About Argument Compiler
The Argument Compiler turns a pasted disagreement into a structured breakdown: what each person claims, which phrases are absolutes or mind-reading, which words both sides share, and a short template for a calmer conversation. It is free, needs no sign-up and runs in your browser, so the conversation you paste is never uploaded.
It is rule-based. There is no AI service behind it and no language model. It looks for simple word patterns such as "always", "never" and "you think", so it is a structured aid and not a judge. It cannot tell who is right, it does not check whether an argument is logically valid, and it is not a fallacy checker or an argument map.
Use it for an argument with a partner, friend, flatmate or colleague that is going in circles. Reading your own words laid out in sections often shows how much of the heat comes from phrases like "you never" rather than from the issue itself.
How to use Argument Compiler
- Paste the conversation. Label each line with a name and a colon, such as "Me: ..." and "Them: ...", so the tool knows who said what.
- Press Compile.
- Read the five sections: Actual disagreement, Hidden assumptions, Absolutes to soften, Common ground and Shortest conversation.
- Fill in the blanks of the Shortest conversation template with your own details, and use it as a script for talking it through.
- Press Copy conversation, Copy full report (Markdown) or Download .md to keep the result.
Features
- Splits the text by speaker, from "Name: text" labels or, without labels, from alternating paragraphs or lines.
- Actual disagreement: up to three claim sentences per speaker, with flagged ones first.
- Hidden assumptions: flags mind-reading phrases such as "you think", "you just want" and "obviously".
- Absolutes to soften: flags "always", "never", "everyone", "nobody", "you don't" and similar.
- Common ground: the content words that every speaker used.
- Shortest conversation: four neutral turns with blanks for your own details.
- Copy the conversation or the whole report as Markdown, or download it as a .md file.
- Also recognises a few common Bangla phrases for absolutes and assumptions.
Worked examples
Me: You never help with the dishes. I do everything around here. Them: You always say that. Obviously you think I do nothing. I washed the dishes on Monday. Me: The dishes pile up every week. gives Absolutes: "never" (Me), "always" (Them) Assumptions: "you think" (Them) Common ground: dishes Shortest conversation: "Me: I want to sort out dishes. When [specific thing that happened], I felt [feeling]."
The report flags the words that raise the temperature and names the topic both people keep returning to.
What each section is
| Section | How it is built |
|---|---|
| Actual disagreement | Up to three sentences per speaker that are not questions and have at least three words, flagged ones first |
| Hidden assumptions | Sentences that match mind-reading phrases such as "you think" or "obviously" |
| Absolutes to soften | Sentences that match words such as "always", "never", "everyone" or "you don't" |
| Common ground | Words of four letters or more that every speaker used, most frequent first, up to eight |
| Shortest conversation | A four-turn template with the shared topic filled in and blanks for you |
Good to know
What it can and cannot do
It matches words, not meaning. "I never said that" is flagged as an absolute even when it is true, and a hurtful sentence with none of the trigger phrases is not flagged at all. A sentence it flags is a prompt to look again, not a verdict.
The topic in the template is the most common shared word, so it can be a little off. Change it to the real issue when you fill in the blanks.
It is not a fallacy checker, an argument-map tool or a logic validator. It does not name fallacies, grade your reasoning or take sides.
Privacy: how it works
- The compiling runs in your browser tab. The conversation you paste is not sent to Ulkaa or to any other server, and no AI service is called.
- This tool does not save your text. Close or reload the page and it is gone, unless you copied or downloaded the report.
Tips
- Label every line with a speaker name. Without labels the tool guesses the speakers from alternating paragraphs or lines.
- Paste the exact words if you can. Paraphrasing removes the phrases the tool looks for.
- Use the Shortest conversation as a starting point and rewrite it in your own voice.
- If the conversation is about something serious or unsafe, talk to a person you trust or a professional. A word-pattern tool is not a substitute.
Frequently asked questions
Is it really free?
Yes. There is no price, no trial and no credit card. Ulkaa Tool does the work with your own device, so there is no server bill to pass on to you.
Do I need an account?
No. There is no sign-up and no login. Open the page and use it. An Ulkaa account is not needed for anything on this tool.
Does it run in my browser?
Yes. The conversation you paste are processed by code running in your browser tab, on your own device.
Is my data uploaded?
No. The conversation you paste are never sent to Ulkaa or to anyone else. You can check this yourself by watching the Network tab in your browser developer tools while you use the tool.
Does it use AI?
No. It is rule-based: a set of word patterns that run in your browser. No AI service or language model is involved.
Is it a logical fallacy checker?
No. It does not name or detect logical fallacies. It flags absolute words and mind-reading phrases and shows the claims and common ground.
Can it tell who is right?
No. It cannot judge an argument. The page says so itself: it is a structured aid, not a judge.
How do I format the conversation?
Put each turn on a new line starting with a name and a colon, such as "Me:" and "Them:". Two or more different names are needed. Without them it splits by paragraph or line and calls the speakers A and B in turn.
What can I export?
The shortest-conversation script, or the full report as Markdown, either copied or downloaded as argument-breakdown.md.
Does it work in languages other than English?
The sentence splitting handles the Bangla danda, and a few common Bangla absolutes and assumptions are recognised. The word patterns are mainly English, so results for other languages are limited.