The 5-section Persona Brief template that captures your voice

Published 2026-05-18 · Updated 2026-10-23

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: What goes in "Who you are"?

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: What are your 5-8 voice DNA 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: Which words and phrases should you ban?

Every banned phrase removes one AI fingerprint. Starter list — phrases that mark content as AI on sight:

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.

Section 4: Which structures should every post follow?

Hook patterns, CTA patterns, sign-off rules. Concrete examples beat abstract rules.

Hook patterns that work

• 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

Keep the hard rules per pattern few. Pile on too many and the output gets rigid.

Section 5: Which 3-5 reference posts should you include?

Include 3-5 of your best posts. The model reads them as examples and tries to produce output in the same shape, so this is where the pattern-matching happens.

How to pick reference posts:

1. Posts that performed well AND felt most like you. The intersection matters more than either alone.
2. Cover multiple formats — at least one short post, one long post, one with a story, one with a framework.
3. Diversity over consistency. The model averages across all references, so you want range.
4. Paste them verbatim, including line breaks and formatting quirks.

Swap in new reference posts whenever your best work shifts. Stale references produce stale output.

What should you leave out of 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 do you iterate the brief?

Treat every edit you make to AI output as an iteration signal and update the matching section of the brief. The process:

1. Generate an output.
2. Edit aggressively.
3. Ask: why did the AI miss this?
4. Update the appropriate section of the brief.
5. Generate the next output. Repeat.

After enough rounds the brief settles, and you start approving most outputs untouched. That is the threshold where autopilot becomes safe.

What does this look like in Kompozy?

A Persona Tweet generated in the Kompozy workspace. Persona formats like this are written with the Persona Brief loaded in context.

Who is this for, and who should skip it?

Best for:

• Anyone setting up AI content generation for the first time
• Teams whose voice document reads like a strategy deck and does nothing for AI output
• Creators who keep rewriting AI drafts the same way

Not ideal for:

• People who will not maintain the brief; it needs updates as your edits pile up
• Brands that want several distinct voices in one brief; use one brief per voice
• Teams that copy another founder's brief verbatim and expect their own voice back

Frequently asked questions

How do we know this?

This page documents Kompozy's own Persona Brief template, which we checked against kompozy.io/brand-voice and kompozy.io/autonomous on 2026-10-23. We removed unsourced figures on brief length, iteration counts and timing.

Sources

1. Kompozy — Brand voice and the Persona Brief (read 2026-10-23)
2. Kompozy — Autonomous content and the 14-day ramp (read 2026-10-23)

Which other brand-voice guides should you read next?

• The complete AI banned-word library (kill every AI tell) — 120+ phrases that flag content as AI-written: hedge words, tricolons, "not just X but Y," vague authority claims. With drop-in copy-paste lists.
• Voice DNA: defining the 5-8 traits that make your writing recognizable — How to extract voice traits from your existing content and codify them so AI can replicate them on demand.
• How to use reference posts to fine-tune AI voice (without actually fine-tuning) — Why 3-5 well-chosen reference posts beat any amount of voice description in a prompt.
• AI voice for X / Twitter: terse, punchy, contrarian — X-specific voice rules — short sentences, no hedge words, contrarian framing. With thread-length and posting-cadence guidance.
• AI brand voice for SaaS founders and product marketers — B2B SaaS Persona Brief examples, including how to balance authority with personality and avoid the "thought-leader AI" voice.

Which related Kompozy topics cover this next?

• Autonomous Content Creation — How Kompozy autopilot really works: one workspace switch that approves and schedules generated posts, the checks that run before they publish, and a 14-day ramp for turning it on.
• 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.

← Back to AI Brand Voice & Persona overview · Get started →