Are the key points written or quoted?

Quoted. Each point is a whole sentence copied from your text, shown with the score that ranked it and the terms that carried it. The extractor never paraphrases, which is why a point can be pasted into notes without changing its wording.

Why do two of my key points say nearly the same thing?

Because the extractor does not cluster near-duplicate sentences. If a document repeats a claim in different words and both sentences score well, both are returned. Remove the one you do not need, or lower the maximum number of points.

Can I use this to extract action items from meeting notes?

No, and it does not claim to. The scoring knows term frequency and sentence centrality; it has no concept of a decision, an owner, a deadline or a speaker. A sentence containing an action item may rank well, but so may a sentence that merely repeats the topic. Treat the list as a shortlist only.

What is the continuous extract for?

It joins the same chosen sentences back into document order. Rank order answers what matters most; document order answers what comes first, which is what you need when the points will be read as a passage or used as a talk outline.

Is the scoring the same as the text summariser?

Yes — 0.55 x TextRank centrality + 0.35 x log term frequency + 0.10 x position, normalised to the best sentence. The two tools differ only in how they present the selection: as numbered points here, as a paragraph there.

Key Points Extractor

Your text

Paste the text to extract points from

Maximum points1-20, default 5

Extraction

Start typing above — the results update as you write.

Scope and limits

Each point is a sentence quoted verbatim from your text and ranked by TextRank centrality (weight 0.55), log term frequency (weight 0.35) and position (weight 0.10). Stopwords are excluded, terms are matched case-insensitively, and the key-term table shows every count the scoring used.

Extraction is not understanding. Two sentences that repeat the same topic can rank above a sentence that carries the actual decision, and the tool cannot tell a key point from a rhetorical aside. It also reads your text as plain prose: headings, captions and list items are treated like any other sentence. Read the ranked list as a shortlist to check, not as minutes.

The key points extractor ranks the sentences of a passage and returns the top ones as numbered points, each with its score and the terms that carried it — extraction rather than generation, so every point is a verbatim sentence from your text.

What is Key Points Extractor?

A summary reads as prose; a list of key points reads as a list, and the difference matters when the output is going into meeting notes, a revision plan or a slide. Use this tool to extract key points from a long passage and get them back as numbered items rather than a paragraph. The key points extractor uses the same sentence-scoring engine as the summariser and presents its result as ranked points. The maximum is yours to set from 1 to 20, with five as the default. Nothing is generated: each point is a sentence lifted whole from the source, so a point can be checked against the original in a second.

Scoring is identical across the extractive tools and is published on the page. Each sentence is scored 0.55 x TextRank centrality + 0.35 x log term frequency divided by the square root of its length + 0.10 x position decay, normalised so the best sentence is 1.000. The point list shows the final score, the word count and up to four of the sentence's highest-weight terms. Below it, the key-term table lists every content term the scoring used with its count and its weight, which is the evidence for why one sentence outranked another.

Rank order and document order are different things, and the tool shows both. The numbered points run from most important to least important as the scoring sees it; the continuous extract beneath joins the same sentences back in their original sequence, which is what you want when the points need to be read as a passage rather than a checklist. Stopwords are excluded from the scoring, terms are matched case-insensitively, and the counts shown are the counts used.

The limits are worth stating plainly, because a ranked list invites more trust than a paragraph does. The extractor does not cluster near-duplicate sentences, so two points can say almost the same thing and both appear; it has no idea what a decision, an action item or a speaker is, so it is not a meeting-minutes tool despite the format; it treats headings, captions and list items like any other sentence; and the sentence splitter is mechanical, so an abbreviation such as Dr. can cut one sentence into two and names with initials can merge text unexpectedly. A document that is mostly bullet lists will produce points that are themselves bullet fragments.

Used within those limits, the ranked list is a fast way through long material. Paste a report and read the top five points before deciding whether the whole thing needs reading. Paste two drafts of the same section and compare which points each one ranks first. Paste a talk transcript and build a speaking outline from the sentences that carry the most shared vocabulary. In every case the output is a shortlist to check against the source, not a substitute for reading it.

Everything runs locally in this browser. There is no model, no training data, no network call and no storage: the points exist on screen and in whatever you copy or download, and closing the tab removes them. The scoring is deterministic, so the same text with the same point limit produces the same list every time.

Use it when the shape you need is a list and the honesty you need is quotation. If you want flowing prose instead, the summariser presents the same selection as a paragraph, and if the passage needs to fit a hard character limit, the TL;DR generator is the one that enforces a budget.

How to use Key Points Extractor

Paste the passage; the ranked points appear as you type.
Set the maximum number of points, from 1 to 20 — five is the default.
Read the numbered points in rank order, and the terms beside each one as its evidence.
Use the continuous extract to check the same sentences in their original sequence.
Copy or download the list and verify any point you plan to quote against the source.

When to use Key Points Extractor vs related tools

Use this extractor when the result needs to be a shortlist rather than a paragraph — meeting prep, revision notes, a talk outline. When you want the same selection written as flowing text, the Text Summarizer presents it that way, and when the output has to fit a strict character cap, the TL;DR Generator enforces one.

If the points will be quoted in published work, the Grammar Checker and the Spell Checker are the mechanical passes to run over the surrounding text first.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.