Tone is the most under-used lever in AI prompting. Most people accept whatever register the model defaults to, then wonder why their marketing copy sounds like a press release or their internal Slack message sounds like a legal memo. This AI tone and register reference maps common writing tasks and audiences to an appropriate point on the formal-to-playful scale, and hands you a copy-ready instruction to drop at the top of your prompt.
How it works
Pick a task type, an audience, and a channel. The tool scores those choices and lands on one of five registers — academic, professional, conversational, friendly, or playful. Each register comes with a plain-language description, a short list of things to do and avoid, and a one-line worked example so you can hear the difference. The generated prompt snippet turns that register into a concrete instruction the model can follow, naming sentence length, person, and contractions rather than just a vague adjective.
The mapping is deliberately opinionated. A regulatory filing for executives lands in academic or professional; a product launch tweet aimed at consumers lands in friendly or playful. When two inputs disagree — say a formal task on a casual channel — the tool nudges toward the channel, because the medium usually sets reader expectations more than the topic does.
Tips and examples
The biggest mistake is stacking contradictory cues, like asking for something “professional but fun and edgy and authoritative”. Pick one anchor register and add at most one modifier. If you want warmth inside a professional email, ask for “professional tone, but warm and human — use the reader’s name and one contraction per paragraph”.
Use the example lines as calibration. If the playful example feels too much for your brand, step one band toward friendly and re-read it. Because register changes word choice and rhythm together, moving a single band is usually enough to shift the whole feel of a draft without rewriting your actual content.