Start here · Lesson 1.2
10 Brand Brain AI Strategies to Pick From
A brand brain AI is only as useful as the habits around it. The folder itself does nothing until you use it well. These are ten tactics for doing that, written as things to try rather than promises. None of them comes with a guaranteed result, so treat each as an experiment and keep what helps.
If you have not built the brain yet, start with Lesson 1.2, Build your brand's brain, which covers the setup. This post assumes you have a folder and want to get more out of it.
Do not adopt all ten. Pick two that match your biggest annoyance, run them for a week, and keep the ones you still use.
Strategies 1 to 3: make the brain accurate
1. Put a dated goal at the top. One line, with a date, such as "Goal this month: views. Next: followers." The assistant then has something to choose between, and you can see when it is stale.
2. Add one example of writing you like. Paste a real post you are proud of. A sample teaches voice faster than a description of voice does.
3. Keep a "never say" list. Phrases you hate, claims you cannot back, topics that are off limits. It is short and it prevents repeat mistakes.
Strategies 4 to 6: feed it good material
4. Sort media into three groups. Ready, needs an edit, still to shoot. The third group becomes your shoot list.
5. Save customer questions. Real questions people ask you are better source material than anything invented. Drop them into a file called questions and let the assistant draft content that answers them.
6. Log what worked, plainly. After a post does well or badly, write one sentence in a results file. Avoid guessing why. Record what happened, and let patterns appear over weeks.
Strategies 7 to 8: build the memory habit
7. End every chat with the memory line. Ask the assistant to save what it learned and to tell you what it changed. Anthropic's documentation describes auto memory in Claude Code, where Claude saves notes itself, and CLAUDE.md files, which you write yourself. Using both gives you notes you control and notes it picks up.
8. Read the changes. The memory line only helps if you check it. A wrong note gets repeated in every later chat, so correct it the first time you see it.
Strategies 9 to 10: use it on real work
9. Start from a brief, not a blank prompt. Give the assistant the job, the reader and the standard. The ideas in How to Brief AI Like a Senior Hire fit here, and the brain covers the parts that never change.
10. Review output against the brain. After a draft, ask: does this match our voice, our audience and our goal? If not, the fix is usually in the brain file, not the single draft.
| If your problem is | Start with |
|---|---|
| Output sounds generic | 2 and 3 |
| You never know what to post | 4 and 5 |
| You re-explain yourself every chat | 7 and 8 |
| You cannot tell what works | 6 |
| First drafts need heavy edits | 9 and 10 |
| The goal keeps shifting | 1 |
How to run a one week trial
Choose two strategies on Monday. Write down, in a sentence, what annoys you now: for example "every draft needs a full rewrite." On Friday, look at the same thing again. Has it improved, stayed the same or got worse? That is a fair, honest test, and it needs no tools beyond a notes file.
Resist the urge to count minutes saved unless you actually timed the work. Rough feelings are fine, but do not turn them into numbers you did not measure.
Which strategies depend on the tool
Some of these work in any assistant that lets you paste text. Others, like folder access and auto memory, depend on the tool. Anthropic's docs say Claude Code can access files in your directory and subdirectories, and that it loads notes at the start of each session. If your tool cannot do that, paste the brain file at the start of each chat instead. It is less smooth, but the habit is the same.
What to do when a strategy does not help
Not every tactic fits every brand. If you try one for a week and see no change in the thing that annoyed you, drop it without guilt. A tactic that helps a bakery posting daily may be pointless for a consultant who writes once a fortnight. Also check whether the problem is the strategy or the execution. A never say list that has three items will do little, while one built from real mistakes you have seen in drafts will do a lot. Keep a short note of what you tried, the date and what you noticed. After a couple of months that note becomes your own playbook, built from your brand instead of borrowed advice. It is also a good way to avoid trying the same thing twice because you forgot the result. When two strategies seem to conflict, such as keeping the brain short and adding more examples, favour short and put the best example at the top.
Where to go from here
For a fuller setup, follow Lesson 1.2, Build your brand's brain. If you want a structure to start from, the free template gives you a page to fill in. And if you wonder whether the effort is worth it, compare it with doing the same job by hand.
Questions people ask
Which strategy should I try first?
Number 1 or 7. A dated goal and the memory line cost almost nothing and shape everything after.
Will these make my posts perform better?
Nothing here promises that. They make your input clearer, which usually makes drafts closer to usable. Results depend on your product, audience and posting.
How many files should the brain have?
Few. One main file and a folder of sorted material is enough to begin.