Artificial intelligence is entering a new stage, driven by an unprecedented speed of adoption of agentic AI.

After the generative AI revolution —which allowed us to converse, create, and automate through language— comes a second wave in which models no longer just respond, but perceive, reason, decide, and act. They are capable of executing complex tasks, collaborating with other agents, and making decisions in specific contexts.

In this new scenario, we must consider what agents will do for our businesses and for us as professionals.

AI agents do not replace people, but rather amplify them. They act as digital collaborators capable of taking on repetitive, analytical, or coordination tasks, while humans maintain judgment, creativity, and responsibility. They operate within human in the loop models, where supervision and joint learning ensure reliable and ethical results.

Their value does not lie in total autonomy, but in the symbiosis between human intelligence and algorithmic capacity: a new way of working in which technology learns from people and people expand through technology.

It is no wonder, then, that in the AI Agent Bible 2025 report, CB Insights describes this evolution as a new infrastructure for digital work. And this is no exaggeration: AI agents are ceasing to be a tool to become a new layer of productivity and operational intelligence within organisations.

1. MARKET

The agentic AI ecosystem has left experimentation behind to consolidate as a new technological and business market with its own dynamics. Companies are no longer experimenting: they are investing, integrating, and competing to dominate this new technological layer. Startups and major platforms are shaping a dynamic and consolidating ecosystem, where innovation, investment, and collaboration redefine the map of digital work

1.1. Expanding

In just over a year, more than 500 startups and nearly $4,000 million in investment have driven a new business tissue focused on agentic intelligence. It is a growing ecosystem that not only automates tasks but integrates operational intelligence into strategic processes and redefines digital productivity within organisations.

At the same time, the major players —Microsoft, Google, Amazon, and Meta— are competing to dominate different layers of this new infrastructure: productivity, data, voice, or user experience. This is not a trend: it is a structural reconfiguration of the global technology value chain.

1.2. M&A on the rise

The market is accelerating with more than 35 corporate operations in 2025, led by players such as Moveworks, Weights & Biases, or OfferFit. Large platforms are buying speed, integration, and talent to position themselves in this new agentic infrastructure.

1.3. Monetisation

The first business cases are already tangible. Cursor ($500M in recurring revenue) or Replit Agents (+$150M) prove that AI agents already compete in profitability with traditional software. The fastest-growing applications are concentrated in code development, customer service, and marketing, where productivity is multiplied.

The processes of the future are not designed for people assisted by machines, but for machines that collaborate with people.

2. A NEW WAY OF DIGITALISING

This situation will bring new technological approaches to the business world. We will start to see agents replacing traditional technological solutions. In a few years, it will no longer be as relevant which tools you use, but which cloud of agents secures your day-to-day and solves your needs

2.1. From co-pilots to autonomous collaborators

AI is evolving from reactive assistance to the capacity to act autonomously, taking on roles that we previously considered exclusive to human talent. Examples like Devin (Cognition Labs) or Adept ACT-1 show how agents can already plan, write code, analyse markets, or coordinate projects with minimal supervision.

This technological change opens up a new framework of collaboration between people and systems. Companies are no longer asking what they can automate, but what they can delegate. And this is the strategic difference we amplify at RocaSalvatella: we champion supervised, human, and constantly learning AI that combines individual and technological capabilities. Our approach puts people at the centre, ensuring that every AI agent is an extension of the organisation's collective intelligence and a catalyst for measurable and sustainable results.

2.2. Birth of agentic commerce

If e-commerce transformed distribution, agentic commerce will transform the relationship between supply and demand. The Agentic Commerce Protocol, driven by Stripe and OpenAI, allows agents to negotiate, buy, and pay on behalf of people or companies.

This changes the nature of the digital customer: the next consumer could be another agent. A disruption that will reshape the entire e-commerce value chain

2.3. Voice as the new interface

Voice is establishing itself as the main channel of interaction with intelligent systems. Following Meta's acquisitions of PlayAI and WaveForms AI, oral conversation is becoming the new "click".

Speaking with technology will be as natural as writing to it.

2.4. Supervision and trust

Autonomy brings new risks and demands trust. Startups like Cekura, Coval, or Larridin are emerging, specialised in observability, governance, and security. Without transparency and control, agentic AI can become an operational risk.

Trust is now at the core of agentic design, giving rise to a new discipline: AI Operations, aimed at ensuring the reliability and safety of autonomous processes.

3. TECHNOLOGY

AI agents do not represent the distant future of technology: they are the immediate foundation of the new model of work and innovation. In just two years, they have evolved from experiments to the operational core of digital innovation, starring in the fastest adoption ever recorded in business technology.

But the advantage does not lie in the model, but in connecting agents designed with purpose-specific data. The organisations that achieve this will be the ones that shape the new economy that agents are already creating.

3.1. Cloud giants' strategies

The big technology players are following divergent routes:

• Amazon, with an open and scalable infrastructure.

• Google, with its Gemini ecosystem.

• Microsoft, with Copilot as a productivity layer.

• Meta, strengthening natural interaction through voice and social environments.

Each seeks to dominate a part of the agentic value chain: data, platform, or user experience.

3.2. The new technological architecture

Agents need their own infrastructure.

In 2025, more than 130 startups are working on systems that integrate voice, memory, payments, and security. The future lies in multi-agent architectures, where various AIs collaborate, learn, and supervise each other.

3.3. The new battle for data moats

In the agentic era, the advantage does not reside in the model, but in the context of data that feeds it. Companies like Snowflake are banking on openness and standardisation, while the big players protect their data moats as a competitive advantage.

The most effective agents will be those fed with relevant, reliable, and ethical data. The challenge, however, is not just technical: it is strategic.

Many organisations understand the importance of agents but still struggle with the complexity of integration, security challenges, and the phenomenon of "more bad faster", where AI simply accelerates flawed processes. Even so, the opportunity is immense for companies capable of governing this transition with judgment and business vision.

Competing in the agentic era

The consolidation of agentic AI accelerates a structural change that redefines the way organisations digitalise, operate, and compete. In a moment of market effervescence, with models challenging traditional big environments, companies must rethink their technological architectures and the relationship between people, processes, and data.

Digital transformation is entering a new stage where agentic intelligence becomes an essential layer of productivity and decision-making.