From: A 40-page brand guide keeps your AI on-voice
To: 15 compressed lines at the top of every prompt kills drift permanently

Theme for the Week: Brand Voice Integrity

So far this week, we've covered why your voice is a compounding asset you can't afford to outsource to generic AI outputs, then built a Voice Vault of 10 gold samples your AI can actually mimic with precision.

✨ In Today's Episode:

Three words. One rule. Zero ambiguity.

That's what a real style directive looks like.

And your AI? It doesn't have one yet.

Which is why every output drifts.

Every prompt costs you a correction pass.

Every "almost there" is a tax you're paying in silence.

I watched a CNBC segment last week where a portfolio manager dropped a line that wouldn't leave my head: "Own it, don't trade it."

Four words. One philosophy. Zero room for misinterpretation.

That single rule kills thousands of micro-decisions before they happen. You don't debate whether to panic-sell on a 4% dip. You don't agonize over the morning headline. The rule did the thinking already.

Now here's the part that hit me as an operator: most brands have a 40-page voice guide that nobody, including the AI, has ever read end-to-end.

And we wonder why GPT keeps writing like a LinkedIn intern on espresso.

The fix isn't more documentation. It's less. It's compression. It's writing the "own it, don't trade it" version of your brand voice - a rule set so tight your AI can execute it on the first pass.

That's what a Do / Don't / Always block is.

Stop treating your brand guide like a novel. Start treating it like a compiled OS - written once, run everywhere.

Here's the template I use, and the one I'm handing you today:

The DO block (5 lines max). What your voice actively reaches for.

The DON'T block (5 lines max). The exact failure modes you keep correcting.

The ALWAYS block (5 lines max). The non-negotiables. The compile-time errors.

That's it. Fifteen lines.

Drop it at the top of every system prompt. Pin it to your Claude Project. Glue it into your CustomGPT instructions. Paste it as the first line of every Cursor `.mdc` rule file.

Then run a test. Same prompt, same input, with and without the block. The drift difference is so loud you'll feel embarrassed you didn't do this six months ago.

Those requests came from non-traded BDC investors in Q1 2026, and most got back roughly half of what they asked for. Moody's U.S. BDC sector outlook: Negative.

On Percent's marketplace that same quarter: new issuances, scheduled payments, 0.44% lifetime net loss rate on asset-based deals since inception.† The difference is structural: concentrated corporate loans with redemption windows that close at manager discretion vs. asset-based finance with 6–24 month deal terms. 14.6% net ABS returns LTM after losses (3/31/26).† Starting at $500.

See how private credit works when it's built for individual investors.

_{Alternative investments are speculative. No assurance can be given that investors will receive a return of their capital. †Past performance is not indicative of future results. Terms apply.}

Here's the part most operators miss.

A rule only holds if it has a trigger baked in. BJ Fogg, the Stanford behavior researcher, has been saying it for years: motivation and ability aren't enough. You need a prompt that fires the behavior.

Your Do / Don't / Always block is the prompt. The system prompt is the trigger. The AI is the executor.

That's a complete behavior loop. And it's the same loop your brand voice needs to survive at scale.

Because here's what's actually happening when your AI drifts: you're trying to enforce voice through motivation (you want it on-brand) and ability (the model can be on-brand). What's missing is the trigger that fires consistently, every prompt, every model, every operator on your team.

A 40-page guide isn't a trigger. It's a library. Libraries don't fire - they sit.

A 15-line block at the top of every system prompt? That fires. Every time. On every model. With every collaborator. Forever.

This is the difference between operators who fight their tools and operators who compound. The compounders write the rule once. The fighters re-explain it every Tuesday.

And the deeper principle underneath all of it: clarity beats volume. Always. A tight rule outperforms a long guide the same way "own it, don't trade it" outperforms a 200-page investment thesis. Compression isn't a shortcut. It's leverage.

Your AI isn't drifting because it's broken. It's drifting because nobody compiled the rules.

Write your fifteen lines today. Paste them everywhere tomorrow. Watch the drift tax disappear by Friday.

That's how brand voice survives the next model release, the next team hire, and the next 10,000 prompts.

Forward to a builder who needs this ➡️

Before you go: Here are 2 ways I can help you scale smarter with AI

AiScaleTips is your founder clarity compass.
Most scale with chaos. You scale by design.

- Justin Glover