Law firms run two kinds of meetings, and today they usually buy two products to do it. For the everyday register — client calls, consultations, team meetings — there's a video subscription that doesn't know what a matter or a privilege is, parking recordings of privileged conversations on a vendor's cloud. For the evidentiary register — depositions, proceedings — there's separate remote-deposition technology that doesn't know the firm's practice. Two line items, two systems, and the testimony that wins the case ends up living in neither one's idea of a case file.
EdgeMeet collapses both into one application with two registers. Most days it's the firm's video platform, running on the firm's own media infrastructure. And when the meeting is a proceeding, one button — go on the record — turns the same room into a deposition machine. Same app, same button; the difference is what happens when you go on the record.
The everyday register: your video platform, on your infrastructure
Start with the unglamorous half, because it's the half that runs every day. EdgeMeet replaces the video subscription: client calls, consultations, and team meetings on media infrastructure the firm controls, so privileged conversations never transit a third-party cloud. Clients join from a plain link — no downloads, no accounts, none of the five minutes of "can you hear me" that starts every consultation on someone else's platform.
And because the meeting happens inside the practice platform, it knows what it's about. Every meeting links to its matter, so the recording, transcript, and action items land on the case instead of in a generic cloud library sorted by date. Retention follows firm policy, not a vendor default. If you've ever hunted through a video platform's recording list for the client call where the settlement authority was discussed, you already understand why the meeting belonging to the matter is not a small feature.
Economically, this half of EdgeMeet retires two line items — the video subscription and the remote-deposition platform. Which brings us to the second register.
Go on the record
When the meeting is a proceeding, deposition mode brings the live transcript, goals, exhibits, and the record itself into the room. Testimony scrolls as text while it's spoken — which changes how the examining attorney works, because "what exactly did the witness say twenty minutes ago" becomes a thing you look up mid-session rather than reconstruct from notes.
Exhibits work the way exhibits should: searched from the DMS — the same document system holding the rest of the case — marked, annotated live, and withdrawn, all on the record. No pre-meeting scramble to upload documents into a separate deposition platform, because the deposition is happening inside the system where the documents already live.
The room's boundaries are engineered for what a proceeding requires. Every recording begins with a room-wide notice. Guests — outside counsel, witnesses — join by personal link through a consent-before-media page, structurally walled off from the attorney panel and the case context, and revocable in one action that also ejects them from the room. "Structurally" is the operative word: the wall between what your side sees and what the witness sees isn't a setting someone might misconfigure under pressure; it's how the room is built.
Didn't plan for the record going in? Any recorded meeting can be converted to a deposition after the fact — the client call where something important was said can become a transcribed, analyzed record retroactively. And the door swings the other way for history: existing Zoom or Teams recordings can be ingested and run through the same transcription, diarization, and second-chair analysis, so the testimony sitting in a vendor's cloud from last year's cases isn't stranded there. The register is a mode, not a separate purchase.
The AI second chair — for your side of the table only
Every litigator knows the second-chair problem. The examining attorney is managing the witness, the outline, and opposing counsel simultaneously — and the contradiction with an answer from three hours ago, or with a document produced in discovery, slips past because no human can hold all of it live. That's what a second chair is for, when the case can afford one.
In deposition mode, Lex listens live and surfaces cited insights to the attorney panel only — contradictions against the discovery record, new admissions, suggested follow-ups. Witnesses and opposing counsel never see any of it. Picture a hypothetical deposition in Hale v. Northstar Logistics: the witness testifies that a vehicle inspection happened in March, and the attorney panel quietly surfaces a contradiction — citing the maintenance log in the discovery record that says otherwise — while the witness is still on the record and the follow-up question can still be asked.
After the session, the insights become typed cards — CONTRADICTION, ADMISSION, FOLLOW-UP — each citing the transcript sequence and the exact media timestamp, quoting the testimony verbatim, and waiting for a decision:
- Accept, and the insight enters the case record with its citations attached.
- Dismiss, and it doesn't — nothing the AI noticed becomes part of the case until a lawyer says so.
Same app, same button. The difference is what happens when you go on the record.
Certified transcripts, and clips cut the way lawyers cite
The live transcript is a working tool; the court reporter's certified transcript is the record. EdgeMeet treats that distinction properly: when the certified transcript arrives, every segment and every insight citation upgrades in place to certified page:line. The contradiction card that cited a timestamp during the session now cites the transcript the way a brief would.
Clips follow the same logic. Cut video by time — or by transcript page and line, resolved against the certified transcript. Ask for 52:3 through 52:19 and you get the video of exactly that testimony, the way you'd designate it in a filing. Every clip comes from the verified session recording, carries an official designation, is hash-verified, and every export is logged for the case record. The clip a jury sees is provably the testimony the transcript records.
Off the record means off the record
One toggle takes the room off the record: transcription stops, the AI's access stops, and the toggle itself is audited. The control is fail-closed, and it extends to export — a clip that overlaps any off-record span is refused. Not watermarked, not flagged for review: refused. Anything said off the record can never leave the system as a clip, because the export path structurally will not produce it. In a tool holding testimony, "best effort" is not a privilege posture; refusal is.
It all files to the matter
When the session ends, the work product doesn't strand in a meetings app. File to Matter puts the recording, the certified transcript, the accepted insights, and every clip on the matter as searchable case records — next to the pleadings they support. Six months later, drafting for summary judgment, the admission is where the case lives: on the matter, cited to page and line, one search away.
Follow the accepted contradiction card from the Hale deposition through its whole life and the design becomes obvious. It was born live, citing the transcript sequence and media timestamp. It survived the lawyer's Accept decision. It upgraded to certified page:line when the reporter's transcript arrived. It's now a searchable record on the matter, and if the testimony ever needs to be shown rather than quoted, the clip cuts to exactly those lines — hash-verified, export logged. At no point did anyone copy anything between systems, and at every point the citation held.
That's the real payoff of one application with two registers. The everyday register means the firm's meetings already happen where the matters live — so when a meeting becomes evidence, there's no export, no transfer, no reconciling a video vendor's library against the case file. The proceeding was always in the right place. It just went on the record.
Try both registers in the live demo
Start in a client meeting, go on the record, accept a cited insight, and cut a clip by page and line — then try to clip the off-record span and watch it refuse.
Explore EdgeMeet →