What does ‘agentic’ AI mean? Tech’s newest buzzword is a mix of marketing fluff and real promise

The rise of “agentic AI” has become the technology industry’s latest buzzword, promising capabilities that go beyond today’s generative AI chatbots. But as industry executives and researchers rush to embrace the term, many consumers and even tech professionals are left wondering what exactly makes an AI system “agentic.”

Google searches for the term “agentic” have surged from near obscurity a year ago to peak interest this fall, reflecting both growing curiosity and confusion about what sets these systems apart from existing AI tools.

According to a new report released Tuesday by researchers at the Massachusetts Institute of Technology and Boston Consulting Group, agentic AI represents a “new class of systems” that “can plan, act, and learn on their own.” The report, based on a survey of more than 2,000 business executives worldwide, emphasizes that these systems “are not just tools to be operated or assistants waiting for instructions” but rather function as “autonomous teammates, capable of executing multistep processes and adapting as they go.”

The fundamental difference between conventional chatbots and agentic AI lies in their ability to take action. While chatbots like the original ChatGPT rely on large language models that essentially perform sophisticated word prediction based on training data, agentic AI systems are designed to execute tasks independently.

“A generative AI-based chatbot will say, ‘Here are the great ideas’ … and then be done,” explained Swami Sivasubramanian, vice president of Agentic AI at Amazon Web Services, in a recent interview. “It’s useful, but what makes things agentic is that it goes beyond what a chatbot does.”

Sivasubramanian, who assumed his role earlier this year to lead Amazon’s cloud computing division’s work on AI agents, believes these systems represent a transformative shift. “I truly believe agentic AI is going to be one of the biggest transformations since the beginning of the cloud,” he said.

At its core, an AI agent functions like a traditional computer program that executes specific jobs but with a crucial difference – when combined with a large language model, it can independently search for knowledge and complete tasks without requiring step-by-step instructions. This means that instead of merely helping draft an email, an agentic system could potentially manage the entire communication process – receiving messages, determining appropriate responses, and sending replies autonomously.

For consumers, initial encounters with AI agents will likely come through everyday applications like online shopping. These agents could operate within set parameters – working with a budget and preferences to make purchases or arrange travel using a user’s payment information. Future iterations may tackle more complex tasks by accessing computer systems while following established guidelines.

Thomas Dietterich, a professor emeritus at Oregon State University with decades of experience developing AI assistants, envisions practical applications: “I’d love an agent that just looked at all my medical bills and explanations of benefits and figured out how to pay them,” or one that functions as a “personal shield” against email spam and phishing attempts.

While enthusiastic about AI systems with “freedom and responsibility” to refine goals and adapt to changing conditions, Dietterich questions companies that describe “any action a computer might do, including just looking things up on the web” as “agentic.” He also notes the potential for these systems to work collaboratively, even forming “coalitions” of “subagents.”

Despite the current buzz, the concept of AI agents is far from new. Milind Tambe, a Harvard University professor who has researched collaborative AI agents for three decades, finds the sudden popularity of “agentic” as an adjective “amusing.” The term, previously more common in fields like psychology and chemistry, has been adopted by the tech industry to describe autonomous systems.

“In the 1990s, people agreed that some software appeared more like an agent, and some felt less like an agent, and there was not a perfect dividing line,” Tambe explained. “Nonetheless, it seemed useful to use the word ‘agent’ to describe software or robotic entities acting autonomously in an environment, sensing the environment, reacting to it, planning, thinking.”

Prominent AI researcher Andrew Ng helped popularize the adjective “agentic” over a year ago to describe a broader range of AI capabilities. Initially, he appreciated that the term was primarily used by technical professionals, noting in a June 2024 blog post that articles discussing “agentic” workflows were “less likely to be marketing fluff and more likely to have been written by someone who understands the technology.”

As the technology evolves and major companies like OpenAI, Amazon, Google, IBM, Microsoft, and Salesforce continue developing autonomous AI systems, the distinction between genuine innovation and marketing hype around “agentic AI” remains a challenge for industry observers and potential adopters alike.