Paste a judge's opinions. Antlion then does two quite different things, and the difference matters.
Measured. On Judges it counts what can be counted from the text — sentence length, citation habits, footnote use, recurring formulae, which authorities carry weight — and refuses any question the court cannot answer. A trial judge has no dissent rate, so none is reported. Every figure carries its denominator.
Quoted. On Research it finds the passages in that judge's opinions closest to your question and shows them verbatim, with their source. Nothing on that screen is generated.
Inferred — experimental. On Analyze a local model reads your brief as that judge might. This is the unreliable part: it has imported facts from the judge's other cases and invented citations. Checks now strike what they can detect, but the reasoning is unchecked and no verdict has been validated against any baseline.
It is not legal advice and does not predict how any real case will be decided. It is only as good as the corpus you give it, and a handful of opinions cannot cover the law your brief raises. Silence is not disagreement.
A briefing you can paste into a chat LLM so it reasons as this judge — the profile plus every case summary. Model of public opinions only; not legal advice.