The 5-section Persona Brief template that captures your voice
The exact 5-section Persona Brief template — who you are, voice DNA, banned words, required structures, reference posts — with line-by-line instructions and copy-paste examples.
The direct answer
The 5-section Persona Brief template is: (1) Who you are — 3 sentences covering role, audience, and the one thing you are known for; (2) Voice DNA — 5-8 traits with one-line descriptions; (3) Banned words and phrases — the highest-leverage section; (4) Required structures — hook patterns, CTA patterns, sign-off rules; (5) Reference posts — 3-5 of your best posts pasted verbatim. The brief loads into context for every AI generation.
Most teams write voice documents that read like brand-strategy decks. They are not actionable for AI. The Persona Brief is engineered for AI generation: every section produces a specific behavior in the output.
Here is the exact template, section by section, with what works and what does not.
Section 1: Who you are (3 sentences max)
Three sentences. Role, audience, and the one thing you are known for. Specificity is the entire game here.
What works:
Founder of Kompozy, the autonomous content composition platform. I help creators, founders, and agencies turn one source into a month of content. I am known for the Persona Brief methodology and treating AI content as an engineering problem, not a writing problem.
What does not work:
I am a passionate entrepreneur who loves to help businesses grow through innovative content marketing solutions.
The second example is generic. It produces generic output. The first is specific enough that the model can infer tone, vocabulary, and even authority markers.
Section 2: Voice DNA (5-8 traits)
Each trait is one line. Each line should produce a specific behavior in output. Examples that work:
Direct, no hedge words. ("It is" not "It may be." "Most" not "many.")
Short sentences. Max 18 words.
Specific numbers always. Cite hours, dollars, percentages — never "a lot" or "many."
Anti-thought-leader. No motivational filler. No "in today's fast-paced world."
Writes like Naval Ravikant if Naval ran a B2B SaaS.
Contrarian openers — lead with a claim that surprises.
Story-led but front-loaded — the punchline goes in sentence 1, not sentence 5.
Reference creators by name. Base models know them — "writes like [name]" is the single highest-signal voice instruction you can give.
Section 3: Banned words and phrases (highest-leverage section)
Every banned phrase removes one AI fingerprint. Starter list — phrases that flag content as AI within 3 seconds:
Hedge words
arguably
it is worth noting that
in many cases
tends to
may often
can sometimes
generally speaking
broadly speaking
one could argue
AI bridge phrases
not just X but Y
this is more than just
what makes this special
in addition
furthermore
moreover
on top of that
Closing summaries
in conclusion
ultimately
to summarize
at the end of the day
wrap up
final thoughts
to sum up
Marketing speak
in today's fast-paced world
leverage
unlock
dive deep
game-changer
revolutionize
seamless
cutting-edge
best-in-class
world-class
state-of-the-art
next-level
mission-critical
paradigm shift
Vague authority
studies show
research suggests
experts agree
data confirms
it is well known that
it is widely accepted that
Add anything you would not say in conversation. The list grows from every edit you make to AI output.
Contrarian claim: "Most founders are wrong about X."
Specific number first: "I tracked 47 launches. Here is what they had in common."
Confession: "I screwed this up for 6 months before I figured it out."
Question with specific frame: "Why does every Y do Z when X works better?"
CTA patterns that work
Implicit CTA via story conclusion: end with the outcome that resulted, link to the deeper guide.
Direct CTA: "If you want the full methodology, [link]."
Question CTA: "What is your version of this story?" (drives replies)
Sign-off rules
Do not include a name signature in social posts (the author is implied by the account)
For newsletters: sign with first name only, no title
For long-form: no sign-off; the conclusion is the sign-off
Set 2-3 hard rules per pattern. More than that and the output gets rigid.
Section 5: Reference posts (3-5 of your best)
This is where the pattern-matching happens. The model reads these posts as examples and tries to produce output in the same shape.
How to pick reference posts:
Posts that performed well AND felt most like you. The intersection matters more than either alone.
Cover multiple formats — at least one short post, one long post, one with a story, one with a framework.
Diversity over consistency. The model averages across all references, so you want range.
Paste them verbatim, including line breaks and formatting quirks.
Update reference posts every 3-6 months as your best work shifts. Stale references produce stale output.
What NOT to put in the brief
Contradictory pairs: "Professional but friendly." These cancel out.
Over-specified formatting rules: "Always 3 bullet points, always end with a question." Output gets rigid.
Vague aspirations: "engaging," "thought-provoking," "premium." No operational meaning.
Long personality essays: longer than 3 sentences dilutes signal.
Cliches about your industry: "We move fast and break things." Empty calories.
How to iterate the brief
Every edit you make to AI output is an iteration signal. Process:
Generate an output.
Edit aggressively.
Ask: why did the AI miss this?
Update the appropriate section of the brief.
Generate the next output. Repeat.
After 15-20 iterations, the brief stabilizes. You will start approving most outputs untouched. That is the threshold where autopilot becomes safe.
Frequently asked questions
How long should the Persona Brief be?
Total length: 500-1,200 words. Shorter than 500 leaves too much room for LLM defaults. Longer than 1,200 wastes context tokens without adding signal. The 5-section structure naturally lands in this range.
Can I copy a Persona Brief from another founder?
You can use it as a template, but the content must be yours. Copying voice DNA verbatim produces output that sounds like the other founder. Use the structure, write your own content.
How often should I update the Persona Brief?
Update sections 1-2 only when your role or audience changes (rare). Update sections 3-5 continuously based on edits. Aim for 3-5 updates per week during the 14-day ramp, then 1-2 per month after.
What happens if my voice evolves over time?
Update the brief. Reference posts especially should reflect your most recent best work, not your work from 2 years ago. Stale references produce stale-sounding output.
Should I have one brief or multiple briefs per platform?
One base brief covering your universal voice. Platform-specific overrides handle tone differences (X terse, LinkedIn long-form, TikTok conversational). The base brief is 80% of the work; overrides are 20%.
Autonomous Content Creation — Most "autonomous" AI content is slop. Here is how 4 quality gates make autopilot output indistinguishable from manually-approved content — and the exact 14-day ramp to flip the switch safely.
AI Content Repurposing — The complete methodology for turning one source into 25-35 pieces of native-format content across every platform — without producing AI slop.