The Meatbag Method · A free video series from Meatbag Made
A ladder for becoming a participant in what is happening with AI, instead of a spectator watching other people argue about it. Plain language, made for people who are not technical.
Leave with a skill, not a subscription
01Where we are
The railroads weren't transformative because trains were interesting.
They were transformative because of what the tracks connected.
A town that had been three days from anywhere was suddenly a morning away. Mail arrived. Produce arrived that nobody local had ever seen. New customers appeared. New jobs appeared. Businesses that couldn't have existed before became possible.
The locomotive mattered. But it wasn't the point.
A lot of the conversation about AI right now is trainspotting.
Which model is fastest? Which company is ahead? Who won the benchmark this week? How many parameters? How many tokens? Which model is smarter?
That's a perfectly good hobby. And, frankly, a pretty entertaining spectator sport.
But the more interesting question is what becomes possible now that the track runs past your house.
What can you do that wasn't economically practical before? What can a small team accomplish? What work can disappear? What new work can exist? What can one person build, learn, create, or operate that previously required an entire organization?
The model is the train.
The infrastructure is the story.
And if the honest answer is that you'd rather ride the train, look out the window, and enjoy the fresh produce it brings to town, that's completely okay.
Nobody has to become an engineer.
But it is worth getting on board.
Because the interesting question isn't how good the train is.
It's where the track takes you.
02The ladder
Nobody has to climb every rung. Most people should climb the first two, and the view from each one is worth the trip on its own.
Every rung starts the same way, with not knowing. That is not a problem to get past. It is where all of this begins. And at every rung, at every step, your hand stays on the output. You are not feeding something into a black box and hoping. You are the one deciding what counts as good.
You use these tools for real work instead of party tricks.
You have heard about this for two years. You typed something in once, got back something vaguely wrong, and quietly decided it was not for you.
Then you give it something real from your actual week, with enough context to work with, and what comes back is genuinely useful.
It is a tool now instead of a rumor. Which raises a fair question: is the one you happened to try the right one for you?
You decide what is true before you send it.
You stop asking which tool is best and start asking which is best for you.
Somewhere there is a leaderboard claiming one of these is the smartest. It is not measuring your work, and you have no idea which one to trust.
You run the same real task through three of them and the differences show up immediately. One needs half the cleanup of the others.
You have a tool you trust for your kind of work. Now you start noticing where your time actually goes.
You set the standard for good, not a leaderboard.
You stop automating the mess and start redesigning the work.
There is something you do every week that takes two hours and should not. You have never once written down what it actually consists of.
You write it out step by step and there it is: four of those steps are just moving information around, and only two of them need you.
Two hours became twenty minutes, and the judgment stayed yours. Now you are wondering what happens if the steps connect to each other.
The judgment steps stay yours on purpose, not by accident.
You connect it to your actual stuff and let it run.
Doing it by hand every week still means doing it every week. You suspect there is a version where it is simply done, but that sounds like programming.
You connect one step to the next, put a checkpoint where a human should look, and watch the whole thing run without you.
You built something that works while you sleep. Which makes you curious about what is actually inside the box.
You build the approval gate before you build the automation.
You run one on your own machine, just to see what that feels like.
Everything so far arrived over the internet from a company. It is easy to assume that is the only way any of this can work.
You download a model onto your own laptop, run it, turn off the wifi, and it keeps answering.
The point was never that the local one is better. It is the difference between AI as a service someone sells you and AI as something you can hold.
Nothing leaves your house. That is the whole point.
03The method
The same four moves work on every rung of the ladder. They do not change as you climb. They just get deeper.
Choose one thing you want, specific enough to have a name, instead of deciding to learn AI in general.
Write the instruction in five parts: role, context, task, format, constraints. The same five you would use to brief a new hire.
Cut what is wrong, keep what is right, run it again. This is the step people skip, and skipping it is how slop gets made.
Give it a home and a schedule, so it still happens during a busy week and not only when you feel like it.
04What gets scarce
These five get more valuable, not less. Each one also comes with a specific way to get cut, so both are listed.
Seeing work as a system instead of a job description. Trigger, gather, interpret, decide, draft, review, send. Once you can see the shape of it, the opportunities are obvious.
Producing an answer is cheap now. Knowing whether the answer is any good is the expensive part, and it got more valuable, not less.
If everyone can generate images, copy, video, decks, and software, then making things stops being scarce. Knowing what is worth making becomes the whole game.
Directing people, tools, and machines toward an outcome, rather than personally performing every task in the chain.
Permission to say I wonder if this could, and then find out. You do not need a strategy first. You need one afternoon and a real question.
Your context. Your standards. Your relationships. The taste you built over twenty years of paying attention to what is good and what is merely fine. None of that is in anybody's training data, which is exactly why it is the part worth protecting.
You can automate a great many things. You cannot automate taste.
05What it is for
Most conversations about AI and time are subtraction. Fewer hours, fewer people, less cost. That is a small idea, and it is the least interesting thing that could possibly happen here.
Here is the honest part. When you actually get the time back, it is genuinely hard to say what to do with it. Most people cannot answer that on the spot, and that is not a failure of imagination. It is the real question, and it is a much better one than how do we do the same thing more cheaply.
Some of the answer is small and personal. An evening back. A thing you make because you can now, that would have been too much trouble before. Something you were curious about at nineteen that you finally have room for.
Some of it is not small at all. Every big jump in what people could make, and could afford to try, ended with more human work rather than less, and most of that work was the kind nobody had thought of yet. Medicine, science, art, and the strange useful things that get built when experimenting stops being expensive. Renaissances tend to happen when making things gets cheap and curiosity gets permission.
You do not have to know which one you are doing. You only have to avoid spending the whole thing on doing your old job slightly cheaper.
06The challenge
Not a course. One loop, run on purpose, this week.
One thing. A simple website. Your family tree. A presentation about the lions of the Serengeti. It matters far less than you think what it is.
Use whichever AI tool you already like. Give it real context instead of a wish. Who it is for, what it is for, and what good would look like.
Look at what came back. Test it. Tell it exactly what is wrong. Run it again. Do this at least three times, because the third pass is where it stops being generic.
Throw the whole thing away and build it again from scratch, now that you know what a good first prompt looks like. This is the part everyone skips, and it is where the skill actually lives.
Do it again next week on something else. The thing you built is practice equipment. The loop is the method.
Prompt it, review what comes back, give it real feedback, run it again, and keep your hand on the result the entire time.
That is the whole method.
07The episodes
A working game built start to finish with these tools, with the wrong turns left in. It shows what the loop produces rather than explaining it.
Watch on YouTube →What these tools are, what they are good at, what to never hand them, and how to keep a person in the loop. Start here if you have never opened a chatbot.
A full build, using fantasy football. In about an hour you will have a weekly digest for your own league that you own outright.
Season 2, Go Make Your Own Tools. Meal planners, job search trackers, the family paperwork nobody wants to touch, and your own learning plan.
Season 3, Level Up the Muscle. Chaining prompts together, building a first agent, and sizing up a new AI product in ten minutes.
There are no release dates, because these go up when they are finished. Season 1 is the commitment.
08Who is making this
I train corporate teams in AI enablement for a living, which means I spend my working hours watching people meet these tools for the first time and learning exactly where they get stuck.
I build my own things with them on nights and weekends, which is how I know the difference between a demo and something that still works on a Tuesday when everyone is busy. This series is the same material I teach at work, given away.
I take on work outside the day job: coaching a family or a small team through their first real use of these tools, building the one weekly chore that is eating your business, advising on where AI fits and where it does not, and writing and content work under the Meatbag Made mark. Send an email describing what is going on, and I will answer it myself.