Text summariser
Pulls out the sentences that best represent a long piece of text.
- Runs in your browser — your text is never sent anywhere
- No sign-up, no quotas, free to use as often as you like
How it works
Summarising means one of two things. Writing a new, shorter version requires a language model and a server. Selecting the most representative existing sentences requires neither, works instantly, and cannot invent anything — which for a report or an article is often the more trustworthy result.
How it chooses
Words are scored by how often they appear across the text, ignoring the hundred or so filler words that appear in everything. Each sentence is then scored by the average importance of the words in it, so a long sentence is not favoured just for being long. Opening sentences are weighted slightly higher, because a piece usually states its subject early.
Nothing is invented
Every sentence in the summary appears verbatim in your text. That is the central advantage over a generative summary, which can state things the source never said. The trade is that it will not read as smoothly as something written from scratch — sentences pulled from different paragraphs sit next to each other without connecting phrases.
What it suits
Articles, reports, meeting notes, long emails and documentation — anything expository, where the important sentences are genuinely the ones carrying the most weight. It works poorly on narrative, dialogue and anything where meaning is built gradually rather than stated.
Ordering
Keeping the original order usually reads better and preserves the argument's structure. Ranking by importance is useful when you want the single most central sentence first.
Text summariser — questions
Does it write a new summary?
No. It selects existing sentences, so nothing is invented or misstated.
Why does it read a bit disjointed?
Because the sentences come from different parts of the text without connecting phrases. That is the trade for nothing being made up.
What kind of text does it suit?
Articles, reports and documentation. It works poorly on narrative or dialogue.
Is my text uploaded?
No. It is analysed entirely in your browser.