Module 2 · section 0 · 599 words · ~4 min read
The AI Production Stack
The module the rest of the guide is trying to earn.
Everything before this was setup. Module 0A gave you the trade. Module 0B gave you a starting point you can defend. Module 1 made the argument. Module 1.5 made sure nobody was pretending to know what a chat window was.
This is the module where you make something.
1What this module is actually about
There is a version of this module that is a list of tools, and it would be out of date by the time you read it. That version is worth very little, and it is not what you paid for.
What holds still is the shape of the work. A finished piece of UGC is four jobs in a trench coat: someone decides what it says, someone performs it, someone cuts it together, and someone checks it before it goes out. Those four jobs existed before any of this, and they will exist after. What has changed is how many of them you can do yourself in an afternoon.
So this module is organised by job, not by tool:
| Section | The job | What you'll be able to do |
|---|---|---|
| 1 | Deciding what it says | Write a brief a model can follow, and get hooks back that don't sound generated |
| 2 | Choosing your method | Tell the difference between work that needs a conversation and work you can hand off |
| 3 | Performing it | Evaluate the voice and avatar category without being sold to |
| 4 | Cutting it | Know which editing passes to hand over and which to keep |
Section 2 comes second on purpose. Am I holding the loop, or handing it over? is the question you'll be asking of everything in Sections 3 and 4, so it's better answered before you meet the tools than after.
2What you will need
- The brand voice document from Module 3, if you have one yet. You don't need it to read this module, but every prompt in the template library has a slot for it, and they get noticeably better once it's filled. If you're reading in order, come back and re-run them after Module 3.
- Something real to make. Pick one product from the daily brand list you built in Module 0B — ideally one you marked would replace, because you already know what's wrong with it. Everything in this module works better against a real thing than a hypothetical one.
- Twenty minutes and a willingness to throw away nine out of ten outputs. That ratio is not a failure state. It's the job.
3What this module will not do
It will not make you fast at this in an afternoon. The tools are fast; the judgement is not, and the judgement is the part clients pay for.
It will also not tell you which product to buy. Section 3 gives you the questions to ask instead, because the answers change every few months and a guide that names a winner is a guide that is wrong by Christmas. The living tool-landscape document is a separate, versioned reference for exactly this reason — see the note at the end of Section 3.
4One line to take out of this module
The tools are good at the passes that are mechanical and reversible. They are bad at the ones that are taste. Learn which is which and you have a career; confuse them and you have a lot of content nobody watches.
Source: content/module-2/00-module-overview.md