33,000+
Calls transcribed
~7,900
Hours of calls
How a Legal Research Firm Reviews Every Call With AI
Legal · Call scoring & coaching
Read the case study →
Case study · Legal · A legal technology company
How a legal technology company launches white-label depositions with automatic recording and transcripts
A legal technology company built its deposition product on Speak AI, using a meeting assistant to record video-call depositions and return speaker-labelled transcripts to its own platform, without building recording infrastructure itself.
As of February 2026Measured from Speak usage records
-
2,900+Recordings captured
-
About 4,500Hours of testimony recorded
-
Almost allCaptured automatically by a meeting assistant
-
About 90 minutesAverage recording length
Summary
The company sells a deposition product to legal teams. Depositions increasingly take place on video calls, and each one needs a recording, a transcript with speakers labelled, and a handoff into the firm's own workflow.
It put Speak AI behind its own brand. A meeting assistant joins scheduled sessions on its own, Speak returns a transcript with timecodes and speaker labels, and webhooks tell the company's backend when each stage is done.
Close to three thousand depositions flowed through the account, almost all captured by the meeting assistant. The account's last recording was in early 2026, so the figures below are shown as of that month.
Three lines to take away
Before
A deposition product meant building and running recording, transcription and delivery infrastructure before a first customer could use it.
With Speak
A meeting assistant records the session, Speak transcribes and labels speakers, and webhooks hand each stage to the company's own backend.
After
Thousands of hours of testimony were captured and delivered through a product that carried the company's brand, not ours.
Building the plumbing was the first hurdle
Before a single deposition could be sold, the company would have needed its own recording, transcription and delivery stack.
Infrastructure to build
Capturing a video-call session, storing the recording and processing it at volume is a project of its own, before any legal feature is written.
Transcripts lawyers can use
Testimony needs clean speaker labels and timecodes. A rough transcript that cannot say who spoke when is of little use for review.
Manual scheduling and handoff
Without automation, someone has to schedule each session, confirm it was recorded and send the files on to the right matter.
The cost: The old way meant engineering time on infrastructure instead of on the product, and a slower route to a first usable release.
The turning point
Buy the capture layer, build the product
The turning point was a simple question: is recording and transcription where this company wanted to compete? It was not. Its value sat in how depositions are scheduled, delivered and reviewed for legal teams.
Two things had to be settled first. The recording and transcripts had to carry the company's brand end to end, so customers never saw another vendor. And the work had to connect to its own backend through an API and webhooks, so no one handled files by hand. Once both were confirmed, the decision moved from build to build on top of Speak.
How they use Speak AI
The flow runs from a scheduled session to a transcript in the company's own system.
- Step 1
Join sessions
Sessions are scheduled in advance, and the meeting assistant joins the video call on its own, with no one needing to press record.
- Step 2
Transcribe
Speak returns a transcript with timecodes in minutes, with each speaker labelled in a consistent format.
- Step 3
Notify backend
Webhooks fire at ingest, processing and completion, so the company's systems start each downstream step without a manual handoff.
- Step 4
Deliver
Transcripts export into downstream legal tools, and branded viewing pages give legal teams controlled access to the testimony.
Honest friction: Connecting a product to Speak through an API and webhooks takes engineering time, and speaker labels on multi-party sessions still benefit from a review pass. Usage in this account has been quiet since early 2026.
Results
Thousands of hours of testimony captured
The account recorded close to three thousand sessions, averaging a little over an hour and a half each.
Under 100 minutes
Average length of a recorded deposition
2,900+
Transcripts in the account
Almost 3 years
Of captured sessions in the account
What the team found, in short
Paraphrased from our own conversations with the team. Not quotes.
For a legal technology company, recording and transcription is plumbing, and teams that buy the plumbing spend their time on what customers pay for.
Paraphrased from our calls with the team
Branded scheduling, recording and review pages matter as much as accuracy, because the customer should never have to know another vendor is involved.
Paraphrased from our calls with the team
Webhooks at each stage turn a pile of files into a pipeline, so no one has to check whether a session was recorded.
Paraphrased from our calls with the team
What is next
- Branded librariesTeams like this often give each law firm its own branded library with fine-grained access.
- Matter-level searchChat across a folder lets reviewers ask questions across many sessions for one matter.
- Summaries on demandPrompts can draft a summary of each session so reviewers start from a structured page.
More customer stories
5,600+
Recordings processed
14
Branded recorders configured
How a Research Agency Launches a White-Label Platform With Speak
Research · White-label platforms
Read the case study →
20,000+
Recordings processed
~2,700
Hours of audio and video
How a Consulting Firm Scores Every Coaching Session With AI
Professional services · Call scoring & coaching
Read the case study →
Customer name withheld at their request. Figures are from Speak usage records, frozen as of February 2026.
Building a product on recorded conversations?
Talk to us about white-label recording, transcription and webhooks for your platform.