Having machines as coworkers: the next revolution coming with agentive AI

BarcelonaImagine that you are a businessman and you have to make a financial decision. Instead of starting to gather all your experts, calling the managers and finding out about the situation of the markets, the competition and the geopolitical risks, you put on virtual reality glasses and, in front of you, the agents you need appear in the form of avatars: one summarizes the current news related to your problem, the other describes the situation.

This scene, which today still seems to us straight out of a science fiction movie, is not so far from becoming a reality. In fact, the Mobile World Congress has witnessed demonstrations like this; this one, in particular, by Sergio Gago, director of AI and quantum at Moody's [in the photo]. It is one of the many applications of agentive artificial intelligence (agentic AI, in English), a step beyond the generative AI that we have known until now, and that allows software to evolve to become our agents or assistants, both at work and in our personal lives.

In fact, agents are already starting to become part of our daily lives, through help chatbots within service websites, for example, but they are not yet fully developed. However, according to the scientific director of digital technologies at Eurecat, Joan Mas Albaigès, "2025 will be the year of agentive AI," he says in statements to ARA.

"Generative AI has a set of algorithms designed to create new content, images, sounds, software code, etc. Apart from that, we continue to develop classic AI, which is what allows predictions to be made. That is what they have been until now," Mas Albaigès explains.

"Previously, agents were pieces of software that could operate under very restricted conditions, but with agentive AI we expand the field of action and make them much more autonomous," explains the expert. These agents can be used for many things, and not necessarily in the form of an avatar.

The industrial field is a clear example: predictive maintenance is already used, which is important in factories because it predicts when a machine may fail; but if agentic maintenance is added here, it goes one step further. "Additional software is added to the machinery that will not only predict failures, but will also tell us what will be needed to fix the machine, will schedule the repair operations, will be able to find the operators to carry out that operation, etc.," explains Mas Albaigès.

"Basically, what it will do is automate processes with an autonomy that they did not have until now," adds the expert. In addition, these software agents "will learn from experience," which is an additional step to the automation we have known until now.

Humans and agents, co-workers

The application fields of the agent are practically all, including the medical sector, where it will be possible to have agents that organize the entire process of patient admission, scheduling visits, operations, etc. "All this is already being developed; for example, Microsoft is making agents to control 5G networks," says the Eurecat expert.

The redefinition of work with agents was a main topic in the talk The web is dead: AI agent and the risk of post-web era, with a debate between experts from Microsoft, Databricks and Citi Global held at the MWC. According to Rick Lievano (Microsoft), "the web is flexible, it is not dead": "We are looking for a world in which perhaps, instead of there being an app for everything, there is an agent for everything," he reflects.

"We want to have a vision where every employee has a personalized assistant for their needs and increases their productivity," says Lievano. "Every business process will have an agent, which will improve any business process." In short, "humans and agents will be partners," she says. She also says: "In the company we have about 100 agents and what we have seen is that they are very good at eliminating repetitive work; this also gives people the possibility to think, to be more efficient, to use their time better."

Meanwhile, Sophia Bantanidis (Citi Global) says that "interactions with the digital world will change completely, they will be more dynamic, based on conversations, with interfaces." She also says that the key is in "the balance in decision-making": "The EC says that customers should not be subject to decisions made solely by a digital process, it should always go through human hands before reaching the customer," she explains. According to Bantanidis, "what companies should do now is invest in human/AI interaction."

"One of the misconceptions we are giving is that AI is totally autonomous, and that is not true. The narrative we give to employees must go in this direction, towards interaction and not replacement, because if workers receive it well, the result will be much better. The key is to all go together," he says.