“Right now we don't have the conditions for AI to develop dramatically. We have neither more data to train it on—everything was annotated long ago—nor any new architectures. They're trying various amalgams of existing solutions, but the progress isn't convincing. For me, there are no preconditions for anything dramatic in the development of AI.”
“The future and development of the IT sector have many more aspects, and some of them, in my view, are even more important than artificial intelligence.”
With these unexpected claims, Krasimir Kostadinov, chief technology officer of Scalefocus and a seasoned IT specialist with wide-ranging interests, impressed me during his previous visit. His words in his talk at the 9th DEV.BG All in One conference were no less provocative. That is why I invited him to continue our conversation about the Bulgarian dimensions of the IT future. About the prospects facing the sector and how it will get along with everyone else. About whether it is worth entering the IT field in Bulgaria today, and which qualities turn out to matter in the dynamic times we live in.
– You're the chief technology officer of a large company, yet here we are talking about history, sociology, even politics if you like. Your view of the future spans very broad areas—you almost step into the role of a futurologist. Why do you do it? Do you consider it part of your job?
– Broadly speaking, there are two ways to grow a business—proactive and reactive. If you want any long-term control over what happens to the company you're in, you have to take the proactive stance. But that's the hard one. Reacting is easy—you don't even need to think, you can just watch what others do and copy their models.
Sometimes that's very bad, because it easily turns into a “cargo cult”—you imitate someone without knowing why they do what they do. At Amazon, Jeff Bezos reportedly put an empty chair in important meetings where decisions were made, because it symbolized the customer. Lots of companies adopted it. But you can't become Amazon by mechanically copying something. It's funny, too, but the point is that you're going about it backwards. You do something not because you need it, but only because someone else does it.
To be proactive, you have to try to predict where things are heading. There's a wonderful story by Nikolay Haytov, one of Bulgaria's best-loved prose writers—“When the World Was Taking Off Its Breeches,” about the old baggy trousers of the Bulgarian village giving way to modern times. Yes,
no one in today's world can predict what will happen in 5 years, but you have to be adequate at least 1-2 years ahead.
– How can the Bulgarian sector take advantage of this situation and the perfect storm the IT world is in, which we talked about last time? How can a sector oriented toward service companies—working mostly for foreign clients—take advantage of it?
– First—by orienting itself strongly toward Europe, which is a safer bet in the medium term. Even if we become friends with the US again, which I at least don't see happening. There's a tendency there that started with fatigue from the endless wars overseas and degenerated into something totally different.
How do we do it? First—we concentrate on Europe, which for the service sector means learning languages. To be close to clients in Scandinavia, for example, you need to know Swedish or Danish; for Germany, Austria and Switzerland—German.
There's another huge opportunity—
the Middle East. Digitalization there is moving at a very fast pace right now,
and things that happened here years ago are arriving there late, so we can be very useful from the position of experience. We can consult, not just provide the service. The risk is that it's an in-between camp that plays by its own rules, has a different way of doing business, and we don't know how the relationship would hold up in the long run if a conflict with Europe arose.
But—focus on Europe, on the problems it has, and they're very different from the problems of the US, in every single sector. Europe, for example, leads in financial technology, and because of the rules in that area there are much bigger opportunities for startups.
The serious defense spending that's coming will be directly tied to the IT sector,
there's no way it won't be. Yes, this is very heavy software in every direction, but defense everywhere is a very undigitalized sector, among the last on that list. Back in the day technology came from there, as military developments, but in recent decades nobody in Europe invested in this sector, and that changed things. At the mall I can buy better cameras than those on some of the most modern tanks.
– It's a very sad observation, that there will be big prospects in the defense sector, but it does sound indisputable…
– It is. Serious money will pour into innovation in this field, and it's clear that every new solution will be digitalized—there's no other option. For the US, reaching 3 percent of GDP for defense is a change, but for most European countries
it will mean a manifold increase in funding, and that can't help but reach the IT sector.
Most IT companies are fairly liberal institutions, and there will be a clash of cultures, approaches, attitudes. We're used to discussing things, but a general isn't used to working that way. In the coming years IT companies will have to learn to work with such clients, and those institutions will learn to work with us—it's a process of mutual adjustment. But it will be interesting.
Of course, one difficulty is that Europe is a regulated market, which creates inconveniences and slows things down. But I don't think the people who think strategically should be underestimated. At some point they'll realize, for example, that there need to be fully European cloud technologies. Since there's no serious investment capital here, and our bankruptcy laws are quite binding, the EU is the only option to fund them somehow.
Europe has a strong starting position and shouldn't be underestimated on the digital field. Linux started in Europe, and to this day a significant part of the core people developing it are here.
– And yet, so far I see no indication that Europe is making a deliberate effort to become a digital player.
– We really shouldn't underestimate it. Many people talk about Taiwan—what happens in a war with China, since the world relies on chips from there. But even there Europe has capacity—yes, over 80 percent of the world's chips are made in Taiwan, but over 95 percent of the machines that make those chips are produced in the Netherlands. If we decide to invest in chip manufacturing, it's not as if we don't already make the machines for the whole world.
We have the capacity to be digitally independent, and we have the money; we just have to decide. So to me it's very obvious where things are heading.
You can see the trend globally—investing in factories for artillery shells, but also for chips. And it started even before this technological nationalism, back during COVID.
All the managers with business administration degrees who make the decisions have been drilled in the ideology of constant growth. Every year more than the last, no matter that, for example, the population is shrinking—which is happening today on every continent except Africa.
We can no longer go on with this way of thinking!
– In my opinion, one of the problems specifically here is that our sector is very cut off from all the others, whichever way you look at it.
– Our sector is based on services, that's clear. Other companies matter too, of course—there are skills we simply couldn't have developed as a sector without them. But a large share of people work in such companies.
The big change is that clients will demand more accountability from us. Models based solely on price won't work. The market leaders, the more innovative companies, are only now introducing new models based on outcomes and shared risk in service delivery. Alas, so far they're applicable only very narrowly. We need a new model—oriented toward the final effect—you pay for the job to get done, yet I stay flexible in delivery. Yes, I don't guarantee that what I build will do the job for you. But I guarantee you're paying me for a specific result.
As a sector we need to start offering more. To share the risk together with the client. To move toward more specific things. Models where you deliver everything, from start to finish,
without caring about the specifics of the work, which again means taking on more responsibility. And risk, but as an industry we have to learn to manage risk.
– If something has changed drastically here in recent years, it's that entry-level specialists are in ever lower demand. Is that a problem?
– Yes, because it means a demographic crisis is looming in the IT industry. For 2-3 years now almost nobody has hired juniors—things flipped from paying people while you train them to people chasing you for an internship and having to pass serious tests.
Another 2 years of not taking young people into this sector, and there will be a demographic problem. It'll be 4 years without young people, others leave the sector, and in 4-5 years there won't be any mid-level specialists.
What will the effect be? A shortage of staff.
AI can't solve the problem. AI improves productivity by up to 20 percent, for some things up to 50 percent. There's no such thing as giving AI to one programmer and him becoming 15 programmers.
– Is that really so? Personally, I can see how in some cases it's entirely possible for me to be three or more times more effective as a programmer with AI.
– It's possible with some types of tasks, like writing tests, or if you start a project from scratch. AI can generate a base project with the latest technologies for you, and in some respects the speedup really is manifold. But in others—fixing bugs, for example—it doesn't help much.
It depends where you work—if you have old software, there's no way it'll take over a large part of the work.
The impact of AI is patchy across different cases.
What strikes me is that there are still no methodologies designed with AI at the start. They're all for people working with people.
And we shove AI into everything we can—code reviews, projects, documentation.
Everything we do will stay the same, we'll just stick AI into it? That's not the right approach. As an industry we need “AI-first” models, and those will open up different possibilities.
The example is how there would have been no industrial revolution if we had “hooked” a steam engine onto every worker, instead of having it do a significant part of the heavy work while the workers tended it.
We keep talking about Copilot—a co-pilot. I'm an aviation buff, and there's a concept called Pilot Flying—the two rotate, sometimes it's the captain, sometimes the co-pilot (there's no such term in aviation, actually; that role is generally called “first officer”). Who flies the plane is chosen on the spot according to a number of criteria. There are specific rules, and they vary from airline to airline.
AI gives you the chance to rotate. It doesn't just help you; in some situations you can give it the main role—you just check and monitor it—and then you switch: you write, it checks.
You get a different dynamic, one that current approaches don't reflect. They're built around interactions between people and their limitations.
Right now we've grabbed the adze by the blade and we're digging.
– What other factors do you see coming “from outside”?
– Twice in my career I've seen services shift to India. After the 2008 crisis there was one, because companies became very cost-sensitive, and IT services aren't a core expense—they're the so-called discretionary spend for many industries.
Economically, we're the cherry on the cake, and before you give up the whole cake, you give up the cherry.
I saw that shift to India reverse back to Europe in 2011, for many reasons—time zones, culture, language. Serious IT work can happen in India, but it's not easy, and American companies can't learn to do it well. They go for cheap, but at some point they realize that if there's no result, cheap is pointless.
There was a saying: “I'm not young enough to know everything.” I see a new generation of managers coming up in the US who know everything, just like the managers before them—the Dunning-Kruger effect. “We go to India, cut costs by 50 percent, collect our bonuses, everyone's happy…” The problem is that one, two, up to 4 years pass, judging by experience so far, and it turns out the program has to come back. The same effect will happen now. I'm even already noticing a trend of some American companies coming back from India.
On the other hand, when the war in Ukraine ends, there will be pressure on the Bulgarian sector. Ukraine has quite a few specialists; however things develop, many people will return, and the sector will be made a priority. We don't differ much in time zones or skills, and we have to prepare for that. There will be a reshuffling of the layers, which is easy to foresee. The point is that we need to have things well baked. We've long been expensive. Well, if you can't beat them, and there's no way to, the approach in my view is to join forces. Bulgarian companies should open offices there.
They've accumulated know-how in flying drones, bridges, how air defense is done, while we, as an industry, have meanwhile built up experience in delivering IT projects. This higher-level experience of ours, combined with the cheap engineering potential that could come from Ukraine, may actually be an opportunity.
– So, to sum up—what should we expect in the Bulgarian IT sector in the coming years? And which decisions is it important to make in time?
– I don't take a gloomy view of things; I'm just a realist. There won't be a big collapse. Yes, many people may change jobs, but
thinking people always find work.
Yes, when a new technology changes things radically, jobs are lost. In that same story by Haytov, the man starts out as a maker of old-style shoes and ends up grilling kebapche, Bulgaria's beloved grilled minced-meat sausages. In his case, beer arrives in Bulgaria and he opens a beer tavern.
New opportunities are born. We've seen how many job titles appeared in the last 10 years, not only in IT. In the next 10, I think it will happen even faster—new professions whose titles would make us laugh today will be perfectly valid jobs in 10 years. That's a fact.
We should think less about what we're losing and more about what new things are being created.
We have to be flexible and adapt quickly, because the changes are beyond dispute.
– Can we learn that?
– A thinking person, whatever you hire them for, is immediately valued.
AI opens up some possibilities that didn't exist before. They cut both ways, for good and for bad. The chance, say, to start a small business. With AI there could be quite a few startups with business models that depend mainly on AI plus human oversight. So it may even open up jobs for the time being.
In many organizations there's internal resistance to AI coming in. That's the good thing about the service sector—you work with every kind of company, you see all sorts of contexts, and in a short time, without changing jobs. The resistance is enormous, especially in more conservative industries where there are engineers with long experience; they're skeptical, they've been through various technologies.
Everyone is talking about AI now, but the real, large-scale change in our lives will most likely come in areas like pharmaceuticals (AI for fast drug development), fake news, AI scams, cyberattacks. Anything can happen, but
cybersecurity has a bright future, and in Bulgaria we have serious experience in that area
It's impossible on the basis of AI alone—it's a very complex subject, with many factors intertwined.
Growth in demand for software will continue. Here, too, we'll hit a plateau in the demand for people, and we have to think about how to carry on. If I go back to the farmers example, AI may be our combine harvester.
– And yet, there are quite a few differences beyond the comparison with that sector…
– That's true. One of the problems farmers don't have, because theirs is mechanical work, is that the capacity of the human brain hasn't increased in the last 10,000 years, or even the last 200,000. Our cognitive capacity hasn't grown. And that's a problem in our work when it comes to speeding things up or making them more complex.
There's another interesting experiment: in the late 1960s a team was asked to estimate how long an IT project would take, as described in The Mythical Man-Month. And it turns out that the answers today don't differ drastically. The same number of people needs a similar number of days to handle the same complexity. Yes, complexity is higher now, a banking system is more complicated, there are different expectations in many directions—user interface or performance, for instance—but roughly speaking the complexity of the system is at the same level and requires the same number of people and hours.
Because of all this, I think the IT sector will stay in a strong position. Will that be true in Bulgaria as well? It depends on us. On whether we make the right decisions about how to change so that we're adequate to the new challenges…
