While the technology behind AI-powered chatbots quickly captured the public imagination, an even more powerful application of generative artificial intelligence has been creating a buzz among business leaders. It’s called agentic AI.
“This innovative technology is not just another industry buzzword; it’s a paradigm shift that’s poised to redefine the boundaries of AI capabilities,” tech guru Bernard Marr wrote Monday in his Intelligence Revolution newsletter.
“At its core, agentic AI refers to artificial intelligence systems that possess a degree of autonomy and can act on their own to achieve specific goals,” he noted. “Unlike traditional AI models that simply respond to prompts or execute predefined tasks, agentic AI can make decisions, plan actions, and even learn from its experiences — all in pursuit of objectives set by its human creators.”
“Agentic AI is the hottest thing going right now,” observed Jason Wong, a vice president analyst with Gartner, a research and advisory company based in Stamford, Conn.
He explained that the technology will not only understand intent and do very simple things like retrieve information and generate some response based on that, but it can also take action. “So, it could retrieve an API or a tool. Or even generate code, like generating Python code to solve a problem,” Wong told TechNewsWorld.
“The agency behind it is highly variable, but it’s AI coupled with tooling,” he continued. “It has the ability to plan how to address your question, your problem, and then activate the tooling and solve your problem.”
Step Beyond Gen AI
Scott Dylan, founder of NexaTech Ventures, a venture capital firm in Manchester, England, maintained that agentic AI takes a significant step beyond generative AI. “While generative AI focuses on creating content — text, images, code — based on existing data, agentic AI has a sense of autonomy,” he told TechNewsWorld. “It can make decisions, take actions, and adapt in real-time without needing constant human input.”
“Think of it as moving from a tool that provides suggestions to one that independently executes tasks, learning from the environment it’s deployed in,” he said.
Agentic AI represents a significant evolution from traditional generative AI by incorporating self-prompted reasoning, dynamic compute allocation and adaptive problem-solving capabilities, added Dev Nag, CEO and founder of QueryPal, an enterprise chatbot in San Francisco.
“Unlike generative AI, which primarily focuses on producing content based on input prompts, agentic AI can autonomously allocate more ‘thinking time’ to complex tasks, employ hidden chain-of-thought search spaces, and utilize reinforcement learning to optimize its reasoning processes,” he told TechNewsWorld.
“This shift allows agentic AI to tackle more sophisticated problems and adapt its approach based on the task at hand, moving beyond mere text generation to more human-like problem-solving across various domains of tokenizable data,” he said. “It’s fair to say that modern agentic AI — like OpenAI’s o1 — builds on generative AI as its infrastructure but can accomplish a wider range of goals.”
Transformative Technology
Agentic AI’s powerful capabilities can be transformational for many businesses.
“Agentic AI can transform industries by automating not just repetitive tasks but also complex decision-making processes. For example, in supply chain management, agentic AI could predict and react to disruptions in real-time, optimizing routes and inventory without human intervention,” Hodan Omaar, a senior AI policy analyst at the Center for Data Innovation, a think tank studying the intersection of data, technology, and public policy in Washington, D.C. told TechNewsWorld.
“Businesses are on the verge of a massive shift due to agentic AI,” Dylan added. “It’s not just about automating processes but empowering systems to handle complex decision-making. In finance, that level of autonomy could drive more personalized customer service and fraud prevention systems that evolve with the threat landscape without the need for constant human oversight.”
“One aspect that excites me is its potential in fields like health care,” he said. “Imagine a health care system that not only diagnoses based on symptoms but actively monitors patients post-diagnosis, adapting treatment plans as it learns from ongoing data. While this is a long-term vision, the groundwork being laid by agentic AI is getting us closer to that reality.”
Nag maintained that agentic AI could revolutionize fields like law, medicine, and finance by automating complex cognitive tasks, potentially displacing jobs involving routine analysis but also creating new roles focused on AI oversight and human-AI collaboration.
“The ability of agentic AI to scale at runtime to solve increasingly difficult problems without necessarily requiring larger models or more training data could democratize access to advanced AI capabilities, allowing smaller businesses to leverage powerful AI tools,” he added.
“This new paradigm of runtime scaling introduces a novel dimension to AI development beyond just hardware and training data scaling, which have been the battleground among AI companies for the last two years,” he said.
Shared Brain, Shared Problems
Like generative AI, agentic AI has its problems. “Inevitably, because AI agents use a language model as their ‘brain,’ they share at least all the problems that generative AI does and then some,” noted Sandi Besen, an applied AI researcher at IBM and Neudesic, a global professional services company.
“Additionally, when you start using multiple agents in tandem and provide them with the ability to work with one another, the innate variability that exists in generative AI is compounded,” she told TechNewsWorld. “However, there are certainly methods you can use to mitigate against this, such as ensuring there is proper evaluation and human in the loop included in the AI system.”
“Agentic AI, like other forms of AI, has the potential to advance users’ productivity. By carrying out the multiple steps involved in many tasks, it is able to automate more work and save users both time and money,” added David Inserra, a fellow for free expression and technology at the Cato Institute, a Washington, D.C.-based think tank.
“While some will inevitably use such an AI tool for malicious reasons or to create content that some find offensive, the many positive applications of this technology means it should be allowed to flourish, free from burdensome government regulation like what we see in the EU,” he told TechNewsWorld. “As a result of such regulations, major tech companies are already withholding new AI tools in Europe, leaving Europeans worse off.”
Closer to AGI?
Since agentic AI gives gen AI the power to act, does it advance the field closer to the holy grail of artificial general intelligence (AGI) and true thinking machines?
“A fundamental trait of general intelligence, whether in humans or animals, is the ability to adapt — sensing environmental signals, responding to them, and learning from those responses. In this regard, agentic AI marks a small yet meaningful step toward general intelligence.” Rogers Jeffrey Leo John, co-founder and CTO of DataChat, a no-code, generative AI platform for analytics, in Madison, Wisc., told TechNewsWorld.
“However,” he added, “we are still far from reaching true general intelligence, which would be capable of applying knowledge acquired from one situation to a completely different context.”
Shawn DuBravac, CEO and president of the Avrio Institute, a technology consulting firm for CxOs and executives, also in Madison, doubted agentic AI would be a path toward AGI. “I would argue that agentic AI is not a precursor of AGI,” he told TechNewsWorld. “It’s not clear we reach AGI through linear progression from current AI technologies like agentic AI.”
“In fact, I think it will be unlikely,” he continued. “If we reach AGI, I believe the path will involve breakthroughs and new paradigms of intelligence that differ significantly from what we have accomplished so far and what we are likely to accomplish in the coming years.”
