Comparison

Built to disagree. On purpose.

Most AI products optimise for being helpful. Counteraxiom optimises for being useful — even when that means telling you your idea is wrong. Here's the practical difference.

What it doesGeneral-purpose AI assistantPolite review toolCounteraxiom
Disagrees with you on demandAlways
Concedes when you push back hardAlmost alwaysOften, to seem fairArchitecturally cannot
Argues the strongest counter, not the politest oneYes — steelmans the opposition
Built-in adversarial modesHard / Soft / Feedback / Off
Pre-built for high-stakes practice (interview, pitch, negotiation)SometimesDedicated rooms with scored debriefs
Live web search during argumentYesYes
Tracks your reasoning over timeProgress dashboard + AI Memory
Saves your best counter-argumentsArgument library
Adapts to your specific context (your pitch, your case, your essay)Per-session onlyPer-session onlyPersistent persona notes
Voice in / voice outSomeYes (Pro+)
Free to start, no credit cardYesOftenYes

Honest disclosure

Counteraxiom runs on the same class of large language models powering the “general-purpose” column. The difference isn't the model — it's the system prompt, the ban list (28 caving phrases auto-regenerated), the adversarial modes, and the purpose-built rooms for high-stakes practice. We're open about that. The point isn't a new model. The point is a new behaviour.