven't heard and actually working for me. When I was starting to do this work at aiOla and I came with a concept, at certain point it was very important for me to get some kind of independent expert to look in and say, 'You're doing something special.' So, I was searching for the best expert in speech I can find. And I found out that one of the very best in the world happens to be an Israeli, nobody's perfect, Professor Joseph Keshet from the Technion. Technion is like Israel's MIT. And actually, when I talked to him, he was involved in the beginning of Siri. And then when I talked to him, he was deeply involved with Alexa. And then, after spending a few hours with me, I convinced him to leave Amazon and to join aiOla as a chief scientist.

And he's now my chief scientist for the last two and a half years. And what he has done here, actually, we developed propriety patent technology that combine automatic speech recognition with natural language understanding that really made a huge difference. And that's something just to give you a sense as a point. OpenAI is an incredible technology. But when we talk about accuracy for the specific processes we're looking at, with less than half of the parameters, aiOla today get 23.8% higher accuracy than OpenAI. And when you're dealing with keyword spotting, Apple is absolutely the world leader, the state of the art in keyword spotting, they developed an incredible technology that recognizes two words predefined, which is 'Hi Siri'. And we were able to develop a technology for any undefined words with greater accuracy and less than half of the time. So, when I'm looking at that, and I like humble people, I like people that fly under the radar, and that's Professor Joseph Keshet, one of the best experts in the world. We are very, very fortunate to have him as one of our team members. So he's my Hansagi hero.

Oleg

That's great. The collaboration between human and AI is evolving, obviously. How do you see this relationship developing, and what role do you envision AI playing in augmenting human capabilities in the workplace? You've already mentioned that it's an amazing opportunity to augment, enhance human capabilities.

Amir

Absolutely. I think, again, that's the right approach to look at that because, at the end of the day, we are not to be replaced by robots or AI. The humans have a special role to play, but it has to be supported properly with a technology we as humans cannot do it. I'll give you a great quote, which I got from the chief strategy officer of one of the leading companies globally in IoT, Internet of Things. And when she saw what we're capable of doing, she said, 'You know what, in an IoT revolution one voice was left out – human voice. And actually what you're doing here is an interesting twist to use. You're turning every single person on a production floor to a super smart IoT sensor.' So, actually the twist here is while a sensor can tell you what's the temperature, or what's the weight, or whatever the environment, the person himself is much richer, multidimensional, and can provide so much more critical information if you can just capture it.

So, if you can talk more, do more. I'll give you an example. We're working right now with a company that has in the production floor people who immigrated from Russia, people who immigrated from Ethiopia, people who immigrated from Palestinians. And right now, all three categories have to fill up forms in English, and it's very challenging. So, actually, we're using the technology to allow the Russian immigrants to speak Russian, the Ethiopian immigrants to speak Amharic, the Palestinians to speak Arabic, and the information is flowing in whatever language you want. So now, you have the ability to capture that. So, I think that augmentation of the multidimension of the human and having the human as also a super smart sensor, I should call it, and actually now, getting the information in a better way through the human let the AI generate the data, and process the data, and come with the insights and intelligence, and then the human can pick and choose what to use and what not to use.

Therefore, you can see, for example, the ability that we offer today to capture surgery reports in real time because the surgeon doesn't need to sit down after the surgery and write down because, actually, aiOla can listen for the entire surgery. I know what is part of the procedure, and what he had for lunch yesterday, and to separate the noise from the signal and have the reports ready to go. Or dealing in pharmaceuticals when you have sterile production. And today, the person that's standing in a sterile production line cannot capture the data because he or she cannot write while they're with gloves and a mask, and the human twin needs to observe and write information. Well, try to imagine if she or he can just talk throughout the process and everything is captured. All the way right now to organizations that using AR and VR, augmented reality and virtual reality, and you can complement it with speech. So for me, I love to see how we marry technology and AI with human capabilities that right now we normalize. I'll give you an example. I came back yesterday from Switzerland. I was sitting with one of the largest cruise line companies, and I said that in every cruise line, they have workers wise, more than 100 nationalities. It's a modern Tower of Babylon.

Now, if you can streamline all the data capturing by speech, regardless of what language, what accent, all of a sudden, and that's the reason I got so excited about the angle we took. And actually, when everybody talks today about LLMs, which is Large Language Models, I realize that we can do it differently. You know, I don't need to go on brute force and take tremendous amount of data. And actually, I don't need to boil the ocean in order to make coffee. So, what we need right now is, actually, take a simple piece of process. I don't care what language is. And on our end, we're using GenAI to generate hundreds of thousands, if not millions, of synthetic samples. So, rather than going very wide in brute force and then narrow it down to the specific need, I go very narrow for the specific process, so I don't need LLM. It's domain-specific language model, or even customer-specific, or process-specific language model. And I'm able to build a language model that speaks the language in the specific jargon, specific process, specific acoustic environment, which is better than 95% accuracy, in two to three days, in any industry. So, I love it because while everybody's going in this direction, you know, we're going to start in a different direction. And yes, we're leveraging generative AI, but rather than going wide, we're getting very narrow, very deep, and very specific.

Oleg

Within the preview of using AI, ethical considerations become paramount. How do you think businesses should approach the ethical aspects of AI, especially in terms of data privacy?

Amir

There are several elements here. I think data privacy is one thing. I was joking with a good friend, which happens to be a very good lawyer, and I said, ‘Always the people, which are going to make most of the money, are lawyers.’ It doesn't matter what happens – always making money. This time, but it's going to be this different type of law because it's a new challenge for everything. I think we've already experienced that in the case of using right now propriety rights because the AI capabilities are running so fast, and connecting so many different data sources, and not providing proper credit, etcetera. You can build law firms, and I'm sure law firms will be built just to go after all these violations. So, I think one is going to be those legal issues. The second part is AI is everything on steroids, but it's also bias on steroids. So, if you misfit the data to AI, you're going to create even bigger bias than we're currently facing on the human side once it becomes artificial.

So, I'm not taking that lightly, and I think, by the way, what happened in Europe with GDPR, et cetera, makes things even more complicated. And people are going to start to see the implications of that because it's going to slow down. Without giving the name of the company, I just came up with an incredible idea for a Fortune 30 something company that loved it. And they said they can deploy it across the world. They said, 'But no, no, no, we're not going to start with Europe because it's way too complicated with GDPR. Let's start it in Singapore. Let's start it in Brazil. Let's start it in different places.' And I think that overprotection that GDPR created, again, without stepping in that landmine, actually going to negatively impact the progression of Europe because people will keep Europe to the very late because there's such a headache just to comply with some of the unnecessary headache that comes with GDPR.

Oleg

Yeah, some countries in Europe even prohibit using ChatGPT.

Amir

Exactly. Right. But it's like once you move from the horse to a car, you're not going to move back to a donkey. It's too late. You know, it's gonna take longer because 'Oh, I don't see you anymore. You're not there.' Of course, you're there. You know, preventing from sticking your head deeper into the sand doesn't solve the problem. But what I really don't like is the fact that we created too many unnecessary obstacles along the way. And I'm always looking for three elements: where I can make the greatest impact, the shortest period of time, minimum obstacles out there. And some of the countries around make it difficult. But going back to your question, ethical AI is a massive problem that is going to get even bigger. And we need to think deeply about it. It's a major challenge not solved yet.

Oleg

Got it. Larger enterprises often have the resources to implement. How do you think the AI technology can be made more accessible and beneficial for small and medium-sized businesses?

Amir

Exactly. Right. Because, you know, many organizations say, 'Hey, I cannot start before I have a data lake, or data pool, or whatever.’ And that's part of the approach I took. I think the challenge is the approach, because you can never be perfectly ready for that. And the needs are the same. It's not exclusively just for large enterprises. The small and medium needs it, but they don't have a CIO, or they don't have a chief analytic officer, and they don't have data scientists. Going back to the point I mentioned earlier about what I really like to do is I can connect, or we can connect the power of AI and impact organizations, regardless of where they are in the journey, and regardless of how small or big they are.

So, all of the other parties, in my opinion, are excuses. It's a mindset and ability to execute because you can do it. If I've just met Oleg and company, and you say, 'I have a tire company in Warsaw. And this is the product process of doing Polish, and I want to do it faster,' I can get it to the point that you have a language model to your specific process, regardless if you're using any technology, or you have a team or don't have a team, in a week time. Which means in less than a week, you're actually gonna have a language model that allows you to do the same process but because of the power of AI more efficient, safer, smarter, in a way you can measure the impact. So, the time to impact, the barrier to impact – we shrank both, underlying any organization in any industry in any size to benefit from not to work.

Oleg

Amir, could you please comment on the problems associated with the lack of qualified specialist in the IT sector, particularly in connection with your business?

Amir

Again, at the end of the day, I believe what happened now with a fast acceleration of technology, in GenAI specifically, it doesn't start and ends with data scientists. This is the tip of the spear, basically, because at the end of the day, you have to go data scientists, to data engineers, to MLOps, and all these IT components. And, actually, the ecosystem was not ready, and it was impossible to be ready to grow as fast as technology is expanding. So, I think right now, the good news is that with AI specifically you have to do it with only data scientists. There were not enough data scientists to start with. And if you're lucky enough to have some, they're far removed from the business. Now we got to the point that data analyst or citizen analyst, or whatever you want to call it, but now more and more people can be part of the game, but more and more people are part of the game. You need to be able to support it with the ecosystem around them. And that's really where the challenge is. So, it's going to be an adjustment time when companies try to ramp up, but you cannot ramp up fast enough to accommodate the expansion of adoption that goes from a very few to the masses now. But without proper IT infrastructure and without proper IT support teams, you cannot do it, unfortunately.

Oleg

Could you share how the development team is usually structured in your company, engineering, development? Have you ever outsourced your techniques to an external vendor?

Amir

Yeah, absolutely. So again, at the end of the day, we are deep tech company. The core, the majority of my company is the tech side of the organization. And it's really broken down to three major components. One is product, and around product, of course, we have a very strong product. The second is the research team. We have a very deep research team, specifically around data science. So we have a very, very strong team of data scientists. Both teams are in eyes. But then we have R&D and engineering, which include, of course, all the MLOps components. So, all that elements right now we started from the very beginning, outsourcing developers initially in Ukraine. And unfortunately, two years ago, when the war started, we had to adjust on the fly, help some of our team members, but expanded to other parts of former USSR or East Europe in general. So, we had some in other places, like Poland. Now we even signed up with a team of developers in Greece. So, we still continue because it doesn't matter how fast we grow, and we grow very fast, we still need to support our efforts. And we did it from the very beginning with high-quality outsource talent that can complement what we're doing internally, specifically around engineering and R&D.

Oleg

What were the precise factors that prompted you to consider IT outsourcing?

Amir

The key component is very simple. Again, we had to scale fast and there was no way that we'll be able to hire the people we needed as fast as we hire and outsource resources. And second is the fact that we didn't need time to ramp up people and train them so we can take highly-qualified, skilled personnel. And actually, even at product, at one point, we had to accelerate, and close some holes, and plug some holes. So we hired an outsource in Israel for product to help us specifically in UX/UI. But it was very simple. First – speed. Second – ramp-up period. So, we don't have the time, we have to jump right away with highly qualified. The third – unit economics. The labor market in Israel is very, very expensive. And for those roles, the price was just unacceptable. So, we had to find something that will address the speed, the scale, and seasoned and proven people, but in a reasonable price, something we can accept. So again, those were the facts, and that's what we have done. And we're still using a substantial amount of outsource in R&D and engineering right now.

Oleg

What are the benefits and drawbacks of outsourcing as per your opinion?

Amir

Again, you know, one of the major challenges when you do outsource, it's not part of your office. And we are very proud of the fact that the majority of the team is working on site most of the days in the week. So, there's a lot of benefit of working together on the same side: collaboration, culture, et cetera, et cetera, which is very difficult or challenging to replicate when you do it outsourced, especially when you do it outsourced internationally. Second, at the end of the day, you can outsource tasks, you can outsource projects, but it's especially when it's an IP-based company, like we are, the core IP has to stay within the company. So, that's another challenge that we could not really lead, you know, with all the respect to our wonderful outsource resources. We cannot let them lead key project and initiative for us. So, one is limitation in understanding that we're using that to complement.

Second, we understand the drawback of they're not with us physically. And it's always going to be remote, or in most cases it's going to be remote. We can visit them from time to time, but not more than that.And I believe, you know, the British have a great phrase that says, 'Culture eats strategy for breakfast.' And culture for me and for us is very, very important. So, the fact that I cannot see the people, I cannot touch the people, I cannot, say good morning and have coffee with them in the office, it makes it more challenging when you do it outside. So, we understand the pros, we understand the cons, and that's the reason we're still using substantial amount of outsourcing. But we had to live within those limitations, and define it to the right rules, and set up proper expectations. And that's the combination when we're doing that.

Oleg

Thanks for the detailed answer. How do you measure the success of your collaboration with an IT outsourcing vendor?

Amir

Again, at the end of the day, we don't compromise on excellence, and we have very clear KPIs, and we have a very clear set of measurement tools that we use for quality, for delivery. So, several things here. One again, not necessarily in that order, is a timely delivery of tasks to make sure that we're moving. We're playing a multidimensional chess game. And if one of the components is not moving properly, it's hitting everything. So, it has to be either orchestrated or synchronized, has to be so timely delivered, super critical. Accountability – very important, quality measurement, because if I'm getting a product, which is full of bugs, and I have to spend a lot of time with QA on my end, I defeat a purpose. So, I have to make sure that whatever outsourcing resources I'm using, it's up to par with the same standards of quality that we use internally. Then we have to make sure that we do it in a timely manner. Then we have to make sure it's working in a fully synchronized part of the other part of the organization. And we have metrics we apply internally, and the same we apply externally for any outsourced source. So, we don't compromise or nothing short but excellence.

Oleg

And finally, what advice would you give to other companies considering IT outsourcing?

Amir

The key there is setting proper expectations because, especially if it's a deep tech-based company, you want to use it because I think it's very wise to use it for all the reasons I mentioned before: acceleration, economics, the ability to expand and retract. Because not always you need the same amount of people. Sometimes you have short-term projects or medium-range projects, but you don't want to hire people full-time for that because all the hustle and bustle of hiring and then firing. That will allow you to have complete flexibility to adjust to different times, and flows, and challenges. So, more than anything else is the realism. You cannot expect the outsource to do what senior management unit teams supposed to do because you want to protect it. You want to make sure you don't compromise on quality, you don't compromise on accountability, you don't compromise on timely delivery.

And again, at the end of the day, from my experience is work with people that have a lot of experience in working with similar people like you. So, if this is a UK-based company that using outsourcing somewhere in Central Europe, make sure that they've already worked with other UK-based companies to understand the UK culture and everything else. So, the key there is make sure that they have a reputation for Israeli entrepreneurs, have a unique DNA and fingerprint. Okay? If you go to a place that never worked with an Israeli startup before, that could be challenging in the beginning because it's very different than entrepreneurs from Copenhagen or entrepreneurs from Berlin.

So, that culture component I was talking about. You don't have the ability to manage them on a daily basis in the office – that's already a challenge. And if they haven't worked with similar culture before, that's a double trouble and a recipe for disaster. So, proven, highly reputable, with experience with similar companies and similar culture. Make sure you set up the proper expectations about what to use them for and what to refrain from using. And making sure that, you hold everybody accountable to the same standard to use internally. Because at the end of the day, it should be a natural expansion of your internal team. If it's anything different, you failed your task by hiring the right people to be your outsourcing.

Oleg

Great. Amir, really thank you for this longest episode. That was the longest I've ever recorded so far, but the answers are super valuable. I really appreciate for giving insights, for sharing vision. You're a very interesting person. I already agree with Julian. Yeah, that was really amazing. Thanks for your time.

Amir

It's my pleasure, Oleg. Thanks for having me, and I enjoyed the conversation and wonderful dialogue. So, thank you for that.

Watch previous episodes

Contact us for a free IT consultation

Fill out the form below to receive a free consultation and find out how Devico can help your business grow.

Get in touch