Skip to content
Free first run · no signup

AI Essay Review

Drop your essay in. Four rival AIs stress-test the argument, tighten the structure, and hand back the stronger version with the reasoning visible.

Your team is ready
🛡️Steelman🧱First-Principles Thinker😈Devil's Advocate🔬Data Skeptic🦉Philosopher
Free first run · no signup, no card · verdict in about 2 minutes
What you get back

A critique of the argument with the holes and the strongest counter named.

Signed off by an independent Final QA audit — with a “verify before you rely on this” list, never a black box.

Why this team

Arguments feel strongest to the person who wrote them, which is exactly when the holes are least visible. The Steelman builds the best version of the opposing case — the one your essay must actually beat, not the convenient one it currently quotes — while the First-Principles Thinker checks whether the conclusion follows from the premises or just sits near them. The Data Skeptic audits every load-bearing fact, the Philosopher names the unstated assumptions doing silent work, and the Devil’s Advocate attacks the argument the way its least charitable reader will — because that reader is coming.

Each seat runs on a different frontier model — GPT, Claude, Gemini, Grok — so it’s a genuine cross-model review, not one AI in costumes. How a run works →

What the debate sounds like

An illustrative excerpt — A policy analyst has drafted an op-ed arguing her city should replace minimum parking requirements with a housing-first zoning code, resting on three claimed benefits and one case study of another city.

🧱
The First-Principles Thinker

The logical spine has one strong link and two soft ones: the claim that parking minimums raise construction costs is well-established and the essay proves it cleanly. But “lower costs” slides to “more housing” without addressing who captures the saving — developers or renters — and “more housing” slides to “more affordable housing” without argument at all. Each slide is where an economist critic will stand. Argue the links or narrow the claims; the current chain asserts twice and proves once.

🛡️
The Steelman

The strongest opposing case, which the essay must answer to persuade anyone not already convinced: in areas with poor transit, parking demand does not vanish when requirements do — it spills into surrounding streets, and the burden lands on existing residents who lack off-street options. Serious critics hold that position with evidence. The current draft rebuts only the weakest version (“people love parking”), which reads as evasion to exactly the readers the op-ed needs.

🔬
The Data Skeptic

The case study carries more weight than one city can bear: the cited city removed minimums and added housing, but it did so during a construction boom, alongside transit expansion and tax incentives — three confounds the essay never mentions. One city with confounds is an anecdote wearing a dataset’s clothes. Either add the broader multi-city evidence, which exists and is more modest, or honestly present the example as illustrative rather than probative.

Questions

What kinds of arguments can I submit?

Op-eds, essays, policy memos, debate cases, thesis chapters, even a pitch’s reasoning section — anything with a claim and a chain of support. State the audience and the goal (persuade, inform, win, pass), because the standard of proof and the treatment of the counter-case shift with them.

Will it strengthen my argument or just find weaknesses?

Both, in order: the genuine strengths named first (so you protect them in revision), the weaknesses prioritised with the specific fix for each, and the strongest counter-argument your piece must answer, stated at full strength. Most drafts get stronger by narrowing claims and facing the real objection — the critique shows exactly where.

How is this different from a grammar or style check?

Entirely — this is a reasoning audit: logic, evidence, unstated assumptions and the counter-case, not commas. A grammatically flawless essay can be a house of cards, and tools that polish sentences never notice. Run the style pass after the argument survives; polishing a collapsing structure wastes the polish.

Your material is used only to run your review — never to train public models. Encrypted in transit and at rest. Security & privacy →