News · Technology
The new artificial intelligence boom: what agentic AI is and why it is attracting so much attention
Agentic AI combines reasoning, planning and tools to perform tasks toward a goal, bringing new possibilities and a greater need for oversight.
— Views
Artificial intelligence is no longer limited to answering questions, summarizing texts or generating images. One of the technologies attracting the most interest now is agentic AI, also called AI agents: systems capable of receiving a goal, breaking it into steps, using external tools and performing tasks with less human supervision than a traditional chatbot. IBM defines it as AI that can accomplish a specific goal with limited supervision, and NVIDIA describes it as a technology that combines reasoning, planning and action to solve multistep problems.
The key difference is simple: a conventional chatbot usually waits for an instruction and responds; an AI agent, by contrast, can plan, decide what to do next and act within a workflow. That opens the door to tasks such as reviewing files, querying databases, writing drafts, coordinating subtasks among several agents and even operating within certain controlled software environments. OpenAI, for example, recently presented an evolution of its Agents SDK for long-running tasks in sandbox environments, and also launched “workspace agents” in ChatGPT for teams that want to create shared agents for complex workflows.
That is why this technology is drawing so much interest. It is not just about “talking to AI,” but about delegating digital work to it. Google Cloud has been promoting the concept of “agentic work” as the next wave of enterprise adoption, while Gartner projects that by 2028, 60% of brands will use agentic AI to offer more personalized interactions. McKinsey, for its part, already speaks of a transition toward an “agentic era,” although it warns that governance still lags behind technical progress.
In practical terms, this means a company could have one agent receive emails, another search internal information, another prepare responses and a fourth request human approval before performing the final action. In programming, these tools are also advancing rapidly: OpenAI describes workflows in which its agents inspect files, edit code and run tests, while Microsoft presented Phi-4-reasoning-vision-15B this year, a multimodal reasoning model designed for tasks combining text, images and interfaces.
Another reason agentic AI is attracting so much interest is that it represents a step beyond traditional automation. Previously, an automation followed fixed rules; now the system can adapt better to changing contexts. IBM explains that these systems combine the flexibility of large language models with the precision of more structured tools and rules. NVIDIA adds that, in more advanced environments, several specialized agents can collaborate on planning, information retrieval and security.
But the advantages come with challenges. The more power to act an AI is given, the more important controls become. McKinsey warns that many organizations still lack sufficient maturity in strategy, governance and oversight to deploy agentic AI safely. That point is critical: an agent can save time, certainly, but it can also amplify errors if it operates on incomplete data, misinterprets an instruction or lacks clear limits.
In other words, the promise is enormous, but so is the responsibility. Agentic AI does not appear to be a passing fad: major technology companies, consultancies and infrastructure manufacturers are steering products and platforms toward this model. The clearest sign is that the conversation is no longer just about assistants, but about systems that reason, coordinate and execute. That could change everything from customer service to programming, sales, research and companies’ internal operations.
Conclusion:
The AI technology attracting the most attention at this moment is not simply a smarter chatbot, but the ability to create artificial intelligence agents that work toward goals. There are still limitations, risks and considerable noise around the subject, but the market’s direction is clear: AI is moving from being only a tool for consultation to becoming a tool for execution.