How it works

You paste a piece of writing. An AI model reads it and returns a short analysis: a one-or-two-sentence summary of how the text reads, plus underlines on phrases worth noticing. That's the whole interaction.

The rest of this page is about why the tool is built the way it is, and why it does some things differently from what you might expect.

The reader frame

When you write something — an email, a message, a post — you can't fully see how it lands, because you already know what you meant. The recipient doesn't. They have only the words on the page. The gap between what you meant and what they read is where this tool tries to be useful.

The analysis reads your text the way someone would who received it cold, without any prior context about you or the situation. Attentive, neutral. Not looking for problems, not smoothing them over either. The summary names how the text reads when read that way. The underlined phrases point at specific words doing specific work.

What the tool doesn't do

It doesn't rewrite your text. There's no "suggested version," no auto-improved draft, no "here's what you should say instead." The tool reads what's there. What you do next is your call.

It doesn't give advice. No "consider softening this," no "you might want to," no "try saying it this way." The analysis describes how the text reads. It stops there.

It doesn't evaluate craft. The tool isn't grading your writing. It's not telling you whether the text is good or bad, well-written or poorly written, professional or unprofessional. Those are different questions, and other tools handle them. This one reads how the text lands, which is its own question.

Why no scoring

There's no number, no grade, no percentage, no five-star rating. The reason is simple: a number would imply a standard, and there isn't one. Text doesn't land a single way to a single reader. It reads warmly to a close friend and coolly to a stranger, professionally to a colleague and stiffly to a sibling. A score would pretend that variance doesn't exist.

The summary names how the text reads when received cold. That's a real observation about a specific reading frame. It's not a verdict on the writing. The difference matters.

Where it works and where it doesn't

The tool is useful when you want a second read on something short — usually a page or less. Emails, messages, posts, replies, drafts, things you're about to send. It works on writing you wrote and on writing you received. The reading frame is the same either way.

It's less useful for long-form work. The analysis is short by design — it's a read, not a review — and a short read of a long piece doesn't help much.

And the analysis is generated by an AI model, which means it can be wrong. It can miss something a human would catch, or weight something a human wouldn't. The output is one read of your text, by one reader, in one moment. It is not professional advice — not legal, not psychological, not editorial. Treat it as input, not as authority.

See it in action

The fastest way to understand the tool is to see what it produces. The examples page shows real text run through the tool, with the resulting analysis rendered the way you'd see it on the homepage. Or paste something of your own at the home page.