Bring your argument.We'll find where it breaks.

A structured adversarial exercise for researchers. Not here to help you write — here to help you think.

This is not AI brainstorming. It's a structured dialogue grounded in formal argumentation theory — Toulmin's model, Walton's schemes, steelmanning — used internally to ask better questions than a friendly collaborator would.

Three phases, 40–50 minutes. We surface the assumptions you didn't know you were making, test your argument against the strongest objection we can build, and ask what evidence would actually change your mind. Then, if the argument survives, we sharpen it.

It's slower than asking AI to write something. That's the point.

Choose your starting point

You can start with a paper, a hypothesis, or just a question. Each leads somewhere different.

Start with a paperPrimary

Upload a PDF — draft, preprint, chapter, grant proposal. The AI reads your argument as constructed, not as you've summarised it.

Start with an abstract or idea

Have a hypothesis in progress? Paste it here and we'll test whether it holds under pressure.

Start with a question

Don't have an argument yet. Sometimes the most important move is to examine the question before you try to answer it.

PDF upload

Click to upload a PDF · Paper, preprint, draft chapter, grant proposal

Ready?

Enter your name or initials so we can track your session.

The three phases

1 · Surface the assumptions ~10 min

Most arguments fail not because the conclusion is wrong but because a hidden premise is doing all the work. We find it. I read your argument and ask questions designed to make the implicit warrant explicit — not "what's your assumption?" but "why does this evidence support this claim rather than an alternative explanation?"

2 · Stress-test the argument ~20–30 min

Before building the objection, we map what you already know. What literature supports your position? What worries you — the study you're hoping no one has already done? From the gap between those two, I construct the strongest objection I can, drawn from relevant literature, not a strawman. Then: what evidence would actually change your mind? We try two or three alternative framings and examine whether they're more defensible. It's not about winning. It's about understanding the shape of the claim.

3 · Rewrite with eyes open ~10 min

We write down what changed. Not a summary of the conversation — a position. One paragraph that captures where you've landed, what you're claiming, and what you now know you were assuming. You stay in control of the voice and substance. The goal isn't a perfect text — it's one you can stand behind with a clear conscience.

What you leave with

A stronger argument Clearer, more defensible, with assumptions stated and scope honest.
A reasoning trace A structured map of what changed — assumptions surfaced, objections named, falsification conditions stated. Generated from the session transcript for your own records.
Falsification conditions The specific evidence that would change your mind, stated precisely enough to be useful.

The reasoning trace is captured for research analysis only. No individual data shared without your consent.

Honest limitations

Takes 40–50 minutes. If you want speed, this isn't it.
The steelman objection is only as strong as what can be constructed from available literature. Real peer review will surface more.
The literature review supplements your own — it doesn't replace it.
If your argument has a structural flaw, we'll name it. This session doesn't guarantee a polished output — it guarantees an honest diagnosis. Whether you came with an abstract, a question, or a PDF, you leave with a position you can actually defend.