AI Agents & Workflows - The Practical Guide
Learn how to programmatically build AI Workflows & Agents. All the theory, plenty of examples.
About This Course
What is this course about?
"AI Agents" is everywhere — but behind the buzzword lies a world of real, transformative opportunities.
From data transformation to content generation, customer service to automated research — AI-powered workflows and agentic systems can revolutionize how you work. Your imagination is the limit!
AI Agents – The Practical Guide cuts through the noise. Instead of chasing buzzwords, this course gives you the crucial theory, plenty of real-world examples, and concrete code snippets to truly understand how LLMs, "normal code," and data work together to bring AI-powered applications to life.
You'll learn exactly what "AI Agents" are, how they differ from (and still relate to) "AI Workflows," and — most importantly — how to build and use both. By the end of this course, you'll be able to create your own AI-powered applications and agentic systems from scratch.
All code examples use Python and the OpenAI API / SDK, but the knowledge you gain applies to any programming language or model you use in your day-to-day work.
What You'll Learn
Master the full spectrum of AI-powered automation — from foundational concepts to advanced multi-agent architectures:
Workflows & Agents
Understand the difference between AI Workflows and AI Agents — and learn to build both from scratch using OpenAI's models via API & SDK.
Data & Automation
Transform input data with AI, build powerful automations, and integrate with third-party services like Slack.
Memory & Self-Evaluation
Manage short- and long-term memory for your agents. Use AI for self-evaluation to create smarter, more reliable systems.
Multi-Agent Systems
Build multi-agent architectures, share data between agents, and split work between universal and specialized agents.
Hands-On Learning
Learn the WHAT, WHY, and HOW — with real examples you can apply immediately.
Real-World Examples
Build practical AI systems across multiple domains:
- Content generation workflows
- Customer support automation
- Automated research agents
- Human-in-the-Loop integrations
Language-Agnostic Skills
Gain knowledge that transfers to any tech stack:
- Detailed explanations of core concepts
- Understand the patterns, not just the code
- Apply to Python, JavaScript, or any language
- Works with OpenAI, Anthropic, or other models
Who Is This Course For?
See The Course In Action
Curriculum
- Welcome To The Course! (1:22)
- What Is An AI Agent? (2:13)
- General vs Task-specific Agents (2:39)
- Where Agents Run / Execute (2:07)
- AI Agent Harnesses, LLMs & Limitations (3:25)
- How Agents Use Tools (5:15)
- Understanding Session Context (2:17)
- Core AI Agent Building Blocks - Overview (2:25)
- AI Agents vs AI Workflows (2:18)
- Module Introduction (0:57)
- Three Main Ways For Controlling AI Agents You Should Know (0:49)
- Managing Agent Tools & The Environment (3:42)
- Choosing The Right Model (1:17)
- Understanding Developer-provided System Instructions (1:39)
- Providing Your Own Instructions & Understanding AGENTS.md / CLAUDE.md (3:29)
- Understanding Agent Skills (8:28)
- Understanding Agent Memory (3:23)
- Sometimes Important: Humans In The Loop (2:57)
- Module Introduction & Expectations (1:35)
- Building Workflows - The Basics (1:20)
- Workflows vs Agents (2:39)
- Building Visually with n8n - First Steps (2:38)
- Running a Demo n8n Project Locally (1:41)
- Exploring n8n & Its Workflow Builder (6:44)
- Exploring & Understanding a Realistic Workflow (4:24)
- Onwards To A Code-based Solution! (3:15)
- Exploring & Understanding A Code-based Workflow (6:20)
- Module Summary (1:11)
- About This LEGACY Section
- Module Introduction (2:21)
- No Code vs With Code (2:07)
- Building AI Apps & Using AI Programmatically (4:12)
- Proprietary vs Open (Local) LLMs (5:51)
- Using Open LLMs
- Understanding Our Development Environment (2:29)
- Creating a New Python Project (using "uv") (1:43)
- OpenAI Setup & Pricing (5:41)
- Getting Started With A First Example Workflow (2:44)
- Preparing HTTP Requests For The OpenAI API (8:40)
- Choosing & Using a Model (2:04)
- Prompt Engineering (4:35)
- Extracting & Using the LLM Response (4:50)
- More on the OpenAI API & SDK
- Code Deep Dives vs Provided Code Snippets
- Use Those Docs! (1:38)
- Using The OpenAI Python SDK (5:49)
- Leveraging Few-Shot Prompting (4:02)
- Generating Prompts Dynamically With Dynamic Content (2:12)
- Building Multi-Step & Multi-Model Workflows (6:47)
- Workflows vs Agentic Systems (1:34)
- Using Locally Running Open Models via Ollama (8:08)
- Enforcing & Using Structured Outputs (10:53)
- More On JSON Schemas & Structured Outputs
- Structured Outputs via SDK & Pydantic (3:56)
- Using Prompt Engineering To Control Output
- Onwards To Another Example (5:38)
- Generating Images In a Workflow (6:27)
- Controlling Workflow Execution with Control Flow Adjustments (2:45)
- Control Flow In Action (8:42)
- Adding a "Human In The Loop" (6:51)
- Integrating External Services - Example: Slack (6:21)
- Important: Potential Problems & Security Risks
- Module Introduction & Expectations (1:55)
- Setting Up & Starting the n8n Demo Project Server (3:08)
- Creating Agents Visually With n8n (3:42)
- Understanding Tools & The Agent Harness (in n8n) (2:51)
- Running The n8n Agent (2:45)
- Finding Key Agent Building Blocks in n8n (0:59)
- Onwards To Code-based Agents! (3:53)
- Defining Tools As Functions (2:26)
- Analyzing The Agent Loop (3:23)
- How The Agent Learns About Tools & Behaves Correctly (6:03)
- Using Provider-native Tooling To Make Things Easier (5:14)
- Onwards To A General Agent (1:43)
- Analyzing Tools, Instructions & Context Engineering For A General Agent (6:19)
- Demo: Using The General Agent (3:15)
- Analyzing An Example Sandbox Implementation (2:44)
- Adding A Human To The Agent Loop (2:35)
- How Agent Execution Is Constrained (3:21)
- About This LEGACY Section
- Module Introduction (1:36)
- How LLMs (Do Not) Use Tools (6:04)
- Implementing Tool Use From Scratch (11:24)
- Using OpenAI's Function Calling Feature (10:23)
- Building a Multi-Tool Versatile Agent (11:16)
- Using Advanced AI Models
- Building Reusable Elements With Classes (7:53)
- Getting Started with a Multi-Agent System (7:14)
- Extracting Website Content
- Building & Connecting Specialized Agents (10:24)
- Universal vs Specialized Agents (3:38)
- Agent Memory: Short-Term & Long-Term (5:21)
- Wrap Up (1:09)
- Module Introduction (0:52)
- Library & Framework Options - An Overview (0:58)
- Libraries vs Frameworks (2:33)
- An Introduction To The "eve" Framework (2:23)
- Diving Into An "eve" Project (2:53)
- Configuring Tools For The "eve" Agent (2:32)
- Connecting The Agent To A Knowledge Base & Skills (3:52)
- Seeing The "eve" Agent In Action (3:58)
- Deep Dive: Connecting Slack As A Channel (9:06)
- More On Tunnels / cloudflared
- Onwards To CrewAI (0:34)
Course Prerequisites
Here's what you need to get the most out of this course
- Basic understanding of programming concepts.
- No prior AI experience necessary to succeed in this course.
All pre-requisites are covered by courses in our "Academind Pro" Membership.
Your Instructor
Maximilian Schwarzmüller
Founder & Instructor
Self-taught developer with 3,500,000+ students and 900,000 YouTube subscribers. I co-founded Academind with Manuel Lorenz to help people master new skills and build amazing projects.
Join 1779 happy students!
Choose the plan that works best for you
Single-Course License
Full access to "AI Agents & Workflows - The Practical Guide"
This is a one-time payment that grants access to this course only, not to any other courses.
Academind Pro Membership
Unlimited access to this and all other current & future courses!
This is a recurring payment. You can cancel anytime from your profile. For more info, contact Academind.
Continue Your Learning Journey
Expand your skills with these hand-picked courses that complement what you'll learn here.
