Paste text and get back its most load-bearing sentences, extracted word-for-word, no paraphrasing, no invention.
This is an extractive summarizer, and the word matters. It does not rewrite anything: it scores every sentence by the frequency of its significant words across the whole text, with a modest bonus for appearing early (where writers put their main points), then selects the top-scoring sentences and returns them in original order. The output is therefore guaranteed faithful, every sentence is the author's own, at the cost of flow: transitions are missing, and a summary of a badly structured text reads jumpily because the source does.
Extractive summaries shine for reports, articles, meeting notes and long emails where the reader wants the gist before committing. They cannot capture an argument that builds across sentences, a legal clause or a proof, and very short inputs (under ~6 sentences) have nothing to compress. Everything runs locally; the text never leaves the browser.
How is this different from an AI summary? An AI summarizer paraphrases and can compress harder, but it can also drift from the source or invent connections. This tool only selects real sentences, verifiable, quote-safe, and instant.
Why did a sentence I consider key get dropped? Scoring follows word frequency and position. If the key sentence uses unique vocabulary stated only once, it scores lower, lengthening the summary is the practical fix.
What text length works best? Roughly 300 to 5,000 words: enough sentences to rank meaningfully, short enough that selection still discriminates.