A live class with me, every week.
In a group · In English · Deliberately small
A recorded course is you alone against a video library. This is a standing appointment: a new live class every week, hot seats on real cases from the cohort, and whatever I am building right now, before it reaches anything else of mine.
AUD 1,500 for the 12 weeks · 10 seats · Applying is not buying
The problem
You do not need another recorded course. You need rhythm.
How many courses have you bought and abandoned at module 2? It is not a willpower problem. A recorded course depends on something almost nobody has spare: solitary discipline, week after week, with nobody waiting on you.
Then there is the second problem, specific to AI: the field moves every month. A class recorded in January is stale by June. New tool, new model, better way of doing it. Anyone learning from a video library is always learning the past.
The fix for both is the same: a live class, on a recurring schedule, with someone who builds this every day. The appointment gives you rhythm. Live gives you the present instead of the archive.
That is exactly what this is: me in front of you, every week, building alongside you.
How it works
A standing appointment, real cases, and the edge first.
A live class with me, every week
Not a video library. A standing appointment: a new class every week, live, me on screen, with room to ask in the moment. The rhythm is what gets the system off the page.
Hot seats: your case on screen
In the hot seat slot, a real case from the cohort gets taken apart live. It can be yours: your vault, your agent, your bottleneck, solved in front of everyone, with everyone learning from it.
Advanced material first
What I am building and finding reaches this cohort before any other product of mine. You follow the edge, not a summary of it months later.
A small cohort on purpose
Seats are capped so everyone fits in the conversation. A small cohort is what separates real mentorship from a webinar with a chat scrolling past.
Builders next to you
You build alongside other people on the same path. A problem that would cost you days on your own is one somebody in the cohort already solved.
The rhythm that keeps you in
Most people do not stop for lack of content. They stop for lack of cadence. A class in the calendar every week is what keeps your system compounding.
10 seats · AUD 1,500 for the 12 weeks
The 12 weeks
Twelve classes, four blocks, a system running at the end.
Every week has a deliverable, not just a topic. It is the same path I build in the 1:1 mentorship, and the hot seat runs inside the 90 minutes, on top of that week’s theme.
Foundation
Your second brain structured so AI can read it, write to it and keep it organised on its own. The base for everything else.
How AI actually works
A language model is a probability engine, not a reasoning one. Why it gets things wrong, why a small error becomes a large one once you stack tasks, what a token is and why some languages cost more of them. The three levels of AI use, and why almost everyone stalls at the first.
You walk out with: A diagnosis of which level you operate at today, and what is missing to reach level 3.
The AI OS
One mother folder for everything. Obsidian as a viewer, not a vault. Local first for speed and privacy, cloud as backup. Markdown as the language models read best, and wikilinks as the thread that connects what you know.
You walk out with: The mother folder created, Obsidian installed, and AI already reading your own files.
Context beats prompts
The three layers: knowledge, context and maps, execution. Writing your own `me.md`, the document that makes the AI know you. Routing rules so it knows where each thing belongs. The `CLAUDE.md` config file every later model reuses.
You walk out with: Your `me.md` and `CLAUDE.md` written. From here you stop retyping context.
Operation
AI operating directly on your files, projects and workflows, the way a professional uses it. Without writing code.
The terminal
Claude Code for people who do not program. What permissions are and what you grant. Running real work against your own files. This is where the copy and paste friction dies and you feel the difference between using AI and operating it.
You walk out with: Claude Code running inside your vault, with one real task executed by it.
Skills
A markdown file that teaches the model to do one specific thing your way, the same way every time. Not a downloaded pack: we build your first skill from the task you repeat most in a week.
You walk out with: Your first working skill, taken from the task you repeat most.
Connectors
MCP explained without jargon, and AI wired into what you already use: Gmail, Calendar, Drive, GitHub. Plus a collection layer for the data your system does not hold yet and you currently fetch by hand.
You walk out with: At least two connectors live, with AI reading real data of yours.
Agents
Agents with different jobs working in parallel, and work that runs without you watching the screen.
Agent architecture
What each agent owns and what it must never touch, which is the decision that separates a system from a mess. Subagents and work running in parallel, so a three hour task runs as blocks at the same time instead.
You walk out with: Your agent map drawn and the first subagent running in parallel.
Harness engineering
The harness is everything around the model that decides whether it succeeds. The seven components. The maths that should worry you: 99% accuracy per step becoming 37% across a hundred steps. And the four question diagnostic, run live against a cohort member's real system.
You walk out with: Your harness scorecard, with the gaps in priority order.
Hardening
Five brakes and one mouth: verification, sanity checks, a spend ceiling, a human approval point, and orchestration. Each brake is tested by breaking it on purpose before you arm the automation, because a brake you have never watched fail is not a brake.
You walk out with: The brakes installed, each one tested by breaking it on purpose.
Autonomy
The system running without you. Off your machine, with brakes on, and with a weekly ritual that keeps it alive.
Deploy
Getting the system off your personal computer. A server, work that runs on schedule with nobody watching, and a report landing back in a channel you already read every day. The difference between an automation that exists and one that works.
You walk out with: One job running off your machine and reporting back on its own.
Security and compliance
Encryption for what is sensitive, separation between what belongs to the business and what is yours, and what the law asks of anyone handling client data. The class most people teaching AI leave for later, and the one that decides whether you can use this at work.
You walk out with: Sensitive data separated and protected, plus the list of what the law asks of you.
Cadence and demo day
The weekly ritual that keeps the system alive: the review, the dashboard, and the roadmap for the next agent. Plus the close of the cohort, where each person shows what they built across the 12 weeks.
You walk out with: Your weekly ritual running and the roadmap for your next agent written.
English cohort · 10 seats · AUD 1,500 for the 12 weeks
The system in action
The week’s class does not end up in a note. It ends up as this.
Every week adds a piece to your system: an agent, a pipeline, an automation. Scroll to see the kind of thing the cohort builds.
$ claude "update my AI OS with what happened today"
reading vault / 02 Context / active projects...
spawning 4 agents in parallel:
vault-keeperorganising today's notes
research-agentpulling what changed in the market
social-agentdrafting 3 posts
decision-trackerlogging the decisions
system updated. zero manual work.
Applying is not buying · I message you on WhatsApp first
Inside the cohort
The class is once a week. The cohort stands all seven days.
No new platform for you to learn. The conversation happens on WhatsApp, and the memory sits organised week by week, so in two months you can still find what was said today.
A closed cohort group
The ten of you and me, on WhatsApp. A question that comes up on Wednesday does not wait for the next class: it lands in the group and gets answered there, or goes on screen in the following class.
A library, week by week
One folder per week on Drive. The class recording within 24 hours, plus the prompts, the skills, the vault structure and the templates from that class. Nothing lives only in your notes.
A written recap of every class
Every class ships with a written recap and the week's task, generated from the transcript of the class itself. Missed it, joined late, want to review before the next one: it is all written down.
The week's questions become a document
At the end of each week I sweep what was asked in the group into a document inside that week's folder. A good answer does not die in the WhatsApp scroll.
Who is in the cohort, and building what
A list with each person's name, what they do and what they are building. It is what makes the cohort help each other instead of everyone queueing for my inbox.
Yours after it ends
The 12 weeks finish and your cohort folder stays yours, with every recording and every material. The alumni group stays open too.
Group or one on one
Two formats, the same destination: your AI OS running.
Group Mentorship
10 seats- A live class with me every week, in a cohort.
- Hot seats on real cases.
- Advanced material first.
- AUD 1,500 for the 12 weeks.
1:1 Mentorship
Open- Just you and me, live, every week.
- Built entirely on your real case.
- Access to my real code, adapted live for you.
- Starts now, by selection.
Want rhythm, a cohort and a smaller investment? Applications for the English cohort are open. Need individual attention on your own project? The 1:1 is open by selection.
Growth Marketing Manager · AI OS Builder
Who runs the class every week
I am Matheus Vizotto. Brazilian, living in Sydney, Australia, working as Growth Marketing Manager at VenueNow. That is more than 9 years in digital marketing across Brazil and Australia, building growth systems on real budgets with results on the line.
What you get in each week’s class comes straight out of my own operation: the Claude Code, the Obsidian, the agents running in parallel and the automations that run my day. I do not teach theory assembled for a course recording. I open what I am building right now, and the cohort builds alongside it.
The good information on this is written for people who already program. My job is turning it into a system you actually build and use.
The group mentorship is for you if:
- You want to build your AI OS with continuous support, without paying for one on one mentorship.
- You have tried learning from a recorded course and stopped halfway. What was missing was rhythm and someone in front of you every week.
- You like learning from real cases, other people's as much as your own.
- You can hold the live class slot every week.
It is not for you if:
- You want recorded content to watch whenever, with no weekly commitment.
- You need individual attention on your case, every week, just you and me. That is the 1:1 mentorship.
- You want a magic button that builds everything without you understanding any of it.
- You want me building your specific project with you. The cohort path is the same for everyone.
English cohort
Ten seats. I speak to every person before the cohort closes.
It starts once the ten seats are taken. The day and time are chosen with whoever is in it, and all 12 dates go out at once before the first class. A cohort of ten is small enough for that, and it is how I make sure nobody joins in order to miss it.
The AI Builders · Group
Applications open12 weeks, a live class with me every week, a cohort of 10.
Paid up front, for the full 12 weeks
- 12 live classes with me, one a week, 90 minutes each.
- At least one hot seat of your own: your vault, your agent, your bottleneck on the cohort's screen.
- Every class recorded and delivered within 24 hours, and the access stays yours after the cohort ends.
- The material from each class: prompts, skills, vault structure, templates.
- A closed cohort group for questions between classes.
- Advanced material first: what I am building reaches here before any other product of mine.
Applying is not buying. I message you on WhatsApp first.
What the cohort is not
I would rather tell you this before you pay than after. The cohort costs less because the path is the same for everyone, not because it is a 1:1 at a discount.
- A weekly individual session just for you. That is the 1:1 mentorship.
- Your specific project being built with me on screen every week.
- Day to day individual support. Questions go to the group and to the class.
If that is what you need, the path is the 1:1 mentorship. It is open by selection.
Frequently asked
What people usually ask before taking a seat.
How much is it?
AUD 1,500 for the 12 weeks, paid up front. There is no self serve checkout: you apply, we talk, and the payment link follows if it is a fit.
When does it start?
Once the ten seats are taken. This is the English cohort and it is filled by application, so the start depends on it filling rather than on a date I picked. Everyone who applies is kept in the loop and hears the date before it goes public.
What day and time is the class?
Set with the cohort. With ten people spread across timezones I would rather choose the hour that works for the people who are actually in it than impose a slot and watch half the room miss it. When you apply you tick the windows that work in your local time, and once the cohort closes I publish all 12 dates at once, before the first class.
Is this the same as the Portuguese cohort?
Same curriculum, separate cohort. The Portuguese cohort starts in September and runs in Portuguese. This one runs in English, with its own ten people and its own schedule. You are not buying a seat in a room that speaks another language.
How do the classes work?
One live class with me a week, 90 minutes, with room for questions in the moment, plus a hot seat where a real case from the cohort gets taken apart live. It is a standing appointment, not a video library. Everything is recorded and the access stays yours after the cohort ends.
Do I need to know how to code?
No. It is the same path I teach in the 1:1: step by step, built for professionals who do not program. If you can install an application and follow an instruction, you can follow this. It works on Mac and on Windows.
How is this different from the 1:1 mentorship?
In the 1:1 it is just you and me, every week, building on your case. In the cohort you learn the same path alongside nine other people, and the hot seats bring real cases to the screen, including yours. What the cohort does not include is me building your specific project with you every week. If you need that, take the 1:1.
Where does the cohort talk between classes?
In a closed WhatsApp group with the ten of you and me. Questions land there and get answered there, or go on screen in the next class. The memory lives on Drive: one folder per week with the recording, the material and the written recap, and at the end of each week I turn the group's questions into a document inside that folder. There is no new platform for you to learn.
What if I miss a class?
Every class is recorded and lands in the cohort folder within 24 hours with that week's material. But be honest with yourself: what makes this work is the live meeting. If you cannot hold the slot every week, the recordings alone will not carry you to the end.
Is applying the same as buying?
No. You apply, I message you on WhatsApp within 24 hours to see whether it is a fit, and only then does the payment link go out. There are 10 seats and I speak to everyone before closing the cohort.
Everyweek,aliveclasswithme.Everymonth,yoursystemstronger.
Ten seats, and I speak to every person before confirming. The cohort starts once those ten are taken, and the day is set with whoever is in it.
Applying is not buying. I message you on WhatsApp first.