According to Sam Altman, 2025 is the year of AI Agents. Agentic AI, or using AI as agents is the big AI push for 2025. But what, exactly, is an AI Agent?
As discussed, there is a broad range of what is defined as AI, but for these purposes, we will look at Large Language Models (LLMs) acting on behalf of a human. In general, an AI Agent is prompted to accomplish a set of tasks to reach a goal set by a human. This could be as simple as helping a customer with technical support or as complicated as carrying out a marketing strategy.
The AI Agent can collect data and interact with humans, APIs or other AIs to accomplish the stated goals, but the path taken to accomplish those goals are solely determined by the AI. This goes beyond simply automating tasks. Automated tasks follow a set path, but AI Agents determine their own path. It’s a little like an old-school phone attendant menu where you press “1” for sales and “2” for customer service (Automation) and an AI Agent who can answer the phone and help the caller right then and there or determine that the call needs to be escalated to a specialist (Agentic AI). (See Euphonia.ai for AI Voice Agents.)
#agentic ai #AGI #AI #AI Agent #chatgpt #euphonia #euphonia.ai #LLM