Code With Antonio says his old format, a polished rebuild, hides the part of coding that matters most now: dead ends, refactoring, and judgment under pressure.
Antonio is trying to retire the clean tutorial as his main teaching form. He says the old model, in which he first solves a project for himself and then records a smooth rebuild for viewers, now hides the parts of coding that matter most: the architecture choices, the debugging, the dead ends, the judgment calls. In the age of AI, he argues, a video that merely gets from point A to point B is less useful than one that shows how a developer decides what point B should be. His answer is a new workshop built around agentic coding, where the mess is the lesson and the goal is to teach viewers how to think like him.
Antonio opens by describing the old formula behind his channel: he builds a project once for himself, settles the architecture, fixes the bugs, and then records a second pass that appears smooth and inevitable. The appeal is obvious. Viewers can copy along and, as he puts it, end up with a finished project after enough time.
I build it for myself first. In this first build, I decide what the architecture is going to look like.
0:26
With this method, everything that I do when I record myself works on the first try.
0:56
He now sees that format as less valuable than it once was. The point of a tutorial cannot simply be to hide friction, because AI already changes the cost of producing code and the appetite for watching someone reproduce a polished path.
His larger claim is that coding education needs to move because the old promise is less useful in the age of AI. Three years ago, he argues, a long step-by-step rebuild could still feel like a practical shortcut. Now it risks training viewers to follow instructions rather than to make decisions.
In the age of AI, this simply isn't as useful as it might have been 3 years ago.
1:27
So it's time to rethink how tutorial should look like.
1:27
The alternative he sketches is more chaotic and more honest. Instead of hiding the first attempt, he would record the first attempt itself, including the places where he gets stuck, changes direction, and has to debug in public. The point is not efficiency alone; it is to make the viewer watch the reasoning that produces the code.
You will understand how I think rather than understand what is the next line of code.
2:04
That argument leads him to a distinction he says many people blur: vibe coding and agentic coding. In his telling, vibe coding is outcome-first and spec-light, often used by nontechnical people who let the model run autonomously with minimal human involvement. The prompts are blunt, sometimes almost comic in their ambition: build me a Facebook clone, build me a YouTube clone, build me a million-dollar SaaS.
Vibe coding is essentially only focusing on the outcome rather than the technical specification that goes into building a new feature or an entire new app.
3:16
They just want to let their AI run autonomously.
3:56
Agentic coding, by contrast, is how he wants to work. The human stays in charge as architect and orchestrator, writing prompts that reflect an existing grasp of the codebase rather than a wish to outsource it entirely. In his example, a prompt about consolidating a scattered service across five files is less a plea for help than a compressed design brief.
You are the orchestrator. You are the architect.
4:27
They are simply using it to speed up what they would do anyway themselves.
5:05
Antonio says that shift finally made his ideal teaching format possible. If AI can compress hundreds of hours of thinking and typing into a few hours of prompting, then the material no longer has to be trapped inside a 10-hour tutorial. He is building a two-week workshop instead, with a new module arriving every few days.
I can now reduce hundreds of hours of thinking and typing into just a few hours of prompting.
5:30
In this workshop, you can digestibly consume this content.
5:38
He presents the workshop as a different kind of learning contract. On YouTube, he says, viewers can listen passively while he lands on the next line of code. In the new format, they have to watch his prompts and his thought process, then reuse that pattern in their own codebase. The project will not be identical from one student to the next, and he sees that as a feature rather than a flaw.
You are almost forced to actively observe my thoughts and my prompts.
6:13
I don't want you to clone what I do. I don't want you to copy what I do.
6:36
His final claim is less about AI than about audience expectations. Antonio wants to keep the high-production tutorials, but he also wants smaller, more targeted videos that lower the barrier to entry. He floats examples like an hour of refactoring or an hour of fixing errors, formats that would show more of the real work without demanding a full day from viewers.
I want to teach you how to code like Antonio, not how to code with Antonio.
7:00
I want to start a new direction of this channel where I still produce high-quality tutorials but also give you a lower barrier to entry.
8:05
He frames the new direction as both a product launch and a test of what viewers actually want from him. The workshop comes with a 40% discount, early access to source code, live office hours in Discord, and a cohort experience that disappears once the program starts. Beneath the sales pitch is a more durable question: whether audiences want a finished project, or the discipline to build one badly, then better.
What is agentic coding, according to Antonio?
Agentic coding is a human-led workflow where AI helps speed up work the developer would still do themselves. Antonio says the human remains the architect, writing prompts that reflect the codebase and its constraints.
How does Antonio define vibe coding?
Vibe coding focuses on the outcome more than the technical specification. Antonio says it is often used by nontechnical people who want AI to run autonomously with minimal human involvement.
Why is Antonio changing his tutorial format?
He says the old rebuild format hides the real work of coding, especially now that AI changes how people learn and build. His new workshop is meant to show how he thinks, including dead ends and debugging.
What will the new workshop include?
Antonio says it will run for two weeks, with new modules dropping every few days. It also includes early access, Discord office hours, and live cohort-style access before the program starts.
Will Antonio stop making long tutorials?
No, he says he still wants to make high-quality tutorials. He also wants shorter formats, such as sessions focused only on refactoring or fixing errors, to lower the barrier to entry.
AI-assisted summary of Code With Antonio's podcast, verified against the original transcript.