Persona switcher
The same instructions delivered in a different voice produce noticeably different output. A “ruthless editor” trims and challenges; a “patient mentor” explains and reassures; a “startup founder” prioritises speed and leverage. This tool takes your existing system prompt and reframes its role and voice to match a chosen persona archetype — without touching the underlying task you defined.
How it works
You paste your base system prompt and pick a persona. The tool prepends a framing block that establishes the persona’s role, characteristic voice, and a few behavioural cues (how it handles disagreement, level of detail, default tone), then appends your original instructions unchanged below it. The result is a complete system prompt where the assistant carries the new personality while still doing exactly the job you specified.
Everything happens in your browser from built-in persona templates — there is no model call, no upload, and nothing is stored.
What each persona archetype actually changes
Different personas shift distinct dimensions of model behaviour:
Ruthless editor: shortens output, removes hedges and qualifications, challenges weak arguments, asks “why does this sentence exist?” The model will cut more than it adds, and will push back on content even when it is technically correct.
Patient mentor: increases explanatory depth, offers analogies, checks understanding, avoids jargon. The model treats every question as an opportunity to teach rather than just answer.
Startup founder: biases toward action and decisions rather than analysis and options. Responses favour the shortest path to a testable result — “ship it, learn” over “consider these seventeen trade-offs.”
Socratic teacher: answers questions with questions, surfaces the learner’s underlying assumptions, guides rather than tells. Useful for working through problems interactively rather than getting direct answers.
Senior analyst: weighs evidence, distinguishes confident conclusions from uncertain ones, explains the reasoning chain. Good for evaluating trade-offs, risk assessment, and research synthesis.
Practical guidance
The persona affects style and approach, not domain knowledge. A patient mentor persona applied to a coding task will explain more and hedge less — but the factual accuracy of the code still depends on the underlying model, not the persona framing. Do not use a persona as a substitute for giving the model correct domain information.
For best results, keep your base system prompt’s task instructions sharp and specific. The persona answers “how should the assistant behave?” while the base prompt answers “what should the assistant do?” — they are complementary, not redundant.
If a persona feels too strong (the editor too harsh, the founder too breezy), add one balancing line to your base prompt, such as “stay constructive” or “explain your reasoning.” And always read the assembled prompt before using it, so the voice and your instructions clearly reinforce rather than fight each other.