AI agents, the new infrastructure revolutionising finance, ecommerce and the entire digital economy

With agents we go from having everything at our fingertips to having a tool that can solve our problems in a comprehensive manner without leaving home

by Pierangelo Soldavini

10 December 2025

5' min read

Translated by AI

Versione italiana

5' min read

Translated by AI

Versione italiana

Americans are real acronym maniacs. Any phenomenon is immediately transformed into a friendlier acronym. So now we also have to familiarise ourselves with Dofm, the 'Do it for me' economy, as a Citi report christened it, focusing on the new frontier of AI that promises to transform the way we live, consume, produce and work.

Just over two years ago, we discovered generative artificial intelligence, learning to interact with tools such as ChatGPT, Gemini, Perplexity and the like. Today we are already at a new, more advanced generation of artificial intelligence, that of agents, destined to surpass Gen AI in efficiency and speed. AI agents are intelligent systems capable of understanding complex inputs, making in-depth reasoning, making decisions and pursuing goals on behalf of the user: in fact, they are tools capable of replicating the way humans act, specialised in certain areas and for specific tasks. If generative AI is the fastest adopting technology in history, agentic AI represents the most intelligent evolution, acting autonomously for specific tasks and changing our daily lives even before professional ones.

The evolution of technology extends its effects beyond mere efficiency and cost reduction to more profoundly affect the way we act, with the prospect of also impacting psychologically and substantially transforming human behaviour. American pragmatists brand it as the 'do it on my own' economy, we Europeans, more philosophical, prefer to refer to it as the 'economy of laziness'.

Economy of Laziness

Already the internet with the addition of mobile access has led humans to think that they have everything at their fingertips: with the telephone we used to phone, today we watch videos, shop, converse, play games, listen to music, get information, pay, order dinner. Telephoning is the last thing we do. In fact, with the smartphone we can do everything without moving a finger, or rather, just moving our index finger. And we take it for granted that this is the case: when we no longer have the possibility to access these services, we are definitely uncomfortable. In the meantime, however, we have gradually unlearnt how to make food, how to leave the house to go shopping, how to find our way around or how to buy a book, but we have also changed the way we relate to people, delve into topics, pay attention, read. The culture of immediacy, of everything immediately, has inevitably transformed the way we behave and think for everyone.

AI agents make a quantum leap in this respect. From everything at our fingertips, we move on to having a tool at our disposal that can solve our problems in a comprehensive manner: no longer simply buying something without leaving the house, but being able to count on an overall guide in the choice, showing options and possibilities based on needs, tastes, budget, conditions. All we have to do is choose, then he takes care of ordering and even paying. It is as if we had a personal assistant that today is for shopping - Amazon has launched its 'Buy for me' -, but tomorrow it could be for finance, savings, information, reading, work, business strategies, and so on, potentially for any human choice, without limits: we submit our needs, the AI agent finds the most appropriate solutions.

"Decide for myself"

Artificial intelligence agents are, in fact, shedding their experimental shoes and entering a more mature phase of their evolution. They are no longer simple chatbots: they are systems capable of reasoning, deciding and performing complex tasks autonomously, relying on cloud platforms that offer scale, resilience and governance.

We are entering a new 'agent era' enabled by the cloud, as Capgemini argues, with AI moving from automation to co-creation of operational models and new customer experiences. "AI agents are not just another technological wave: they are a paradigm shift, comparable to the rise of the internet," observes Emmanuel Néré (Generali France) in Capgemini's World Cloud Report, which forecasts an economic value for these tools of up to $450 billion by 2028, highlighting the transformative potential for the financial sector in particular.

The adoption curve appears sharp: 87% of financial institutions, be they banks or insurance companies, have invested in AI, 32% are generative and only 10% have gone so far as to deploy AI agents at scale. Adoption, however, is set to grow rapidly: 80 per cent are still at the ideation or pilot testing stage, but there remains plenty of room for development, especially given the current low implementation rate.

Companies are optimistic that agent-based AI can effectively address critical issues, offering key benefits such as real-time decisions, greater accuracy and faster response times. "AI agents ensure autonomous goal-oriented decisions with minimal human oversight," the report explains, highlighting the impacts on productivity, accuracy, response time and scalability.

In addition to efficiency benefits, financial services managers recognise the concrete potential of AI agents to generate tangible business results: expansion into new markets without huge infrastructure investments, dynamic pricing and offers, multilingual support in line with regulations and local specificities, customisation, and generally improved profitability and competitiveness.

These are all arguments that also apply to many other sectors, such as, for instance, ecommerce, another area where AI agents are at an advanced stage of experimentation and implementation, providing consumers with the possibility of greatly shortening choices and decisions, right up to the conclusion of the entire transaction by delegating a large part of the entire process to agent models.

The stakes

Capgemini suggests that agents will shift the centre of gravity of operational models: less dependence on physical hubs and more 24/7 services, enabled by multi-agent ecosystems and intelligent data centres. On the organisational front, a hybrid 'human-agent' workforce will emerge, with leaner teams, augmented decision-making, continuously monitored risk and real-time operations.

The message is clear: those who manage to integrate cloud-native AI agents into core processes and embed them in new embedded channels, hidden within processes to make them usable automatically and easily by consumers, can achieve double-digit cost cuts, +25% on satisfaction metrics, acquisition costs down by up to 30% and compressed timeframes - numbers that translate into market share where the experience weighs as much as the product.

The opposite risk is disintermediation: if 'third party' agents live in super-apps or platforms, the company could turn into a mere supplier of commodities. But the agent 'embedded' within the process becomes a valuable tool to keep control of the customer relationship, before someone else does.

There remains a kind of responsibility for companies in this process, vis-à-vis consumers, who are increasingly inclined to delegate their decisions. Like any technology, the potential goes hand in hand with the risks, all the more so when confronted with the enormous value of AI. Algorithmic personal assistants can be of great use in supporting more efficient and faster problem solving, but they cannot replace the human capacity for thought and reasoning. From laziness to passivity, to adapting and submitting to the machine's choices, the step is very short. And it risks being irreversible.

Pierangelo SoldaviniVicecaporedattore
Luogo: Milano
Lingue parlate: Italiano, Inglese
Argomenti: innovazione, digitale, didattica, critpovalute, blockchain
Scheda autore
Trust project