AI Tone Detector

An AI tone detector reads a piece of writing and tells you how it is likely to come across to the person receiving it. Paste an email, a chat message, a post or an article and a language model reports six things: the primary tone, the sentiment with a confidence level, formality on a scale of 1 to 10, the emotional undertones, how a reader is likely to feel, and specific suggestions that quote the phrases worth changing.

It exists because tone is hard to judge in your own writing. You know what you meant, so you read your intent rather than your words. The sample on the page shows the problem: "Per my last email, the report was due Friday. I'd appreciate it if we could avoid this happening again." The writer may think it is firm but polite. Most readers hear irritation, and the detector should point at "Per my last email" as the phrase doing the damage.

The analysis runs on the Toolz server. Your message is sent over HTTPS, the server builds the prompt and runs Gemma 3 1B, a small open model from Google, with Ollama, and the report streams back within a few seconds. That matters here, because the messages people want checked are often the sensitive ones: a complaint, a reply to a manager, a breakup text. The text is used to write the report and then discarded: it is not stored or logged, not used to train any model, and not sent to any third-party AI company.

Treat the result as a second opinion, not a verdict. Tone depends on the relationship and the context the model cannot see, sarcasm is easy to misread, and the confidence level is the model's own estimate rather than a measured statistic. A 1B model can also over-read neutral text as negative. Where the stakes are high, ask a person you trust.

It is free with no signup, within about 20 checks per 10 minutes. The alternatives are reading the message aloud, waiting an hour before sending, or asking a colleague. This tool is quicker than all three, and it names the exact words to soften, which you can then reword with the paraphraser.