META24 WORKFORCE LAB · GRADED TO THE CORE · PROJECT-BASED · A COURSE IN AI & MACHINE LEARNING · MIDDLE SCHOOL · UPPER SCHOOL · GRADE 6–12 · 0.5 CREDIT · META24 LAB COMPETENCIES · THE AUTOMATION BENCH
AI & Machine Learning · Applied Automation
Understand it · Build it · Put it to work · in Understand it. Build it. Put it to work. · AI & Automation
13 weeks of Understand it. Build it. Put it to work. · AI & Automation: It Is Not Thinking. It Is Predicting. · The Loop · Four Companies, One Trick, Different Bets · Where It Belongs in a Business — and Where It Does Not · Your First Call …
The weeks are the course’s own, read from the lab; the catalog pads the last weeks with the capstone and the floor so the course fills a whole term.
THE MIX · HOW MUCH OF EACH CORE SUBJECT THIS COURSE CARRIES · IN ORDER
CODING & AIREADING & WRITINGBUSINESS & ENTERPRISEMATHTECHNOLOGY & TRADESHISTORY, GEOGRAPHY & HUMANITIESSCIENCEART & DESIGN
The squares are a picture of a course — how much of each core subject its own rooms carry, read from the rooms’ text — never a score of a student. Every course in a lab is graded to the core across the board; the squares say which columns it feeds most.
WHAT YOU WILL BE ABLE TO DO
- A language model is a very large function that predicts the next piece of text — and everything it can do, and every way it fails, follows from that one fact.
- An agent is a model in a loop with tools — it observes, thinks, acts, observes again, and stops when the job is done or a person must decide
- The major models are the same mechanism trained with different data, different priorities and different business models — and the right one is decided by the job, the volume and the risk, never by the brand.
- AI earns its place in a business at the steps that are repetitive, information-heavy and low-stakes-per-instance — and it is kept out of the steps where a person must be accountable.
AT A GLANCE · 18 WEEKS
Q1 It Is Not Thinking. It Is Predicting. → One Plug for Every Tool — The assignments · weeks 1–9
Q2 Automate One Thing in Your Own Lab → The floor — The assignments · weeks 10–18
ALSO IN AI & MACHINE LEARNING · UNDERSTAND IT. BUILD IT. PUT IT TO WORK. · AI & AUTOMATION
A semester
18 WEEKS · TWO QUARTERSQuarter 1 · It Is Not Thinking. It Is Predicting. → One Plug for Every Tool
WEEKS 1–9 · IN THIS OFFERINGTHE WEEKS · THE ROOMS EACH ONE OPENS
WK 1It Is Not Thinking. It Is Predicting.UNDERSTAND IT · BUILD IT · PUT IT TO WORK
A language model is a very large function that predicts the next piece of text — and everything it can do, and every way it fails, follows from that one fact.
WK 2The Loop
An agent is a model in a loop with tools — it observes, thinks, acts, observes again, and stops when the job is done or a person must decide. METTIE is not one yet. You will see exactly why.
WK 3Four Companies, One Trick, Different Bets
The major models are the same mechanism trained with different data, different priorities and different business models — and the right one is decided by the job, the volume and the risk, never by the brand.
WK 4Where It Belongs in a Business — and Where It Does Not
AI earns its place in a business at the steps that are repetitive, information-heavy and low-stakes-per-instance — and it is kept out of the steps where a person must be accountable.
WK 5Your First Call
Every AI product in the world is, underneath, a program sending a request to a model over the internet and reading the reply.
WK 6A Prompt Is a Function. Test It Like One.
A prompt that runs ten thousand times is code — it has inputs, outputs, edge cases and bugs — and the professionals who build with AI write it, version it and test it like code, with structured output and an evaluation…
WK 7Give It Hands
A tool is a function you describe to the model; the model asks to call it, your code runs it, and the result becomes the model's next observation.
WK 8Write the Loop Yourself
An agent is a while loop around the tool call from week 7, with a stop condition, a turn limit, a budget, and a gate.
WK 9One Plug for Every Tool
MCP is the standard that lets any tool be plugged into any agent — write a server once and Claude, ChatGPT, Gemini and your own loop can all use it.
THE PROJECT · WHAT THE QUARTER IS FOR
The assignments · weeks 1–9
The question. What does Understand it ask you to be able to do by week 9?
The make. Every week's assignment and the quiz as practice: It Is Not Thinking. It Is Predicting. · The Loop · Four Companies, One Trick, Different Bets · Where It Belongs in a Business — and Where It Does Not …
The evidence. Handed in through Understand it inside Understand it. Build it. Put it to work. · AI & Automation, kept in My Work; the Report reads it from there.
Quarter 2 · Automate One Thing in Your Own Lab → The floor
WEEKS 10–18 · IN THIS OFFERINGTHE WEEKS · THE ROOMS EACH ONE OPENS
WK 10Automate One Thing in Your Own LabUNDERSTAND IT · BUILD IT · PUT IT TO WORK
Take the workflow you mapped in week 4, the tools from week 7, the loop from week 8 and the server from week 9, and ship one automation that runs against your lab's sample grid — with an eval, a log, and a gate a…
WK 11What an Agent Must Never Do
An agent is trusted in proportion to what it cannot do. Permission levels, an audit log, and defence against instructions hidden in its inputs are not features added to a finished agent — they are what makes it…
WK 12The Same Loop, Inside This Site
Everything you wrote in Python has an exact twin in TypeScript — and this site is written in TypeScript. This week you read the real METTIE route, write the same agent loop in the site's language, and see how close the…
WK 13Ship It to a Lab That Is Not Yours
The proof that you can build automation is not that it works in your lab — it is that it works in someone else's, from the runbook, with their data, and that they would keep it.
WK 14–18The floor
The floor: the work is done in the room, logged in the room, kept in My Work.
THE PROJECT · WHAT THE QUARTER IS FOR
The assignments · weeks 10–18
The question. What does Understand it ask you to be able to do by week 13?
The make. Every week's assignment and the quiz as practice: Automate One Thing in Your Own Lab · What an Agent Must Never Do · The Same Loop, Inside This Site · Ship It to a Lab That Is Not Yours …
The evidence. Handed in through Understand it inside Understand it. Build it. Put it to work. · AI & Automation, kept in My Work; the Report reads it from there.
A course is a quarter, a semester or a year — nine, eighteen or thirty-six weeks — a project a quarter. Offered as the year, a semester or a quarter, it is the same course seen through a window. A lab is a family: its year, each of its tracks as a course, and its rooms. The evidence is what a student hands in through the room; the Report reads it from there. Nothing in the catalog is a grade.
THE PASS · THIS COURSE AND EVERY OTHER ONE
One pass includes AI & Machine Learning · Applied Automation and every course in the catalog. Pick a pass, or enter a school code; checkout stays on this page.