Although 71 per cent of business owners declare that artificial intelligence is their investment priority and most plan to allocate more than 10 per cent of their budget to it, the corporate reality draws a very different picture. The market is entering a phase of acute dichotomy: on the one hand, boards are pushing for innovation; on the other, they are losing control over how the technology is actually used. The ‘shadow AI‘ phenomenon is becoming a silent killer of data security. According to WEBCON analysis, as many as 70 per cent of companies do not monitor the AI tools used by teams, while 8 out of 10 employees use them on their own. As a result, company data ends up in external models without any oversight, and only 29 per cent of organisations are able to keep this process within approved procedures.
The year 2026 will bring a brutal verification of past declarations. The rate of adoption of the technology is unprecedented – ChatGPT achieved in two months the reach that the Internet had worked four years to achieve – but the business success rate remains alarmingly low. An MIT report indicates that only 5 per cent of Generative AI-based prototypes end up in production deployment. The problem is not a lack of technology, but an inability to embed it in business processes. Łukasz Wróbel, CBDO at WEBCON, points out that the key to success ceases to be mere access to models, and becomes the construction of internal ecosystems where data is structured and verified. Only then is it possible to safely scale bottom-up innovations, which are often created in small teams and can be implemented in an iterative model.
A paradigm shift towards the ‘agentic enterprise’ is already on the horizon. Competitive pressures are driving 69 per cent of managers to reach for autonomous AI agents to not only generate content, but to realistically perform tasks in back-office systems. However, this enthusiasm will collide with tough financial requirements. Investors’ patience is running out – according to KPMG, 67 per cent of managers now expect a return on investment in AI within a horizon of 1-3 years, a drastically shortened outlook compared to previous years. Meanwhile, 91 per cent of companies still lack hard data on the impact of AI on their efficiency. Those that fail to move from intuition to measurable business metrics risk dropping out of a race in which the stakes – according to the PFR report – are an increase in GDP of €55 billion per year.
