Course

AI Agents in Production

The engineer's guide to shipping agents. Start from the one distinction that matters — an **agent** decides its own steps, a **workflow** follows yours — then go deep on the machinery: function calling, MCP, memory, the agent loop and the 2026 stack, the ReAct / Plan-and-Execute / Reflexion patterns, and why agents fail in production. Every screen is a drawn schematic, grounded in Anthropic's *Building Effective Agents* and hard-won production lessons. No hype, no pinned model versions.

7 lessonsIntermediate52 min

Start the course Get the app

Course lessons

What is an agent (vs a workflow)

The one distinction that decides how you build: who owns the control flow, you or the model.

›

Function calling, deeply

How a text model actually acts on the world: it never runs your code — it asks yours to.

›

MCP for engineers

One open protocol between agents and tools — turning N×M custom glue into N+M reusable pieces.

›

Agent memory

Why long-running agents forget, and how real memory is engineered around a finite window.

›

The agent loop & the 2026 agent stack

Strip away the frameworks: an agent is one small loop, wrapped in a production stack.

›

Agent patterns — ReAct / Plan-and-Execute / Reflexion

Three ways to run the loop, and how to pick the one that fits your task.

›

Guardrails, agent safety & why agents fail in production

The failure modes a demo never hits — and the guardrails that make an agent shippable.

›

Take the full course

Learn it properly — tracked & personalized

Follow the whole course as a guided journey — sequential lessons, progress, streaks and a personalized path, all in the app.

Get the app