Rewrite text in any tone, with your own key
Sometimes a draft says the right thing in the wrong voice — too stiff for a landing page, too breezy for a legal notice. The Tone Adjuster takes a piece of text and a target tone and rewrites it while keeping the facts intact, using your own OpenAI or Anthropic API key so nothing passes through a third party.
What each tone option produces
Formal steers toward longer sentences, Latinate vocabulary, complete constructions, and the third person where possible. Appropriate for legal notices, reports, executive communications, and official documentation.
Casual shortens sentences, favors contractions, uses everyday vocabulary, and permits fragments. Appropriate for app copy, onboarding flows, chat responses, and consumer-facing landing pages.
Assertive eliminates hedging, qualifications, and passive constructions. Every claim stands on its own without “perhaps,” “it might be,” or “one could argue.” Appropriate for calls to action, sales copy, and persuasive arguments.
Empathetic centres the reader’s perspective and feelings. Uses “you” and acknowledgment language before information. Appropriate for rejection emails, support responses, and sensitive announcements.
Concise removes filler, collapses redundant clauses, and trims until each sentence carries only its core meaning. The character-count diff is often the most dramatic here.
Playful uses lighter vocabulary, wordplay, occasional humor, and a conversational cadence. Appropriate for social media, games, campaigns, and brands with an informal personality.
How it works
You paste your API key, your text, and pick a tone. The tool builds an instruction that tells the model to rewrite the passage in the chosen tone while preserving every factual claim, then calls the provider’s REST endpoint directly from your browser — OpenAI’s chat completions or Anthropic’s messages API. It shows the rewritten text next to a character-count diff so you can see whether the new tone expanded or tightened the copy. Loading and error states are handled explicitly, and your key is never stored.
Tips for good rewrites
- Pick a cheaper model for drafts. GPT-4o mini and Claude Haiku rewrite tone well at a fraction of the cost of the flagship models.
- Proofread for meaning drift. A tone change can subtly reframe a claim — always read the result against the original.
- Keep inputs focused. Rewriting one paragraph at a time gives tighter, more controllable results than dumping an entire document.
- For documents too long to fit in context, summarize first, then adjust tone on the summary.
- Use the character-count diff as a tone signal. Formal typically expands; concise always shrinks; casual and playful vary. A large expansion from “concise” is a sign the model ignored the instruction — re-run or try a different model.
- Run detection before and after. If you used the passive-voice detector before adjusting tone, re-run it afterward — assertive and concise rewrites typically reduce passive-voice count; formal rewrites sometimes increase it.
- Tone is cumulative across a document. Adjusting individual paragraphs in isolation can produce an inconsistent result. Adjust the whole document in one pass, then tweak specific sections.