You don't have a tool shortage. You have a decision surplus. Every subscription you add creates one more choice to make before you can start working — which tool, in what order, with which export settings, into which folder.
The decisions are the expensive part. Making the thing is comparatively easy.
So the stack below is organised by layer, not by product. Five layers. One tool each. If a layer has two tools in it, the second one is either a hobby or a mistake.
The five layers
| Layer | What it does | Skip it and… |
|---|---|---|
| Capture | Where ideas land the second they occur | Good ideas die in the shower and DMs |
| Make | Image, video and text generation | You're renting someone else's taste |
| Assemble | Where raw material becomes a finished asset | Everything looks like a draft |
| Distribute | Getting it out, on a rhythm | You post in bursts and vanish for ten days |
| Remember | Briefs, prompts, references, results | Every week restarts from zero |
1. Capture — one inbox, or none
Not notes app plus voice memos plus screenshots plus saved posts. One place, searchable, that you can open in two seconds.
The test: when an idea arrives, can you record it in under ten seconds without deciding anything? If not, it isn't a capture layer — it's an archive you're frightened of.
2. Make — one model you know well
Knowing one image model deeply beats sampling five. The knowledge that transfers is not button locations — it's knowing how the model misreads a brief, and which words you use to correct it.
You cannot build that model of a model if you switch every fortnight.
3. Assemble — optimise for speed, not features
This is the layer you spend the most hours in, so it should be the least interesting one. Templates, keyboard shortcuts, a known export setting.
If your editor has features you've never opened, that's not capability — it's a menu you're paying to ignore.
4. Distribute — manual first, always
Post by hand twenty times before you schedule anything. You learn the platform's rhythm, what the first line looks like in a feed, and where the crop cuts off. A scheduler hides all of that from you, and then you're optimising blind.
5. Remember — the layer almost everyone skips
This is where leverage actually lives, and it's the least exciting, which is why it's empty for most creators.
Four things belong in it:
- Your brief. The six fields: audience, idea, look, format, no-list, success test.
- Your prompts. The ones that worked, with the variables pulled out, kept in one place instead of scattered across chats.
- Your references. Ten images that define the look, named clearly enough that you can point at one and say "this".
- Your results log. One line per post: what it was, how it did, and your best guess why. Sixty of those lines will teach you more than any course.
Without this layer, every week is a cold start. With it, week thirty is faster than week one — which is the only compounding a solo creator actually gets.
A new tool must replace one, and you must be able to name the specific friction it removes in one sentence. "It's meant to be good" is not a friction. "Every export takes four clicks and I do it eleven times a week" is.
Tool stacking is a socially acceptable form of procrastination. Nobody's work got better because they changed editors. If you're reorganising your stack more than once a quarter, the stack isn't the problem.
What this looks like in practice
Five tools, five layers, one file each. A brief you can open, a model you know, an editor with two saved templates, a posting rhythm, and a folder of prompts and references.
That's it. Everything else is optional, and most of it is noise wearing a subscription.
If you want the prompts layer to start from, the prompt index is organised the same way — by job, not by product.