EP601: How AI is reshaping business operations | Denis Romanovskiy, Chief AI Officer at SOFTSWISS

**Cyrielle Delmas** (0:06)
Artificial intelligence is transforming the way businesses operate. But for many organizations, the biggest challenge isn't the technology itself. It's adapting people, processes, and culture to make the most of it. Joining us today is Denis Romanovskiy, Chief AI Officer at Softswiss. We look at the challenges and opportunities of AI adoption, where businesses are seeing the greatest returns, and how technology is reshaping the modern workforce. Denis, welcome.

**Denis Romanovskiy** (0:35)
Thank you so much. It's a pleasure to be here.

**Cyrielle Delmas** (0:37)
Amazing. Let's jump right in.
You've described the biggest challenge with AI to be organizational adaptation rather than the technology itself. So where do you see the biggest barriers for companies who want to introduce AI adoption in their daily basis operations?

**Denis Romanovskiy** (0:58)
You know, it's not like you have a single barrier. It's like you have a single barrier, then you have another one, then another one. Yeah.
If you get back like six months, the main barrier was that people were afraid of AI, and they didn't trust it, and it hallucinated a lot, and we didn't have any methodology how to overcome it. The models were not that good.
But since then, we did a lot of trainings, and there are more trainings like online, and people are doing better. But these days, actually people started to use it a lot, and they sort of automated some things here, there. They made better artifacts like these presentations, like PDFs, whatever kind of documents.
They automated their flows with content, with even some, you know, business processes. But I see that we are stuck now in teams. So we all use it AI well for individual work.
But when you need AI in a team, it's absolutely a different story. How do you pass these, you know, AI instructions and artifacts between different roles in the team? And if you have a cross-functional team, you know, there are expectations between the roles, how they communicate to the together. And now, because we don't have this kind of, you know, methodology or whatever, yeah, this is where very often AI stacks and AI adoption stacks. And it's not that easy just to improve your processes. Yeah, you have like 20, 30% improvement of work. But then, you know, you have to pass this work to another role. Like, for example, in software development, yeah, your product manager, you create what you want to do in the product. Then you pass it to developer, then developer pass it to tester, yeah. And because we don't have like a common approach how to do it with the AI, yeah, this work is stuck. And we cannot like speed it up like three times, five times, ten times faster, yeah, because then we need to get back, let the other role to review the work, yeah, and understand the results. And actually, there are already like approaches how to do it, but we again need to train people, yeah. People should choose the best tools for that and implement. So, you know, these days we need to train people to use AI tools, we need to train them to use tools like Git, yeah, GitHub, GitLab, yeah, and they just have to pass instructions, they should do these instructions of AI agents and run these AI agents together, monitor them together and understand how can they improve it. So, I think this is the kind of level of implementation of AI now, that is very important to gain the most of the wealth.

**Cyrielle Delmas** (4:10)
And talking about methodology and approach, how do you measure a successful AI initiative? And also, why do you think companies should keep creating these AI leadership roles just like yours?

**Denis Romanovskiy** (4:22)
Okay, about initiatives.
A good initiative these days with AI is a small initiative, yeah. If you try to build something big, you will definitely have a lot of problems. And it will just, you know, with equality, with security, with governance, with just passing this work from one role to another, like I explained before. So keep your initiative small and figure out how you will evaluate, like, what is the economic effect of it? Yeah, like, for example, this initiative saves some hours of a particular person through their day versus how much tokens they spent.

**Cyrielle Delmas** (5:02)
What's an example of a small business initiative in AI?

**Denis Romanovskiy** (5:06)
For example, one of the initiatives we had in marketing recently when we had to collect data on our competitors.
Previously, it was like research, research, research. But now, we have an agent that goes through multiple websites, multiple systems, collects all data together and combines it in a specific way and find trends and find risks and so on. And create a final report that saves probably maybe 15 to 20 hours per week.