01
Pick your judge
Select the named judge you’re appearing before from the live calibrated bench. If your judge isn’t yet live, fall back to a court-style general bench calibrated to the forum (Commercial Division, Court of Chancery, etc.).
How It Works
From brief to debrief in under thirty minutes.
Mootly.ai sits between the work you’ve already done — a brief, a hearing date, a judge — and the moment you stand up to argue. Below is the full workflow, end to end.
02
Upload your brief
PDF or DOCX, both sides. We extract the tension map — the specific points of disagreement between the briefs — automatically. Takes about thirty seconds.
03
Review anticipated questions
Before you argue, you see 4–6 anticipated questions this judge is most likely to ask, grounded in how she has phrased similar questions in prior proceedings. You can edit, prioritize, or skip.
04
Argue in real time
Step up to the lectern. The judge interrupts you the way this judge actually interrupts. Pushes on the weak points this judge actually pushes on. Quotes your record back to you.
05
Get a transcript-grounded debrief
Six skill scores: responsiveness, command of record, doctrinal precision, handling weak points, concision, composure. Each weakness comes with a citation to a real prior proceeding.
06
Replay, share, repeat
Audio playback of every turn. Shareable link to a partner or coach. Run the same case three more times before Friday.
Under the hood
What’s actually happening when you press “Start argument.”
A Mootly session is not a generic chatbot wearing a robe. Three things happen behind the scenes that distinguish it from generic AI legal tools.
1. Speaker-attributed corpus retrieval
Every transcript in the corpus is parsed line-by-line and tagged with who said it — the judge, counsel for petitioner, counsel for respondent, court clerk. That gives us a clean index of this judge’s words, separated from the words of the lawyers in front of her. The simulation queries that index by your case’s tension map.
2. Judge-partitioned vector search
When you upload your brief, we embed the issues and search the judge’s own questioning history for analogous moments. The simulation’s questions, hypotheticals, and pushbacks are weighted by what this judge has actually done — not what an LLM thinks a judge would do.
3. Behavioral exemplars in every prompt
Each turn of the conversation injects fresh exemplars from the judge’s record into the LLM’s context. The character does not drift. The voice does not become generic. The questions stay grounded in actual transcripts.