Free with an account · 3 lengths in one run

Summarize any text in three lengths.

Paste essays, transcripts, emails, or long articles — get a TL;DR, a standard summary, and a detailed version, plus key points and keywords. One run.

0 words · 0/50,000

Focus

Audience

Style

See how it works — click any example

Works on
  • Articles
  • Research papers
  • Transcripts
  • Long emails
  • Meeting notes
  • Book chapters
  • Lecture notes
Length & tiers

100 to 50,000 characters per run. Three summary lengths produced in one pass — TL;DR (20-55 words) / standard (80-160 words) / detailed (180-320 words). Six focus modes (general / academic / news / business / technical / key-facts) and three audience levels tune what gets emphasized.

You get

Headline, TL;DR, standard paragraph, detailed summary, 5-10 key points, 5-12 keywords, named entities, sentiment, and tone — all from one credit charge. Convert to a mind map or flashcards, or pull from a URL with the article summarizer.

Won't work on

Under 100 characters — too little to summarize. Over 50,000 characters — split into chunks, or use the PDF summarizer for longer documents. Will never invent facts or neutralize opinionated source text.

Pricing

Free with an account30 runs/month. Upgrade to Pro for 1,500/month.

Sample output

Here's what a real summary looks like.

Input: 470-word study on remote-first engineering teams · English output, trimmed.

Headline

Remote-first teams ship more individually but erode mentorship, collaboration, and connection unless deliberately protected.

TL;DR

A 1,200-engineer survey found remote-first work raises individual output 11% but reduces mentorship by a third and cross-team collaboration by 22%. Companies that fund retreats and structured mentorship reverse the trend.

Standard summary

Across 47 companies, remote-only engineers produced 11% more individual output than hybrid peers but reported a 33% drop in mentorship and a 22% drop in cross-team pull requests. Forty-one percent felt isolated most days. The companies that avoided this trade-off did not mandate return-to-office — they funded in-person retreats, paired new hires with tenured mentors for 90 days, and protected one meeting-free day per week. These companies saw 18% higher retention with no drop in individual output. The takeaway: remote is not a neutral default; second-order effects like mentorship and cross-pollination must be actively protected as engineering investments, not HR soft skills.

Key points

  • Individual output (commits, tickets, shipped features) rose 11% for remote-only engineers versus hybrid peers.
  • Informal mentorship between junior and senior engineers dropped by roughly a third in remote-only settings.
  • Cross-team collaboration measured by shared pull requests fell by 22%.
  • 41% of remote-only respondents reported feeling isolated "most days" of the work week.
  • Companies that paired new hires with tenured mentors for 90 days showed 18% higher retention without output loss.
  • One meeting-free day per week and periodic in-person retreats were the two interventions that correlated with reversal.
  • Junior engineers at remote-only companies without mentorship rated career growth 1.4 points lower on a 5-point scale.

Keywords

remote work engineering teams mentorship retention cross-team collaboration workplace isolation asynchronous work

Detected

Content type: research / survey · Language: English · Sentiment: mixed (positive on output, negative on connection) · Reading level: college · Source: 470 words · Compression: 53% (standard), 17% (TL;DR)

Try it with any text · Results usually in 10-30 seconds · Copy as markdown or share via link

Questions & answers

What does this tool actually produce? #

You paste any block of text (100-50,000 characters) and we return six things in a single run: a headline, a TL;DR (1-2 sentences, 20-55 words), a standard summary (80-160 word paragraph), a detailed summary (180-320 words), 5-10 key points, 4-8 short skim bullets, 5-12 keywords, named entities mentioned in the text, overall sentiment, tone, and the source reading level. You get all three length tiers from one API call — no rerunning to get a different length.

How is this different from your Article Summarizer or PDF Summarizer? #

The Article Summarizer takes a URL and fetches the page for you. The PDF Summarizer takes a file upload. This one takes raw pasted text — any source, any format. Use it for transcripts, emails, pasted book chapters, meeting notes, anything where you already have the text and just want a summary.

Does it invent facts or add commentary? #

No. The prompt explicitly forbids adding facts, numbers, names, or context that are not in the source. If the source hedges ("may", "could"), the summary preserves the hedge. If the source argues for a position, the summary conveys that it argues for it — it does not neutralize into "the text discusses both sides" unless the text genuinely does.

What counts as "banned phrases" in the output? #

We strip about 30 tired phrases that signal AI filler — "In conclusion", "In summary", "The article discusses", "In today's fast-paced world", "delve into", "leverage", "paradigm shift", "game-changing", and similar. A summary should read like a sharp editor wrote it, not an overcaffeinated B2B blog.

Can I pick a focus or audience level? #

Yes. Six focus modes — general, academic, news, business, technical, key-facts — tune what the model emphasizes. Three audience levels (general / beginner / expert) tune how much jargon is defined versus preserved. The combination handles everything from "summarize this research paper for my PhD committee" to "explain this SEC filing like I have never seen one".

How long a text can I paste? #

Minimum 100 characters (about 20 words). Maximum 50,000 characters (roughly 8,000-10,000 words, or a full research paper chapter). Larger texts are truncated. For longer content, paste in chunks or use the PDF Summarizer.

What languages work? #

Any language the model reads — English, Chinese, Japanese, Korean, Spanish, French, German, Russian, Portuguese, Italian, and dozens more. The summary is returned in the same language as the source by default, or you can set a target language in the request. The English UI comes first; other UI translations roll out next.

What about transcripts with ums, filler, and speaker turns? #

Set focus to "general" or "news" and we collapse verbal filler (um, uh, you know) and repetition automatically. We flag in the output that the source was a transcript. Speaker attributions are preserved if the transcript has them.

Is the word count honest? #

Yes. We count the words in each summary tier server-side and show the true numbers on the result page, including compression ratio versus the source. LLMs are known to undercount their own output by 10-30% — we never trust the model's self-report.

Is it really free? #

The first 3 summaries per browser per day are free with no signup. Sign up for 30 summaries per month free, or go Pro for 1,500 per month plus history, export, and no rate limits.

Why three length tiers in one run? #

Because skimming is layered. You look at the TL;DR to decide whether to read further. If yes, you scan the standard summary to triage. If you need detail, you read the detailed summary. Getting all three in one pass means one wait, one credit charge, one share link — with the length choice made at the moment of reading, not at the moment of generation.