Paste a post, report, or policy summary. This page highlights the claims, the evidence, the assumptions, and the weak spots. It runs in your browser. Nothing you paste is sent anywhere.
No extraction yet.
Paste text above and click Extract to see claims, evidence, assumptions, and weak support highlighted here.
These notes walk through the extractor from first paste to final report. Read them once, then come back whenever you hit a snag.
The extractor works best on writing that argues for something. Opinion columns, policy briefs, research abstracts, and long posts all work well. Very short notes, raw data tables, or poetry usually do not produce useful results. If you are not sure, paste it anyway. The labels will tell you whether the text carries enough structure to break apart.
Strictness controls how picky the detector is. Easy mode catches only strong claim and evidence cues. Normal mode balances precision and recall for everyday articles. Strict mode flags more sentences, including borderline ones. Start with Normal. If you get too much noise, switch to Easy. If you miss important lines, try Strict.
Do not scan randomly. Read the claims first. Then read the evidence. Then check the assumptions. Finally, look at the weak support. This order shows you what the text wants you to believe, what it offers as proof, what it takes for granted, and where it leans on confidence instead of reasons.
The summary panel gives you a quick count of each label and a short written overview. Use it to decide whether the argument is evidence-heavy or claim-heavy. A long article with many claims and little evidence deserves more skepticism. A short post with one claim and three pieces of evidence may still be wrong, but at least it tried to support itself.
Click Save to browser to keep an extraction on your device. This helps when you are comparing several articles for a research project, a fact-check, or a class discussion. Saved entries stay in your browser storage. They are not uploaded anywhere. Clear them with the Clear history button when you are done.
Use Copy summary to grab a plain text version you can paste into a note, email, or chat. Use Download .txt if you want a file you can share, print, or attach to a report. Both exports include the labels, the original sentences, and the summary counts so other people can read your extraction without visiting this page.
Imagine you are reading a long article about a city proposal to replace parking spaces with bike lanes. You paste the article into the extractor. On Normal strictness, the page flags the sentence "bike lanes will reduce downtown traffic by 20 percent" as a claim. It flags a reference to a study from a transportation institute as evidence. It flags the sentence "shoppers will simply choose to bike instead of drive" as an assumption, because the article never proves that shift. It flags "everyone agrees this is overdue" as weak support, because "everyone agrees" is a vague appeal rather than a source.
Now you have a clear map. The article makes one main claim, backs it with one piece of evidence, rests on one big assumption, and leans on one weak appeal. You can decide whether that structure holds up. You might go find the original study. You might ask whether shoppers really switch modes that easily. The extractor did not answer those questions, but it showed you exactly where to ask them.
These short samples show how the detector labels different kinds of sentences. Open any preset in the extractor to see a full result.
Remote work should become the default for office jobs. A 2024 study of 12,000 workers found a 13 percent productivity gain. Most managers can learn to trust their teams without constant oversight. Everyone knows that commutes waste time.
School start times should shift later for teenagers. Sleep research shows adolescent brains need eight to ten hours. Districts that shifted times reported fewer absences. Bus schedules can absorb the change without major cost.
Soil carbon increases when cover crops are used for three or more seasons. Across 41 trials, soil organic carbon rose by an average of 0.3 percent per year. Farmers will adopt cover crops at scale if seed costs stay low. This approach is clearly the future of sustainable farming.
AI writing tools are making students lazy. Everyone can see it happening. Students who use AI never bother to think for themselves. One professor reported a drop in original arguments in essays.