Ten questions about who you are, who you serve, and how you sound. Answer once. Save as one file. Paste anywhere AI reads context. Watch the output stop sounding like everyone else's.
Not your job title. The actual thing you spend your week on.
What you make or sell, for whom, with what outcome. If it takes two sentences, the first is probably noise.
What keeps your customer up at 2am. What they are tired of explaining to other vendors. Their words, not yours.
Two or three things a competitor cannot copy by hiring better. "We care" is not a real answer.
Picture a single human with a name and a problem. Generic = generic AI output.
Industry, role, company size, geography, life stage. The more specific, the more your AI can pattern-match.
The thing they are quietly working on this quarter, that nobody is selling to.
Podcasts, newsletters, Slack communities, the one peer they text. AI needs to know who you sound like to them.
Three traits. The opposite of "professional and approachable."
Direct, warm, technical, blunt, irreverent, calm, urgent, playful, no-nonsense. Pick three that fight a little.
The "never" list is more useful than the "always" list. Words you would not be caught dead writing.
A LinkedIn post, an email, a newsletter. The AI will reverse-engineer your rhythm.
The corporate version of you. So the AI knows which way NOT to drift.
Three steps to make these answers actually change your AI output.
Name it Core/brand-brain.md inside your VaultAgents vault. Plain text, no formatting drama.
Add one line under "Core context" pointing at the file. Every AI session will now auto-load it before responding.
The next time you ask Claude to draft anything customer-facing, it will sound like you, not like a 2023 GPT. If it does not, the file needs more specifics. Iterate.