Ai in enterprises has entered its mature stage but Italy runs at two speeds
Three years into the artificial intelligence economy, the adoption of AI in businesses is accelerating not only in scope, but also in depth. It is transforming the way people work, how teams collaborate and how organisations create and deliver products. This was stated by OpenAI, the company behind this new industrial revolution in November 2022, in a report published in early December entitled 'The State of Enterprise 2025'.
According to data from OpenAI's chatbot, 2025 marked the end of innocence for artificial intelligence. It is no longer a toy for enthusiasts, but the invisible infrastructure on which growing companies rest. But how much? Over one million companies are actively using AI tools. The volume of messages on ChatGPT Enterprise grew eightfold year-on-year. The most important figure is the increase in the consumption of 'reasoning tokens' (the ability of AI to think before responding), which grew 320 times. This indicates that companies no longer use AI just to write quick emails, but to solve complex problems. In a nutshell, this means, according to OpenAI, that employees save between 40 and 60 minutes per day.
Even rival Anthropic has calculated the fundamental difference between before and after. The AI would reduce the 'task completion' time - the expression is awful but we have to get used to it - by about 80%. The percentage is gigantic.
Looking up and extrapolating these figures to the entire US economy, it is estimated that the current generation of AI models could increase annual labour productivity growth by 1.8 per cent over the next decade. This rate is roughly double the growth experienced in recent years. We are at the levels of the economic boom of the late 1990s or the 1960s and 1970s.
There is, of course, a catch. As with electrification or information technology, the real transformative gains will come not just from doing the same things faster, but from reorganising production processes entirely around new AI capabilities. And this is probably the reason why many AI agent and AI projects have not translated for all of them, in the past three years, into increased financial results.
