Persona prompting
Role & persona
Level 1 · FoundationsGive the model a fully fleshed-out character — background, personality and voice — not just a job title.
Persona prompting goes further than role prompting: instead of naming a professional role, you describe a whole character — their background, personality traits, values, and way of speaking. The model then stays 'in character' throughout the response, which produces a distinctive, consistent voice rather than just domain-appropriate content.
When to use it
Use it for creative writing, chatbots, brand voice work, or any content where a recognizable, consistent personality matters more than raw domain expertise.
When to avoid it
Avoid it for factual, technical or high-stakes tasks (medical, legal, financial) where a strong personality could distort accuracy or professionalism, or where consistency with a house style matters more than character.
Advantages
- Produces a distinctive, memorable voice
- Keeps tone and word choice consistent across a long response
- Useful for brand voice, characters, and conversational products
Disadvantages
- Can drift into caricature if the persona is too extreme
- Adds tokens describing the persona before the actual task starts
- Personality can occasionally leak into tasks that need a neutral tone
Examples
You are Mara. Background: 15 years as a senior engineer, mostly on backend systems and code review. Goals: help people write clearer, more maintainable pull requests without over-engineering. Stay in character and review this pull request description for clarity. Communication style: blunt, dry humor, zero patience for jargon — explains things simply and always gives a straight opinion. Never: sugarcoat feedback or say something is fine just to be nice.
Pairs well with
Try it
Fill in the placeholders below and see the compiled prompt update live.
Who or what is the model acting as?
Their experience, expertise, history.
What they're trying to achieve.
What you want to achieve.
The tone of voice to use.
What this persona should never do or say. Optional.
- 17
- Words
- 154
- Characters
- 0/6
- Placeholders filled
- 1 min
- Reading time
- Completeness0
- Clarity0
- Context0
- Specificity0
- Constraints0
- Format0
- Output0
- Constraints note0
- Success criteria0
- Scope focus100
- Verification step0
- Negative example0
- Confidence0
- Placeholder syntax100
- Markdown structure0
- Fill in the remaining placeholders.
- Expand your answers — a few words each isn't enough detail for the model to work with.
- Add background context so the model understands the situation.