*By Jorge Moskovitz
For many, the future seemed distant. But, observing the latest discussions, we can see that it is already knocking on our door! And it involves Agentic AI, the new technology hype, which represents an evolutionary leap in Artificial Intelligence (AI), combining cutting-edge innovations to enable autonomous agents to analyze data, set goals and act with minimal human supervision. Although most AI-related speeches seem like “more of the same”, these agent systems will truly transcend the automation of specific tasks to achieve near-human cognition, transforming machines into problem solvers that learn and evolve with each interaction. When we talk about Agentic AI, I fully believe that we will move from hype to practice.
The capabilities of autonomous agents come from the combination of several technological approaches. Large Language Models (LLMs) allow systems to understand and respond to commands in a fluid manner, while machine learning applications identify patterns and make decisions based on huge amounts of data. Reinforcement learning techniques enable the evolution of these agents based on the consequences of their own decisions, improving their next moves. When integrated, these technologies are moving towards transforming entire industries and redefining our interactions with machines. However, following this path involves facing major challenges, both operational and cost-related.
The depth involved in implementing Agentic AI requires innovative and structured procedures for process changes, which in many cases can overshadow the benefits of use, for example. The chances of poorly supervised autonomous systems making harmful decisions are high, especially when they have access to sensitive data or large networks.
Vulnerability to external attacks is a critical issue. Hackers can exploit autonomy to cause damage that can replicate and amplify its effects. Robust measures, such as the implementation of emergency shutdown systems, advanced cybersecurity and encryption frameworks, and the definition of clear operational boundaries, in addition to continuous governance and rigorous monitoring, are essential to protect agents against failures and manipulation.
In the ethical field, concerns about algorithmic biases and lack of transparency are even more evident. Agent systems are capable of perpetuating discriminations present in training data, while the complexity of their decisions makes it difficult to track the criteria used. The advanced automation brought by Agentic AI can also generate economic and social impacts, such as the unequal displacement of jobs. Therefore, adopting principles of impartiality, accountability, transparency and explainability, and carrying out constant audits to ensure that decisions are understandable and aligned with social and organizational values is imperative.
Another major challenge is regulatory compliance, such as the General Data Protection Regulation (GDPR) in Europe and the General Law on the Protection of Personal Data (LGPD) in Brazil, which require data to be processed rigorously and autonomous systems to operate within clear limits. The lack of a unified global standard for regulation increases risks, exposing companies to legal risks, especially in cases of ambiguity regarding responsibility for incorrect decisions or damages caused by agents, resulting in severe penalties and compromising the company's reliability. For this reason, investing in compliance, with recurring external audits, shared responsibility mechanisms and data management protocols, is essential.
Agentic AI relies on robust and reliable data to operate, making it essential to implement data integration, quality, and accessibility platforms that eliminate silos and promote a data-driven culture. Tools that enable real-time analysis and intuitive visualization create the foundation for implementing Agentic AI solutions safely, accurately, and efficiently.
These solutions will play a central role in enabling companies to address the challenges of Agentic AI, while promoting responsible and sustainable implementation of the technology, with robust infrastructure and a focus on delivering value. This facilitates automated decision-making based on clear, real-time insights. Only in this way will Agentic AI drive significant advances in emerging technologies, such as the use of extended reality and autonomous vehicles, for example.
But its real impact will be in software development. Agentic AI will completely transform the project lifecycle. Tools will allow developers to take on the role of supervisors, speeding up the delivery of solutions that would otherwise take years to months or days.
Given this context of advances and challenges, businesses need to position themselves in digital transformation and in the adoption of Agentic AI, seeking solutions that democratize access to insights based on quality data and enhance the intelligence of organizations. As Agentic AI shapes the future, it is necessary to choose strategic partnerships and build an ecosystem in which humans and technology will collaborate harmoniously, enabling companies to innovate with confidence, overcome challenges and lead an increasingly competitive market. I see that the future has already arrived and we must be prepared to face it with responsibility and innovation!
*Jorge Moskovitz, Account Executive at Qlik
Notice: The opinion presented in this article is the responsibility of its author and not of ABES - Brazilian Association of Software Companies
EN 
PT 
