Eduard Ashrafyan: Why 'If It Works, Don't Touch It' No Longer Works

In recent years, the clunky term 'digitalisation' has taken hold, though it doesn't sound great in Russian. 'Digital transformation' is a bit more elegant, but rarely clarifies much. So what lies behind this trend? Who sees digitalisation as a lifeline — and who considers it an unwanted burden? We spoke with Eduard Ashrafyan, a specialist in digital leadership and AI transformation, business architect at Reksoft, Director of AI Export at NeuroLab, and Associate Professor in Business Informatics at the Financial University's Faculty of IT and Big Data Analytics.

Leaders in AI transformation are becoming business leaders

— In the business world, it's trendy to talk about digital transformation. But what does that really mean? Is it just about new tech, or something deeper?

— Digital transformation means different things to different people. Some recall large-scale IT projects; others are learning the tools. But there's a group of pragmatists — owners and top executives — who have implemented these technologies and genuinely increased company profits. For them, it's not just a buzzword, but a practical instrument for business growth.

In 2025, the classic idea of digital transformation is evolving into AI transformation. Those who laid the groundwork through earlier digitisation are now rapidly integrating AI systems. This isn't just a tech initiative — it's a business-driven approach led by executives focused on profit.

Look at Sberbank and Herman Gref. A decade ago, he introduced agile principles into the mainstream. Now, Sberbank doesn't just play with tech — it leads the financial sector. In 2024, it earned ₽1.58 trillion in net profit — a record. That success stems from strategic tech investment, process reengineering, and knowing how to realise ROI on a sensible timeline.

— Many leaders still say: 'If it works, don't touch it.' Why should businesses change processes that are already profitable?

— That mindset comes from two scenarios: either the current leader wasn't involved in building the system and avoids altering it, or they built it themselves and are reluctant to start over, even knowing the benefits. So 'don't touch it' becomes company gospel. But the tech landscape changes fast — and with it, the market. Competitors who adapt quickly grab more value. Even a successful company eventually faces a choice: give up market share or evolve. Tools now exist to forecast these shifts — through management accounting and growth models — providing hard metrics that justify transformation.

Take T-Bank. Instead of building a branch network, they went digital from the start — mobile app, remote services. By February 2025, with modest starting assets, they reached 32 million active users and 49 million in their ecosystem — joining Sberbank and VTB in the top three. Or Yandex Go. It took on Russia's traditional taxi fleets with an aggregation platform, creating a more convenient service. The result? ₽227 billion consolidated in RideTech — taxis, car sharing, scooters. That's profit from a once-stagnant industry. So yes, even when business runs well, external pressure demands evolution. As the Red Queen said: 'It takes all the running you can do, to stay in the same place.' Transformation is a calculable task — not blind faith, but forecasted ROI. Regular reviews help predict risks and losses. That's why smart companies innovate before things break.

Digitalise This!

— In which areas of business and governance is digital transformation most successful? And where does it still struggle — and why?

— Each industry experiences digitalisation differently. The 'Digital Vortex' concept maps out which sectors were swept up early, and which are only now approaching transformation. According to recent IMD research, the most digitalised sectors are tech services, education, and financial services. Telecoms, media, and retail follow — still highly digital but with slower recent growth. At the bottom are industries like energy and heavy manufacturing — digital laggards. That's no surprise. Internet-based companies and mobile services are born digital — IT is in their DNA.

Same with mobile networks: they were designed in the computer age. But manufacturing and energy sectors rely on decades-old systems — even over a century in some cases. These legacy landscapes are fragmented and complex, requiring phased tech implementation. The shift to Russian-made software has added friction. Replacing familiar foreign tools with local alternatives has increased the number of 'integration seams' across platforms. That's why digital transformation in traditional sectors moves slowly — you're maintaining the old while building the new. Even so, the shift is underway. Industrial IoT, digital platforms for logistics and extraction — investments are rising. The effects may not be as immediate as in online services, but they're coming.

— Is there a risk that in some sectors, digitalisation becomes a trendy buzzword without tangible impact?

— That risk is real. Some companies pursue digital projects simply because it's 'what you're supposed to do' — without a clear goal. Money is spent, time is lost, and results fall flat. Managers quietly admit to failed attempts: a system is deployed, doesn't take off, the team burns out, and the project is scrapped. But it's important to see this for what it is — normal. Transformation rarely succeeds on the first try. Early projects are experiments. Pioneers stumble, gain experience, and find success in later iterations. Once a company wins, nobody remembers their early failures. As Mark Zuckerberg said: 'In our society, we often avoid big moves because we're afraid of making a mistake... but that shouldn't stop us from starting.' That's how experience is built — by failing smart and trying again. In Silicon Valley, failure is seen as a necessary step to success. So, a failed digital project isn't a disaster — it's a valuable lesson. The danger lies in making digitalisation a PR stunt or an end in itself. It should always tie back to business goals. When it does, real value follows.

AI Is Changing the World Before Our Eyes

— In which areas will AI and digital tech grow strongest in the coming years? And where will human expertise still be vital?

— It's already clear: projects with small teams can now deliver outsized results — and that shift highlights how essential human expertise becomes when paired with powerful digital tools and AI. By 2025, we're hearing job interview questions like, 'Show me your best prompt' — meaning the most effective and complex prompt someone has crafted to work with AI tools. Every professional will soon use LLMs or AI agents — from junior staff to top executives. We already see AI in support centres, decision-making systems, and analytics.

A new wave of AI agents and multi-agent systems is on the rise. Over 30% of the world's largest firms are already piloting or deploying such systems. The best outcomes come when humans focus on judgment and creativity, while AI handles routine, analysis, and generation of options. People make the final calls, apply intuition, and solve atypical situations. But the volume of tasks that AI can perform keeps growing monthly. One striking case: Petrobras in Brazil deployed a generative AI agent for tax calculations and, within three weeks, found ways to save $120 million. They're now scaling it to HR, procurement, and finance, expecting annual savings of up to $1 billion. BP and other global energy giants are following a similar course — adopting generative AI and traditional algorithms across the board. So even in hardware-heavy industries, AI is proving itself — in logistics optimisation, predictive analytics at extraction sites, and office task automation.

AI will accelerate everywhere: from agriculture, where robots and field image recognition are used, to government, where big data analytics supports national 'data economy' initiatives. But in every case, a person is still needed — now enhanced with digital tools. In tech circles, we talk about employees with digital exoskeletons: people augmented with AI capabilities. Human expertise isn't going anywhere — it's just becoming digitally supercharged.

— Digital tools make life easier — but do they make people complacent? Are we losing skills as machines handle our routines?

— Yes, some skills will fade — just as driving replaces walking. The World Health Organization recommends 10,000 steps a day to offset this. Similarly, business activity will shift toward more complex tasks, and leisure or wellness routines can make up for less physical or cognitive effort at work. There's already a term for this: 'automation complacency'. Pilots overly reliant on autopilot can lose their manual skills. In business, employees may over-trust software and stop understanding how it works. When systems fail, they may struggle to respond appropriately. But automation's purpose isn't to make people idle. It's to free them up for higher-value tasks. Studies show that employees using AI tools feel more effective.

A Workday survey found that 93% of those using AI felt it freed up time for strategic work. Instead of filling out reports, they interpret results and make key decisions. Of course, this demands new skills. That's why training is essential — so that people don't passively follow machines but evolve alongside them. When done right, automation becomes a launchpad for learning, not a threat. The company gains efficiency, and the individual finds renewed purpose.

Managing Business with AI: Analytics, Automation, and Beyond

— What does running a business with AI and digital systems look like today? Is it more about analytics, automation — or something else?

— It's largely about automation — but not only that. AI-powered systems give managers an unprecedented level of insight. They can collect and process vast datasets on business performance and customer behaviour. AI helps integrate this data, detect patterns, and support fact-based decisions. More and more decisions today are data-driven — not just based on expert judgment but grounded in real evidence. Banks and retailers analyse transactions to personalise offerings. Manufacturers use sensor data for predictive maintenance. Routine managerial tasks — scheduling, screening CVs — are now mostly automated.

A global survey showed that the top three use cases for AI in business today are: data analysis (51% of companies), security monitoring and fraud detection (43%), and HR and recruitment (39%). These fall into two major buckets: deep analytics, and automation of repetitive operations. A third area is growing: forecasting and decision support. BI systems and AI assistants now highlight trends, flag anomalies, and suggest actions. These aren't systems to blindly obey, but tools that help executives manage complexity. The portrait of a modern manager? They have real-time dashboards with key metrics, forecasting models that anticipate problems, and digital tools that streamline communication — from emails to chatbots. Many routine controls and reporting tasks are now handled automatically.

The human role is to interpret insights and make strategic calls. Compare a CEO from 2000 and one in 2025. The former relied on rigid hierarchies and offline updates — emails, Excel summaries, intuition. The latter operates in a cloud-based ecosystem with real-time data, low-code bots, AI copilots, and synchronised remote teams. Their job isn't to issue orders, but to create an environment — to be a curator of perspectives, not a dictator of decisions, as Harvard's Linda Hill puts it so well.

What's Next?

— What will business management look like in 5–10 years? Are we moving toward people-free companies — or a new model of human–AI collaboration?

— I believe the future isn't 'business without people,' but rather business powered by the model of 'human + AI.' This is what Gartner calls 'augmented intelligence' — a human-centred partnership, where AI is seamlessly embedded into tasks, like email is today, and significantly amplifies human decision-making. A helpful analogy: autonomous KamAZ trucks on the Moscow–St. Petersburg highway. AI systems can handle traffic lights or lane changes, but destination planning still rests with human experts. As of March 2025, KamAZ-54901 autonomous trucks had driven over 7.2 million km and transported 800,000 cubic meters of cargo — all accident-free.

Practically, we'll see autonomous AI agents managing planning and decision-making in narrow areas, while top managers retain authority over key choices. Some firms are already testing 'autonomous enterprises' powered by smart contracts and AI. But we won't see people-free companies any time soon. Instead, human roles will shift toward creativity, strategy, and ethics — the areas AI struggles with. Meanwhile, machines will handle operational and partially analytical tasks. Imagine logistics departments run by AI platforms that manage deliveries and reroute traffic automatically.

Human managers set KPIs, track exceptions, and fine-tune the system. Sales teams might use AI to draft personalised offers in real time, while people step in for the final touch. In one GO4AI showcase, a 500-person company trained a neural net to generate commercial proposals (CPs), which employees then edited. That's the direction we're headed — systems doing the heavy lifting, with small human teams captaining the ship. Crucially, human strengths will become more valuable — empathy, creativity, moral judgment. These are skills algorithms don't have. So future business won't be cold automation — it will be teams delivering more value through smart use of AI and distinctly human qualities.

— Which professions are most likely to disappear or radically change because of AI and digital technologies?

— I wouldn't say professions will vanish — but many will evolve. Expertise will still be in demand, though roles may shift. Jobs that are routine, repetitive, and cognitively simple will change most dramatically. Take call centres: chatbots now handle most queries, and humans only step in for edge cases. In transport, autonomous vehicles may transform the roles of taxi and freight drivers. Entry-level office roles — like junior accountants or analysts — may also shift, as software handles bookkeeping and reporting. Translators are another example. Neural networks can now do near-instant translation. The profession won't vanish — but some translators will move into more complex specialisations. Even high-skill jobs are evolving. Junior and mid-level developers are seeing their workflows change with tools like GitHub Copilot and ChatGPT. More code is auto-generated, and team leads are using algorithms more intensively.

Systems architects and analysts — people who define problems and design solutions — remain highly valued. Writers and journalists are also adapting. GPT models can generate basic news stories. So journalists are shifting toward analysis, investigations, and exclusives — things AI can't do. Designers are impacted too. Tools like DALL·E and Midjourney generate images from prompts. Simple illustrations and logos might no longer need human input. Legal assistants may also shift, as AI handles precedent searches and contract drafts. In healthcare, specialists like radiologists now work alongside AI that can read scans — with doctors focusing on edge cases and quality control. So, we're not seeing jobs disappear — but transform. Think of a 1970s pilot versus one today: the job remains, but many controls are now digital. AI is changing the tasks within professions, not eliminating them. People will need to reskill, expand their capabilities, and work with AI — not against it.