Stop Vibe Coding: AI Agents for Elixir Developers

Learn structured oversight. Integrate AI into your Elixir workflow without losing architecture decisions, without dependency, and without the “yes loop” that breaks codebases.

This is not a course about saying “yes” to AI faster. It’s about knowing when to say “no” — and what to do instead.

Available Now: Modules 0–5 are complete.

Bruce Tate
Instructor:

Bruce Tate

8 courses
45 videos
12 hours

Wait, I thought this was vibe coding?

We get it. Most AI courses teach you to say "yes" faster.

Bruce teaches the opposite: the Ask → Plan → Agent framework that keeps you in control. This is about structured oversight, not autopilot acceptance.

If you've been avoiding this course because the words "AI agents" sounded like vibe coding wrapped in jargon — please come back. This is the anti-vibe-coding curriculum. The framework that keeps the architecture decisions yours.

What You'll Learn

This is anti-vibe-coding for Elixir developers. By the end of this course, you'll be able to:

  • Integrate AI into your Elixir workflow without losing the architecture decisions that make your codebase reliable.
  • Apply the Ask → Plan → Agent framework to prevent the "yes loop" that turns AI assistants into a crutch.
  • Use AI in Phoenix and LiveView apps without second-guessing your design, knowing exactly where AI fits and where it doesn't.
  • Recognize when AI-generated code fits your codebase and when it should be thrown away.
  • Build production-ready AI agents in Elixir using OTP's concurrency model, with the same discipline you'd apply to any other Elixir system.
  • Develop the mental models that let you evaluate AI output instead of approving what you can't fully review.

This is deep conceptual understanding of where AI fits in your work, not generic prompting advice.

Why This Course Exists

When AI coding tools first arrived, developers had a choice: use them to write more code faster, or use them to think more clearly with the time they gave back.

Most courses teach the first version. Bruce teaches the second.

The cost of writing code used to be the act of writing. That cost forced you to think before you typed. AI removed that cost — and with it, the design conversation that used to happen inside the work. This course is about putting the conversation back. Not by typing less. By thinking more.

Course Details

Learn to use AI coding agents on real Elixir code without surrendering control over your architecture, your codebase, or your judgment.

You’ll work with tools like GitHub Copilot, Cursor, Claude, and TideWave — not as autopilot, but as structured collaborators inside a workflow you design. You’ll learn to structure your code so AI assistance fits the system. You’ll learn to build your own AI agents in Elixir, using OTP’s concurrency model as the architectural backbone.

By the end, you’ll understand how to integrate AI into your Elixir workflow without it becoming a crutch — and where AI helps versus where it quietly hurts.

What You Get

Video lessons taught by Bruce Tate
Hands-on Elixir code exercises
Project examples you can fork and adapt
Mini-books for each major concept
Future course updates included
14-day money-back guarantee

Course Contents

Overview

How to think about AI agents before you adopt them. The Ask → Plan → Agent framework. What anti-vibe-coding means in practice.

AI Agents Fundamentals

Where AI fits in the development loop. How to evaluate AI output instead of approving what you can't review.

Elixir-Specific Strategies

Why Elixir's functional programming model and OTP architecture are unusually suited to disciplined AI-assisted work.

TideWave and Project Setup

Project rules, agent instructions, and scaffolding that keep AI assistance aligned with your team's conventions.

Iteration, Checkpoints, and Source Control

The discipline of small, reviewable chunks, and knowing when to throw AI output away.

Building AI Agents in Elixir

Design and build your own AI agents using Elixir, OTP, concurrency, supervision, and fault tolerance.

Overview

AI 0.1: Five AI Development Approaches
29:06
AI 0.2: Programming 2.0 and LLM Fundamentals
13:48

AI Agents Fundamentals

AI 1.1: Introduction to AI Coding Agents
13:48
AI 1.2: Choosing AI Agents
7:39
AI 1.3: Code Completion Mastery
24:17
AI 1.4: Editor Agents Deep Dive
21:15
AI 1.5: AI-Assisted Planning
0:00
AI 1.6: Reviewing AI-Generated Code
11:52
AI 1.7: Claude Rules and Guidelines
5:14
AI 1.8: Claude Organization Strategies
8:07
AI 1.9: Sub-Agents and Delegation
15:13
AI 1.10: Slash Commands and Workflows
10:39

Elixir-Specific Strategies

AI 2.1: Elixir Infrastructure Overview
8:30
AI 2.2: Organizing AI Infrastructure
12:45
AI 2.3: Installing Claude Marketplace Plugins
15:20
AI 2.4: Customizing Development Workflows
18:30
AI 2.5: Meta Workflow Generation
22:15
AI 2.6: Research and Planning Workflows
25:40
AI 2.7: Critical Planning Review
20:30
AI 2.8: Functional Core Skills
18:45
AI 2.9: Evaluating Generated Functional Cores
16:20
AI 2.10: Context Management and Ecto Schemas
19:10
AI 2.11: Game Contexts and State Architecture
21:30

TideWave and Project Setup

AI 3.1: Understanding TideWave - Teaching AI About Your Elixir Application
15:30
AI 3.2: Why Elixir Excels for AI-Assisted Development
18:45
AI 3.3: Mastering Context Management and AutoCompact
16:20
AI 3.4: Installing and Configuring TideWave
22:10
AI 3.5: Running Claude Code with TideWave as MCP Plugin
19:35
AI 3.6: Choosing Your Interface and Implementing Contexts
17:50
AI 3.7: Mastering project_eval - Your IEx Swiss Army Knife
12:45
AI 3.8: Reviewing Context Layer Code Quality and Architectural Patterns
16:20
AI 3.9: Checkpoint - Preparing for User Interface Development
14:30

Iteration, Checkpoints, and Source Control

AI 4.1: Understanding Checkpoints and Three Sources of Truth
12:45
AI 4.2: Mental Models for Managing State Across Editor, Agent, and Repository
15:20
AI 4.3: Building Project Infrastructure with agents.md and claude.md
18:35
AI 4.4: Connecting Stable and Dynamic Guidance Through claude.md
16:45
AI 4.5: Test-Driving the GitHub Integration Workflow
19:25
AI 4.6: Introducing Claude Desktop, Completing the PR, and Intentionally 'Vibing' the UI
22:10
AI 4.7: The Discipline of Throwing Code Away
17:30
AI 4.8: Building Dictionaries with Claude Co-Work and Improved Prompting
20:15
AI 4.9: Building a Stateful Component Skill from a Working Example
18:45
AI 4.10: Building the UI with Proper Infrastructure
16:20

Building AI Agents in Elixir

AI 5.0: Making AI Work for Your Customers
1:30
AI 5.1: Building a Phoenix Prompting GenServer
17:00
AI 5.2: Hardening the Prompter with Timeouts and Concurrency
16:30
AI 5.3: Course Wrap-Up and Capstone Challenge
2:00

About Bruce Tate

Bruce Tate

Bruce is the author of more than ten programming books, including Seven Languages in Seven Weeks, Programming Phoenix, and Designing Elixir Systems with OTP. He has trained thousands of developers and currently writes the Stop Vibe Coding article series, including posts that introduce the Ask → Plan → Agent framework this course teaches.

"I don't want you to write Elixir faster with AI. I want you to write better Elixir, with AI, without losing what makes your codebase yours. That's a different goal. That's what this course is about."

- Bruce Tate

Read Before You Buy

If you want to see Bruce's teaching style and the framework this course is built on, start with the foundation post:

How I Work With AI Coding Agents in Elixir (And Why "Vibing" Is a Trap)

That post introduces the five-mode framework that stops vibe coding. It's from this course. Read it free before you commit.

Pricing

Individual Course

$500.00

Lifetime access to the AI course.

Buy AI Course

Yearly Access

$350.00

Yearly access to all 8 courses. Best value if you take 2 or more courses.

Subscribe for $350.00

14-day money-back guarantee on all purchases.

FAQ

Is this just about using Copilot and Cursor?

No. The tools change every six months — what doesn't change is the discipline. This course teaches the framework (Ask → Plan → Agent, structured oversight, prompt layering) that works with whatever tools come next.

Will this teach me to write code faster?

Sometimes. But that's not the goal. The goal is to use AI without losing the architecture decisions, mental models, and learning that make you a better developer over time. Speed is a side effect, not the point.

I've been avoiding AI tools. Is this for me?

Yes, especially. Bruce takes seriously the fears that come with AI-assisted development. This course gives you a framework for engaging without becoming dependent.

I already use AI heavily. Is this for me?

Also yes. The structured oversight framework will likely catch patterns you've absorbed without examining. Most developers using AI heavily are making subtle mistakes they don't yet have language for. This course gives you that language.

How does this fit with the other Groxio courses?

This course assumes Elixir fundamentals. If you're new to Elixir, start with the Elixir course or grab the yearly subscription. Everything cross-references.

Ready to stop vibe coding?

Anti-vibe-coding for Elixir developers. From the author of Seven Languages in Seven Weeks.

Video Thumbnail
$500.00
Buy AI Course
Or subscribe to all courses: Subscribe for $350.00

What You Get

12 Hours in 45 Videos
A book per course
Hands-on Exercises
Course Certificate