Agentic AI Explained
NVIDIA Deep Learning Institute
Last updated June 18, 2026
Agentic AI Explained is a short, lecture-based introduction from NVIDIA's Deep Learning Institute to the idea of AI agents — software that perceives its surroundings, reasons about them, acts, and learns from the result. Across about an hour of videos, it builds up the core loop that distinguishes an agent from an ordinary program, walks through the building blocks (perception, reasoning, action, and feedback), and surveys where agents are being put to work across industries. It closes by sketching how you would begin building a simple agent of your own. A concept-first orientation rather than a coding workshop.
What you'll learn
- What makes a system "agentic" versus a conventional program
- The agent loop: perception, reasoning, action, and feedback
- How agentic and generative AI complement each other
- Real-world agent applications and where to start building one
Frequently asked questions about Agentic AI Explained
Who is Agentic AI Explained for?
Curious newcomers and technically-minded professionals who want a clear conceptual picture of AI agents without committing to a hands-on build.
Is Agentic AI Explained free?
Yes — Agentic AI Explained is completely free to take.
What are the prerequisites for Agentic AI Explained?
A basic understanding of programming (especially Python) and familiarity with core concepts in mathematics or data science; some knowledge of machine learning or AI principles is helpful but not required.
Why we suggest this course
A fast, plain-language grounding for anyone who keeps hearing "agentic AI" and wants to understand what it actually means before going deeper. The distinct takeaway is the perception–reasoning–action–learning loop — the mental model that makes every later agent course easier to follow.