#303 Building SaaS Products with AI: Lessons from Hakob Sharabkhanyan's Entrepreneurial Experience

0:00:01 - Mehmet
Hello and welcome back to any episode of the city or show with Mehmet. Today I'm very pleased joining me a guest who came from Armenia. He's currently in the UAE, but going back, but I love to speak always to fellow city O's and entrepreneurs. So, hakaob, thank you very much for being on the show today. The way I love to do it is I keep it to my guests to introduce themselves and tell us a little bit more about what they do. 

0:00:27 - Hakob
Okay, thank you, mehmet. So I'm Hakaob XCTO a couple of successful and unsuccessful startups, and currently I'm the CEO of Hakaob. I founded my company when I was 19 years old and grow into a team of 80 people. Right now, we are mostly working with Western clients having pretty big multi-billion dollar or public companies in our portfolio, and I'm currently a little bit out of the CTO role. I'm more on a CEO, executive role, but I still have my technical knowledge and I still have my passion into technical work. 

0:01:07 - Mehmet
That's amazing, hakaob, and I'm interested to know a little bit more now during our conversation about this. So the first thing, what attracted you to the tech and then how this transition happened that led you to start HACTech today. 

0:01:31 - Hakob
Yeah, actually I was having patience since my early childhood. At that time I was more like, oh, I'm going to build games and this kind of stuff. But then I entered to the university applied mathematics and informatics and there it really started to grow into the career, into the profession. At that time, when I was starting HACTech, there were a lot of IT companies growing. It was this boom of outsource and outstuffing. I clearly could see that there's something missing. There is a lot of people working but creating less value than it could be created. So that was the main driver. 

I found it HACTech to bring something new to the outsource world. That's what we are positioning ourselves now. So we are not an outsource, neither an outstuff agency. What we are doing is the software engineering partner. So we are basically becoming the team and the main difference here is that we are coming with our processes of product management, project management, engineering, best practices and as a team I believe we are creating much bigger value than you can create just by outstuffing a couple of developers or outstuffing a PM. So that's what was the main driver to start the company. 

0:02:48 - Mehmet
Now I would ask you something, and because I interviewed a lot of people and they had different point of view. So do you think, as start up, I need a full time CTO, full time product manager at the beginning, or is it like something that they can do it later on? And this is where actually your role comes into the place. Right, and tell me a little bit more about the struggles. I know that many people know, but just to highlight for someone who might be he's not technical and he wants to start, what are also the struggles they face with offshore teams usually? 

0:03:32 - Hakob
Yeah, yeah, actually I love this topic. I call this like how to run a company without the CTOs. So if you are a startup, if you go to outsource or outstuff, you need to have a CTO because, like in outs, you are just getting, like people, but you don't get the knowledge, you don't get the base, so you don't know how to run these people. Cto is a person who should coordinate, put processes, put some quality gates, some infrastructure, architecture, etc. And if you go with outsource, you are getting something done but you don't own the knowledge again in your company. So I think, as a startup, if you don't have a CTO, you need to work with a company that coming with the processes and implementing it into your company, and that's what we are doing In our partnerships. 

We create the notion or confluence with all the information, all the processes, we create the CI CD pipelines, we put the right product management processes, because I'm also very big fan of product management myself. I believe that even if you have amazing development team, if you have a bad product management standard or like no standard at all, then you will end up with a great product that no one wants to use. So, on my perspective, product management, product manager, is a very important role in every startup. 

0:04:59 - Mehmet
So the question, haco Black, here is when would be the right time to hire a full-time studio? So, because you are providing this at the maybe baby stages, let's call it. So when would be the right time later to have a leadership within the startup itself? 

0:05:23 - Hakob
Yeah, I think it's either when you are starting to create multiple product lines or when your product is. So for me, cto is more executive person. It's not the person who should go and like check the code, accept merge request etc. While in many companies CTO doing that job. So CTO should be more strategic position and as soon as you need that person to go check for companies to mergers and acquisitions etc or create a long-term technical vision, that's where you need the CTO. In other cases, companies might call it a CTO, but it's more like a director of engineering position or engineer and lead position Great. 

0:06:06 - Mehmet
That's, you know, looks like very logical to me, and you know I love when you said like the CTO doesn't have to write the code right, so this is well said. Now I would love to you know, because you said, like you have done some startup, some of them have they failed. And you know, today the topic usually that people wonder about. You know, on top of mind of everyone, I want to build a SaaS product right. So, based on your experience, what are the common mistakes usually? Have you seen, you know people do when they are building SaaS products, development and on a high level also, like how these mistakes can be avoided? 

0:06:50 - Hakob
Yeah, in our case, I would say the most common failure was the lack of proper product management. So one of the startups where I was co-founder and, like I, was the CTO was a solution for Europe. It was similar to Twilio, but we were having, like some different approaches, different building processes, and it was pretty good until GDPR, because our model was completely relying on some phone books and phone numbers that we were also providing to the customer. So the main reason the product failed was GDPR. But the reason why it didn't like go very well not to rely on this one GDPR model was a bad product management. So we were not concentrating on the core and, I think, on site as a service type of business. 

When you are startups you have limited budget. You need to always go with only building what really matters. So, like we ended up creating tons of gold, tons of complicated logic to find out the cheapest SIP provider, etc. Which was not making sense at that point because we were not dealing with like millions of dollars to try to save some five or 10 percent there. So this was one of the mistakes that we started to think too much about the corner cases that really don't matter on a startup. Another big mistake was that we were just doing on our bias, so I think these users will not like this. We need to change this, we need to change this. 

And we ended up redesigning the whole platform which was very big platform, I think three, four times, completely redesigned without getting any feedback from the user. So it was just our opinion that it's bad. We need to redesign to make it better. I think these are the common mistakes. When there is no product management processes, there are no user. You are just building what you think is right and then you are not able to sell and you are just thinking that it's bad design and going and investing more and more on redesigning. And another mistake is that, like you are not concentrating on the exceptional value that your product should give. 

0:09:03 - Mehmet
I love this, you know, and a couple of weeks back someone asked me this question. I said you have to go out. So people usually ask how we can first validate the idea and second, you know, start to take these feedback. So any kind of a success recipe. I know it would be different from product to product, but I mean how they can start this actually even from the early stages, from validation. 

0:09:29 - Hakob
Yeah, I think it's not super easy to understand this on validation, because when you are talking to the users, they might say, oh yeah, we like this idea, but actually, when it comes to using, they are not so eager to use. 

For me, one of the best things in successful startups is continuous delivery. So, for example, aws, amazon, like if I'm not mistaken, they are releasing new changes every one second. With this continuous delivery mindset, tools and techniques, what we are doing is we are releasing to production every day, every two days, and this is giving you a big advantage, like because in traditional scenario, let's say, you do a sprint two weeks and then you release something, and then you go another two weeks trying to find out your customers love it or not, and then you go to another cycle then putting it into the backlog, the changes, etc. With continuous delivery, you can release every day. You can release something today, get feedback tomorrow, add it to the next item in the backlog and then you shape it and so this feedback loop, the most important thing to shorten this feedback loop, the delivery loop so you can build the product that your clients really need. 

0:10:43 - Mehmet
Yeah, 100 percent. I agree with you on this and this sometimes also, I heard from some people People get excited at the beginning and they tell them yeah, yeah, you know it will be nice, but when they launch it, no one comes to it. And I think this is where there are different techniques. So, whether you maybe let them pay free, pay maybe the offering or like, maybe pre-book it sometime, so it all depends to see if this is where they will be willing to pay money for that. Now, something I want to ask you here related to, like the culture. Okay, so, because I know, like you, you, you, when I was preparing, I've seen you talk about this topic as well. So how important is in a startup also to build the culture that would help you to keep attracting the talents, because, especially today, it's very competitive. Like engineers, they can shift from one company to another very easily. So what's your philosophy, I would say, on this one? 

0:11:44 - Hakob
Yeah, it's definitely the most important thing. Yeah, even if we are from technical perspective, not even from retention perspective, if you want to do continuous delivery use you should have a like super good culture where people don't afraid to make mistakes, because continuous delivery itself means it's like releasing every day. It means that there is a bigger chance of failure, even though you are putting a lot of gates, like security gates, quality gates, etc. But it's still like can cause a lot of failures. And as a company, you should be. You should have not blaming culture, you should be culture. 

So I think in IT overall is the industry where you can't do anything. 

You can't push a person to work because otherwise you need to put like 10 pm to measure their work, understand if they are performing well or if there is some slight decrease in their performance. 

So the best way is to have a culture where people really want to work, where people really are getting motivated by the impact they are making. So this is another thing that a tactic in our model we appreciate is that in Outstop you can't control the culture because, yeah, you can put the values on the wall, you can talk about them, but you have these individuals like 80 people working on 80 different products and you can't control the culture in that products and the person is like working 8 hours on the client's project. So even if you have very nice office with nice people, it's still the culture of the project they are working. In our case, with the team of 80 people, we have just nine projects right now, so we have a big team on each project. We are selecting our partners based on the culture. Like we are talking to them, seeing if they think the way we think, if they are open to this model of work, and only then we are working. So I would say culture is one of the most important things in this business. 

0:13:37 - Mehmet
That's 100% also very resonating with me, because the feedback I receive when I talk with engineers they care about the work environment and they want peace of mind, Like it's not the perks and the benefits anymore, it's like some place where they can feed their home. Now let's shift a little bit to AI and, of course, on this podcast we discussed AI maybe more than any other podcast ever. Like every episode almost, I asked something related to AI. But now, from a software development perspective, do you see AI as a thread? Do you see it as a tool? You can leverage it. What are your, I would say, opinions on different aspects of using AI with building software and even for the model that you use today? 

0:14:37 - Hakob
Yeah, I'm looking on AI as absolute supporting tool, not a thread, and I'm comparing this to the era when the electricity was invented or when the computers became accessible, because AI is not something new. Like the GPT, models are there for a very long time and as soon as it became a cloud, as soon as it became accessible I mean I was playing with GPT two couple of years ago and this GPT-3, when it was in the cloud, it was accessible to everyone it started this hype that everyone started oh, this is writing so good text. It's going to replace those. But actually, if you think of it, it's just text generation. So it's just generating text. 

It doesn't have the soft skills that emotions people have, which makes the biggest difference. So it's just a tool to make people more effective. I'm sure it will push us to the point that will create more value. It will push us to the point that maybe it will work a little bit less, but it will never be able to replace human, the same way as the computers didn't replace human, but they make like human more productive. So, for example, if on an accounting, maybe like 30 years ago, it was necessary to have 30 people, now you can have three people because of the computer, but they didn't took the jobs, it just shifted people to different new positions. It opened up a lot of new jobs, so I'm super positive about the AI impact. 

0:16:04 - Mehmet
Out of curiosity, because in software development, in building any software product, there are a lot of things people think about it, only the coding part itself, which is one part of it. So where do you see the AI contributing much more? Is it on, let's say, the framework part? Is it on the QA part? Is it on like, where do you see it like? At least in the couple, I would not say years, at least in the couple of few months, maybe one year from now, taking over in a sense? I mean we as human leveraging it? And are we, you know, later, seeing AI doing let's say, maybe it will not do 100% of the jobs, we know this but at least maybe doing 80% of the job? What do you think about that? 

0:16:55 - Hakob
Yeah, actually, when talking to AI, again, most people just thinking about GPT, about generative things, but I think that AI potential overall, the other, completely other type of models the potential is much bigger. So one of the biggest places that I see AI can contribute is the data Analyzing because, again back to the product management, the most important thing to analyze the data, understand what your users doing, understand what your users feeling and sometimes, even though there are tons of tools that supporting these heat maps etc. That's still, as a human, very difficult to understand all these data and even though you are like merging it with Tableau, with other like these kinds of PI tools, trying to get something is very time consuming and it's not giving 100% picture. So I'm sure with AI, with this kind of data analyzing models, at some point we can have much better picture about what really needs to be improved in our website where our users really struggling. So I would say the biggest contribution that I see can be generated is data. 

0:18:04 - Mehmet
Yeah, that's a good point. Now, any by the way, just like one quick thing about what you mentioned when it comes to AI, people think AI is only chat GPT and yes, ai is not chat, gpt only chat GPT is like one application of the AI thing. So thank you for reminding us about this hack. What are like other technologies you are seeing really, you know might Give the same impact same as what GPT have gave us lost here and the year before. Actually, anything else. Are you seeing any trend you're closely following other than the AI? 

0:18:48 - Hakob
Yeah, actually, I would say again related to AI. What I really like in this recent articles that I was checking is the AI tools working with voice. So I think from now on, everything will be related. Whatever there will be credit people somehow related to AI, it will have some AI components. So I really like this voice voice changing technology, which I find really really useful. 

So there are these AI trends that I more can real time change your AI. I speak like Armenian accent, you speak maybe Emirates accent, so there are tools that can real time convert this, which I find very interesting, very useful alongside with regular AI, even GPT tool, so you can build some nice applications for call centers. You can write some nice like bots that are making calls closing to human voice. So I think the biggest trend will go from text to voice video, which already we are seeing many tools that can create video, many tools that can create voice, et cetera. And another big potential IC will become the tools that will be kind of anti-AI. So it will be like finding out GPT or written content. It will be finding out these deep fakes because it's a real threat to the famous personalities. You can use AI to replicate their voice their video. At some point this is going to become a problem, and the anti-fake, anti-gpt tools are going to be very useful at some point as well. 

0:20:33 - Mehmet
Two things, and I think you spotted that very right. So the first thing is regarding the use of voice. So we started to see these products. Back in December, there was this human, the device that supposedly they want to market it as a replacement to the phone. We started to see some other form of products I mean physical product that they are AI powered and it's all about interacting with AI, whether with voice or with video. So I think, yeah, this is very interesting. Time will show us if they would succeed. So, people, because they are liking, I think, this customization the AI feels that you feel that it's talking to you only as Hakob or me as Mehmet. So this is interesting. 

And the second thing you mentioned about deep fake and how AI will be used to deep fake, and maybe a couple of weeks back by the time this would be aired, like it would be almost one month maybe, or one month and a half. So there was this story about how they use a deep fake of a CFO to get I think it was a big amount of money $25 million or something like this from one of the banks and it was like, wow, everyone was talking about it. It's becoming harder and harder to spot these differences, but, as you said, time will show us. It's very interesting times. Now, hakob, one thing before we close. 

Actually, this question came to my mind and because you are actually doing it from your side. You mentioned about the CTO role, like maybe he should have, or she should have, knowledge in the infrastructure and there are a lot of debates currently about the cloud and which cloud to go to. So are you seeing more a tent to go with one cloud, multi-cloud, a mix between multi-cloud and maybe some still on premises? What are your observations, I would say, from that part? 

0:22:47 - Hakob
Yeah, actually, as we are mostly working with site as a service type of companies or some digital transformation companies, we are always going with cloud, so we never have real dedicated in-house servers to work with and I think there is just no point, even the very. I know that banks in Qatar started to move to cloud, so it's a big sign that there is a very big trust towards the clouds and I think it's just not making sense to spend so much resources. Again, this is the part of concentrating on what really makes sense and concentrating on core value that you are giving. So there is just not making sense to build the cloud from scratch. And, by the way, about the startup that I told when we were doing for the ever of this voice over IP thing I created, I was also coding that time because, like it was six, seven years ago, I created the infrastructure which was automatically scaling based on the calls number, and it was like I think I spent three, four months purely on this one task which could be done, like which is already built in AWS and every other cloud, so it was just not making sense. 

In terms of multi-cloud, if you need like 99.99% off time, then it makes sense to go to multi-cloud solutions, but usually in our projects we just go multi-region, so we were 95% in AWS cloud with our clients In AWS we just use multi-region infrastructure which is again giving very, very high scalability and we never faced any big down times et cetera. So unless you are again building some projects where there is absolutely necessary to have some backup of the backup of the backup, then I think one cloud with multiple regions should work totally fine. And in terms of trends, like at least in US, I'm seeing big trend towards AWS. In Middle East I'm seeing a big trend for Microsoft Azure just because Microsoft having these local servers in Middle East and I think it's giving almost free for startup. It has lots of programs. So yeah, I'm seeing a big trend in Middle East to go to Microsoft Azure and US to AWS. 

0:25:17 - Mehmet
Sorry to your point. You're right. So for the hyperscalers, so whoever was the pioneer in being in one country, so they have the edge because they arrived before. So 100% this is what we are seeing, although here in the UE, majority of the hyperscalers are here. So we see Microsoft, we see AWS, like, as we mentioned, in Qatar. It's like now more Microsoft in Bahrain because AWS they start their first region there. So it's like AWS, now Google in Qatar and Saudi Arabia. So it's interesting times of this cloud migration projects that will go on. Small advice from my side if you're just thinking about it as a shift and lift, think again because otherwise your bill will become very high. So you need to plan it well. Just pushing the servers to the cloud is not the thing you need. Maybe to think about the re-architecturing also of your applications over there. Now have just final advice from your side to fellow entrepreneurs and also where they can find more about you. 

0:26:30 - Hakob
They can find about me in LinkedIn. Just, I would say super responsive there. So, anytime, just any advice I would be happy to share. And final advice is just again to pay very much attention to product management processes. If you are just CEO, not having CTO or product manager, you need to take some education to learn how the right product management being done. And it's not super complex, it's just the processes that you need to understand and you need to follow. So, yeah, this is my advice to, because I'm seeing this mistake so much in our clients that they are coming and, hey, our CEO wants this button to be powerful and there is no logic behind it. So it's super important that if you are IT company, or if you are a company that wants to go to digital transformation and become an IT company, you need every person in the company should have some level of education to understand what is, because IT is different culture, different environment. So, yeah, that's my advice to other entrepreneurs. 

0:27:38 - Mehmet
Thank you very much, jacob. I will make sure that also your LinkedIn profile and the link to your company is in the show notes. Thank you very much for the time today. It was like a pleasure to talk to you and the topics we discussed I think they should be exciting for fellow I would say startup founders, entrepreneurs and even tech people. So thank you very much for sharing your insights, jacob, with us today, and usually this is how we end each episode, so this is for the audience. 

If this is your first time here, you just discovered this podcast. Thank you very much for passing by. I hope you like it so you can subscribe on all your favorite podcasting platforms. And if you are one of the loyal followers that keep sending me their notes, their messages, thank you very much and keep them coming. I really appreciate that and, as usual, I repeat this at the end of each episode. If you are interested to be on the show, you are working on something different. You have an idea you want to talk about. You have a new technology you came up with. You're doing something that might help other founders. I would love to hear that from you, so don't hesitate to reach out to me. You know where to find me and thank you very much for tuning in and we'll meet again very soon. Thank you.