Studycase

Watch a hearing.
Know what argument to make.

A study cockpit for law students in virtual court: a live transcript that scrubs the video, an AI coach that spots the moment — the objection, the pressed concession, the bench's hypothetical — and case-law cards grounded in real authority, never invented.

A study tool, stated plainly · Studycase is for moot-court and mock-trial preparation and courtroom literacy. It is not for assistance during real proceedings, and not for graded exams or assessments where outside aid is prohibited — it builds courtroom judgment in practice, it does not supply it during something that counts.
▶ Watch a demo sessionRead the guide

The demo replays a scripted, clearly-labeled illustrative hearing — no sign-up, no microphone, no screen sharing. Live sessions need Chrome or Edge; the demo runs anywhere.

The cockpit — three panes, one join key

The Studycase cockpit mid-session: a live transcript pane on the left, two-ink notes and AI analysis in the center, and case-law suggestion cards — including a verified genius line quoting Terry v. Ohio — on the right.
Illustrative — a scripted demo session (Dalton v. City of Kerrville, parties and dialogue invented for demonstration; the precedents cited — Terry, Graham, Scott — are real opinions). Captured from the actual product, not a mockup.

The transcript settles — and never walks backward

Committed words never reflow, captions break at clause boundaries, and every word is a timestamp anchor — click it and the video scrubs there.

Your notes stay yours

Two inks: your typing in full ink, AI analysis in muted gray, labeled with when it was written. Edit a gray block and it becomes yours; every export labels which was which.

Provenance or silence

Case cards quote the moment that triggered them and cite real opinions from a real corpus. A “genius line” has its authority verified word-for-word before you see it — or it says nothing at all.

Tell the coach your goal (optional)

State what you're trying to accomplish and the coach orients its suggestions toward it from the first moment. It steers emphasis only — it never invents authority, and adverse law you have to deal with still surfaces.

Choose your model
How do these two compare? Private ties Opus on safety, edges it on quality (0.925 vs 0.881) — see the benchmarks ▸

Measured on our own 30-scenario legal exam (7 hearing types, real cases, deterministic scorer):

  • Pooled quality — Private 0.925 vs Opus 0.881 (+0.044 (14B leads by ~5.0%))
  • Safety axes tie — say-nothing 5/5=5/5, citation-gate 29/30=29/30, adverse authority tied.
  • Cost — Opus meters ~1.65¢ per coaching moment; the private tier bills ~$0 marginal on owned hardware.

Opus edges no-invention; the private tier edges coaching directness. Our own eval on real cases — educational, not legal advice. Full per-axis breakdown on the Benchmarks page.

0/600
Start a session with this goal →Optional — you can start without a goal.

Honesty is the brand

No fabrication is architecture here, not policy: suggestions are retrieval-grounded, confidence is coarse (Strong / Relevant / Speculative — no fake percentages), AI text is always visibly labeled, and anything unverified is refused rather than guessed. What the tool does not do yet, it says out loud — see the accessibility statement for the same stance applied to conformance.