The AI Takeover: Insights from a Chief AI Officer

Arham: Hello everyone and welcome to another episode of the AI and Beyond for Business podcast! Today we have with us Yuri Yakulev, who is the Chief AI Officer at Zazmic. This is probably not a designation you’ve heard very commonly over the last few years, but I think at the moment it’s probably the most important position in a company. So welcome Yuri!

I want you to just kick off the podcast by sharing your reflections from the past year of AI. Did you expect 2024 to be such a significant year for AI, and did you expect this growth?

Yuri: First of all, hello! Thank you for inviting me to this podcast. Last year was promising, and I think we achieved most of our expectations. We spent the most time working on the expansion of our AI experience, working on prototypes, proof of concepts, workshops. We increased our partnership with Google in that direction, so that was promising and we fulfilled that.

Arham: Awesome! And now before we get more into AI, I want to talk a bit about the origin of Zazmic because you were part of that founding team at the beginning. We’ve already gotten Yann’s side of the story. Yann was in the first episode, if anyone wants to go check that out. But what was it like back in 2013 in San Francisco? You know, all of you started from a garage and now you’ve made this into a multi-million dollar company. So just talk us through that story.

Yuri: Yeah, it was different. Times were different. We started back 10 years ago with the remote-first approach. It was something unusual actually. When I met Yann, I already had about 10 years experience in software development in different roles – as a developer, as an architect, as a manager. Until that time, I already had an idea to set up my own company where I would build processes as I see them. And then I met Yann. It was a recommendation of our common friend. That friend recommended me because I successfully helped him to deliver one of his projects. He got a lot of investments then. I will not name that project, and then we started with Yann. That’s how it was.

Of course, it was difficult to start. It’s always difficult to start something new because only you know how to make it and you have no definition of whether you will have success or not. Yann was always the visionary in our team, and I was always the technical specialist. So that’s how our combination works until today.

Arham: Could you talk a bit about that chemistry you and Yann had because Yann was, of course, that visionary guy who told you what to make, and you were, you know, the tech genius at the back making those things. So how did you guys work? Could you talk a bit about the relationship you guys had while working?

Yuri: Yann doesn’t admit that, but he is a great sales manager. He loves to talk to people. He increased his network significantly, while I’m mostly an introvert. I work, I don’t like to talk too much, but recently that changed. With our years of experience in Zazmic, all of us got this experience, improved our skills.

It’s never that Yann is telling me what to do. He brings opportunities. He’s the visionary in the company. He expands his network, he brings new opportunities, new clients. We brainstorm how to do it better, and it is open-ended. It’s always open-ended. I implement this. It was at the initial stages, of course. Now we have 400 employees and multiple projects and departments. We are doing every project differently, but initially we started like that and still continue.

Yann was never happy with the achieved results, so we were working up to 200%. He’s the person who wakes up at 5:00 AM to not miss important meetings even if it’s not a convenient time for him. The same he expects from everyone. So we didn’t have any issues with our company, at least between the two of us. It was normal to wake up at night, early morning, to solve one of the tasks. All the tasks were critical because we owned the responsibility of the success and the result.

Arham: Fantastic! That sounds like a killer team combo to me. I wonder, if you could perhaps sum up very briefly from that almost 11-year journey now, what is one thing which you think you’ve learned which is very valuable to you or something you’ve held for the last decade?

Yuri: For the last decade, I’ve learned how to test your code, how to evaluate the definition of done. I’m not asking clients if it is good to go. I’m not asking clients about the feedback, is it well developed, because I tested myself. I always test the quality of produced deliveries. That’s very important. You should put yourself into the shoes of your client. You should wear the hat of the client and the customer of your product and think, is it done well? Would you use this? What kind of feedback would you have? And that’s very important.

Arham: Zazmic has been at the forefront of AI for the last few years, both internally and externally. I want to talk a bit about where that started from. What was the first time where you really saw the potential and power of AI?

Yuri: A couple of years ago, we had one potential client who asked to provide an estimate for a mobile project. Usually I spend several days preparing the architecture, estimate documentation, but this time I decided to try to build a prototype using AI. I was impressed how fast it was. I spent the weekend, and in a couple of days I had a working prototype which I shared with the client. Of course, there were still some improvements to be done, but the initial phase of development of that prototype took so fast!

And I shared my feedback with my colleagues, and we started to think about how to do it better, how to expand this, what else could we use. And that continued to expand, and now we build a lot of plugins, development tools, to improve performance of our software engineers. We develop multiple tools for Visual Studio Code, for IDX, WebStorm platforms to boost productivity. And not only develop, we do it for automation of other departments as well.

Arham: I want you to talk a bit about this idea that you constantly also talk about within Zazmic, that is, to calm the fears of a lot of your engineers, that AI won’t end up replacing you, but people who use AI will replace people who do not use AI. So can you talk a bit about that?

Yuri: Yeah, of course, AI will not replace people, but it makes life easier. At the company, we are practicing the continuous improvement model, so we always look for weak chains or processes which could be automated, where people spend enormous time, and that could be improved using tools instead of manual processes. I think that’s important for companies to scale, to grow, to automate processes.

If you’re talking about marketing, recruitment, there are a lot of manual processes which could be automated. If before people did that using scripts, tools, now we use AI to produce some content, documentation. So multiple marketing teams, recruitment, even migration teams are using AI tools on a daily basis. Finance team. So that’s all which is possible to automate. Multiple examples in different domain areas where those tools could be used. Still people don’t lose their job, they continue to work, but they work better and faster.

Arham: I think one thing I’ve really picked on all the users you mentioned of AI is going against this myth about AI that it’s kind of limited to just chatbots or very advanced chatbots. So if you could share any more specific examples of how AI is being used, perhaps by software engineers, and you can mention any tools you use. Where does AI fit into your workflow at a more granular level?

Yuri: I started with chatbots. So just to mention that if before developers in daily work did use a combination of keywords for faster code development, now they use chat. That’s a new interface which is provided for software development.

But let’s think about cybersecurity. What about AI-based security which will prevent spam and scam messages? It will recognize the message and understand that that’s spam. Of course, it works both ways. Maybe you noticed now we get a lot of personalized emails, cold emails, from different marketing teams of other organizations who also use AI tools. So this works in both directions.

But what else could we use AI for? To summarize information. Imagine the recruiter is doing a pre-screening meeting with the candidate, and we need a fast way to understand if this meeting was successful or not. As for developers, it’s not only chat. We use tools to generate auto tests. Most developers know how important auto tests are for reliable and scalable products, but it would also be true to say that most developers don’t like to write unit tests. It’s about 30% of development time that’s easily automated using AI tools as well. And so on. So it’s not only chats, it’s automation of different processes like summarization, classification, content generation, assistant tools which help to make decisions.

Arham: And are there any sort of tools you would like to call out which you personally use, or people on your team use, which you think are extremely useful for engineers?

Yuri: I use those tools on a different level as a developer. I’m using some tools which wouldn’t say anything to non-developers, like the Hen Face platform for models, or Coro, or Code Pilot. Most often I’m using Gemini code assist. I like to use AI Studio from Google because there is grounding from Google. You have the possibility to get answers which are up to date. There are thinking models which not only provide you an answer, but they provide you a well-defined answer. There are models which are dedicated for coding improvements. That’s as well as tools for usage. As a client, I prefer to develop tools myself.

Arham: I wanted to get into some of the pitfalls, potential pitfalls, you see of using, or would you say overusing, AI. Are there any sort of disadvantages you see of relying too much on AI within organizations?

Yuri: A year ago, I saw there were higher expectations for AI from customers. For example, if it is summarization, customers expect 100% precision. None of the publicly available LLM models give 100% precision. That’s not possible. Also, how models work with information, what they consider important, what is not, it’s not precise. So that’s why it’s important to note that AI tools are assistant tools. They help in our work. Still, this year is about to be promising. New models are appearing. Now I work on experiments with text-to-video, images-to-video generation, or to synthesize different texts, conversational dialects. If you saw Notebook LM from Google, it’s a great example of how conversational AI works. Last week I built something similar and we are using it for internal usage.

I wanted to add that soon companies will stop mentioning that they use AI to build their product. It will be by default. Most of us will use AI in their daily life and their work. There will be no reason to mention that it’s built using AI, as we do not mention how bread is made or that a car has wheels or tires. There will be no reason to say that it’s built on top of AI.

Arham: Well, I think Notebook LM has also already shown us the capability of making podcasts through it. But I guess there’s something about the human touch which still keeps me employed, I guess.

Yuri: There is a local community of IT specialists in Austria. I generate a quick digest for them, and last week we noticed AI constantly generating horses with six legs. If you use AI too often, people will forget how real horses look, especially people who live in big cities, and they will believe that horses should have six legs. That’s the danger of it. You can use AI to manipulate people’s opinions. That’s why it was so stressful for American companies to see competitors like Deep Seek last week. It’s the way to play with information which you would like to show users.

Arham: Yeah, and I don’t really know if there’s a solution to the question I’m about to ask, but it’s just a thought that came up. Going to more of the ethical and misinformation sides of AI, where you said that people will at some point just start believing that horses have six legs. And that’s where I’m wondering, we talked about a post-truth era a few years ago when data was becoming such a massive thing and the use of big data. What are your thoughts on how AI is going to change that? You know, there are all those organizations like Google, they have implemented a lot of safety mechanisms to make sure that the transparency is there when people know something is developed by AI, but not everyone’s going to do that. So how do you see the future of security, ethics related to AI? I’d like to hear what you have to say about that.

Yuri: I think there will be a switch from classical software development. The way how people write software will be changing. Now we work with high-level software development languages. I started my education, not career education, with assembler. Nobody is using assembler these days. But we then switched to C++, then we switched to C, Java, and so on. We use high-level languages now, but it will be changing very soon.

And I think there will be big companies… Of course, I don’t know what the companies will do, I can only guess. I think they will decrease the amount of software engineers who do classical software development, and they will push in the direction of cybersecurity, cloud ops. Still, there is high demand now in most countries for data specialists, data science specialists, people who work with data, optimizing and fine-tuning the models, and so on. So there will be a shift, there will be changes. But I think we should always learn something new. It’s great that these things are happening these days. It’s not a tragedy for developers, it’s an opportunity.

Arham: We have a couple of questions. These are more audience questions, so I’d like for you to share a bit of your personal insights here as well. The first question is, what’s the most important lesson you’ve learned as an entrepreneur or leader in the AI space? Were there any mentors or influential figures who played a significant role in your development?

Yuri: Lessons I learned as an entrepreneur, as a leader in the AI space… You should constantly learn and adapt to new technologies. Last year I submitted my documents to university again to start learning data science.

Arham: Wow!

Yuri: There is always a lack of skills, lack of knowledge. We are a little bit behind. Technologies are evolving and growing extremely fast, and it’s critical to learn constantly on a daily basis. That’s why when you see Bill Gates recommending reading some books, don’t miss that opportunity. Pay attention to this.

Arham: Do you have any books to recommend?

Yuri: Most recently, I’ve been reading documentation for API usage of different platforms, like how to work with Vertex, how to work with the Hiking Phase. I would recommend API documentation for Meta and for Microsoft. That’s a great resource.

Arham: Awesome. And then the second question was, were there any mentors or influential figures who played a significant role in your development?

Yuri: Influencers for me are entrepreneurs or people who build something, some real products or services, and found their niche in the market. That sometimes looks easy, simple, but there is always a lot of work behind all of these stories. And that’s really great when people know how to build their technical skills and to use them to produce something which is useful for people. The same can be said about OpenAI or similar companies for producing something which is changing people’s lives. So I think that’s the best.

Arham: Any name you’d like to mention here?

Yuri: No. No advertising. (Laughing)

Yuri: I like the phrase of Snoop Dogg, “I would like to thank myself, and only because of me I became who I am.”

Arham: So, looking ahead, what are some of the most transformative possibilities you’re looking for AI in 2025? Which industries do you think it’s going to really shake up?

Yuri: In recent years AI was mostly used in the tech sector, but now I think AI will be used more in other sectors: in logistics, in medicine, finance, banking, mil tech and so on. So I think AI will be widely used in other sectors and we will see it in our life. I see the changes will be significant and even difficult to predict where we will be in one year. I see huge competition between big companies in this area and new products, new services appearing on a weekly basis. So let’s see where we will be in one year.

Arham: Yuri I really appreciate your time. Thank you so much for joining us today and thank you everyone for listening until next time, bye-bye.

Yuri: Thank you, bye.