← sources

Are we learning to build AI agents, or just learning how to use agent frameworks?

source: reddit (reddit:AI_Agents)published 2026-11-19 09:33ingested 2026-11-20 02:35

retained text

I've been exploring AI agent architectures, and one thing I've noticed is how quickly most learning resources jump into frameworks.

You start with LangChain or LangGraph, define a few tools, connect an LLM, add some memory, and suddenly you have a working agent.

But I wonder how much we actually understand about what's happening underneath.

For example:

* When an LLM decides to call a tool, what's actually happening behind the scenes?
* Why do we need state management? How is it different from just maintaining conversation history?
* When does a simple LLM workflow become an agent?
* What happens when an agent makes a wrong decision halfway through a multi-step task?
* How do we evaluate an agent when the same input can lead to different execution paths?

I feel these questions become much more important once you move beyond basic demos.

Another thing I've been thinking about is whether beginners should learn agent architectures independently of frameworks first.

For example, implementing a basic tool-calling loop in Python before learning LangGraph. You'd probably understand much better why concepts like nodes, edges, state, and conditional routing exist.

Of course, frameworks make development easier, and there's no reason to reinvent everything from scratch. But relying on them too early might make it harder to understand what's going wrong when something breaks.

I've actually started putting together a free AI Agents Fundamentals series where I'm trying to explain these underlying concepts before getting into frameworks. I'm also working on a follow-up series where we'll build a complete AI agent using LangGraph, applying these concepts to a practical application.

**I'll share the channel link in the comments for anyone interested in following along.**

But I'm also curious how others approach this.

If you were learning AI agents from scratch today, would you start by building a simple agent without frameworks, or jump directly into LangGraph/CrewAI and learn the underlying concepts along the way?

And for those already building agents in production, which fundamentals do you think beginners tend to overlook?

document fetched 2026-11-20 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1y05akc/are_we_learning_to_build_ai_agents_or_just