Romanian businesses are hurrying to put AI to work, while their customers (and their budgets) take a more leisurely approach.
Romanians still pay for most things in cash. According to Elena Gheorghe of PayU, a payment processor, notes and coins account for around 55 to 60 per cent of payments in the country, only slightly down from roughly 65 per cent in previous years. Meanwhile the industry is already exploring “agentic commerce”, in which AI helps consumers search for products and takes part in some of the purchasing and payment process. Technological availability does not automatically translate into consumer readiness, particularly when trust, security and old habits still shape how people buy.
That gap between what technology can do and what businesses (and their customers) actually do with it ran through the discussions at InnVolve 2026, an event organised by Innobyte, a Romanian e-commerce company. Executives, investors and entrepreneurs discussed how AI is moving from experiment into everyday operations while consumer habits, financing and internal processes evolve at a different pace.
Then there is the question of who pays. Robert Berza, executive director of The Edge Institute, a think tank, and a former general manager of Fashion Days, said that more than half of the companies covered by the institute’s research expect some form of government support for AI and financing. His advice was not to build a strategy around that hope. Firms may well back the creation of public funding mechanisms, but with public budgets under pressure they also need a plan B. The practical question is who finances adoption, and under what conditions it delivers a measurable impact on profit and loss.
For some Romanian firms the effects are already visible. Marius Petrache, founder of Marcoser, an agricultural consultancy, explained that the company’s location makes it hard to recruit certain kinds of specialist. AI eased the pressure by increasing what the existing team could do, with employees taking on extra activities and projects. “We simply did not need to add more people. There was never any question of anyone leaving because of AI,” he said. At Marcoser, AI became a productivity tool and an answer to a real labour shortage, rather than a means of trimming headcount.
Sergiu Neguț, a co-founder of FintechOS, made a similar point. He described strategy as a “logic of success” and argued that some of the best opportunities lie in applying technology to markets and business models that companies already understand (“technology applied to something”, as he put it). With information noise growing and technology cycles getting shorter, the challenge for managers is less about access to more tools than about spotting the signals that justify a business decision.
Customers are getting harder to follow, too. Ioanna Christopoulou of L’Oréal cited company research showing that a beauty shopper is exposed to an average of 6.1 touchpoints in a single day before making a decision about a product. That number is expected to rise as social media, marketplaces and AI-based discovery tools become more important, which makes measurement as much of a headache as marketing. “GMV [gross merchandise value] is an important indicator, but it can be very dangerous if we look at it in isolation,” she said. Paid traffic, discounts and promotions may deliver immediate sales, but channels also need to be judged on profitability, retention, customer lifetime value and incrementality. For social commerce she offered three priorities: speed, relevance and trust.
In search of operational value
Carlos Davidovich, who writes about leadership and management, came at the problem from the perspective of neuroscience. The brain, he said, needs about 50 milliseconds to decide whether to stay on a website, with visual cues shaping that response before any text is read. But conversion is not simply a matter of design (worries about delivery, returns, refunds and the trustworthiness of the merchant can all kill a sale). The same goes for AI. “Give me something simple, something I can use,” said Davidovich, arguing that one barrier to adoption is often not rejection of AI itself but the fear of not understanding it.
Radu Georgescu, an investor and entrepreneur, drew a line between AI’s long-term relevance and the valuations of companies in the sector. Capital, in his view, is flowing towards a shrinking group of firms that beat investors’ benchmarks, while those below the bar find money harder to come by. A correction in valuations should not be mistaken for AI losing its relevance. Operational value and the hype cycle are different things.
Nor does country risk affect all capital equally. Marius Ghenea, managing partner at Catalyst Romania, said that for pre-seed, seed, Series A and parts of the growth segment, a country’s rating makes little fundamental difference, because many of the companies being financed are built for regional or international markets. For larger investments, however, investment-grade status can become “decisive”. The same macroeconomic backdrop can therefore treat firms very differently depending on their stage, size and growth model.
Taken together, the discussions suggest that AI, consumer behaviour, financing, talent, payments and investment are becoming hard to treat as separate issues. Decisions that once sat with technology or e-commerce teams are increasingly becoming management decisions, involving cost, risk, growth and organisational design. “When we talk about e-commerce today, we quickly get to questions about costs, AI, infrastructure, people, risk and customer experience,” said Cătălin Șomfălean, the boss of Innobyte. “This is what interests us at InnVolve: bringing into the same conversation perspectives that, inside companies, already end up at the same decision-making table.”
Photo: Dreamstime.