Getting Started

Welcome to Joeven

What Joeven is, how each page works, what you will build, and how to learn agents without skipping the boring checks.

What Is an AI Agent?

An agent pursues a goal by observing, acting, and looping. The model is not the agent. The loop around the model is.

The Agent Loop

Think, act, observe, repeat. Each turn adds tokens and a trace. Stop on success, budget, or a human handoff.

Agents vs Chatbots vs Workflows

Chatbots reply. Workflows follow your flowchart. Agents choose the next step. If you can draw the path, do not use an agent.

The Learning Path

Joeven is ordered like a degree, compressed: Python, math, models, the agent stack, then projects. Skip around only after the foundation.

Setup: Python, Keys, and Safety

Install Python, use a virtual environment, load keys from the environment, and treat every model output as untrusted.

Tokens, Tools, and Goals

Every agent spends three scarce things: context tokens, tool permissions, and a goal you can check with a function.

Your First Tiny Agent

A complete think-act-observe loop: fake model, two tools, a transcript, a step budget, and a final answer for job 17.