Your firm's AI persona
The Agent learns how your firm works. Your drafting style, your clause preferences, your legal reasoning, captured once and applied automatically in every session.
An agent that works
the way you do.
Your AI persona built from your own documents at onboarding — how you argue, what you cite, how you conclude.
Writing style drafts read like you wrote them — your phrasing, your paragraph rhythm, your headings.
Matter context knows the client, the court, the stage and the next date, so you never re-brief it.
Common questions
A profile of how your firm works, built from your own documents at onboarding. How you structure an argument, which authorities you reach for, how you phrase a conclusion, how long your paragraphs run. Once it exists, drafts come back reading like your work rather than like generic legal English.
A handful of documents you consider representative — opinions, pleadings or agreements you have actually filed or sent. The twin is derived from those. The more characteristic the sample, the closer the match.
Yes. The twin is stored as readable notes — persona, structure, authorities, tone — that you can open, edit or delete. Nothing about how the agent writes for you is hidden from you.
No. It learns the shape of your writing, not its content. Your documents are not reproduced in other people's work and are never used to train a shared model.
That is the normal case. Twins can be personal as well as firm-wide, so a partner's opinions read like that partner while the firm's standard agreements follow house style. You choose which applies to a given piece of work.
The twin is how you write. Memory is what the firm knows — the clause positions you take, the client facts, the matter history. The twin shapes the prose; memory supplies the substance. They work together and are managed separately.
Yes. Any session can be run without the twin when you want a neutral draft, for instance when writing for a client whose house style differs from yours.