Evidence search

Search thousands of hours of recordings like Google.

Every answer links to the exact moment in the original recording — file, speaker, timestamp, play button.

Ask, don't scrubsemantic, not keyword luck
Provenanceevery result plays the tape
Private cloudyour recordings never train a model

The idea

A search result here isn't a claim. It's a moment.

"According to your recordings, X happened" is a summary. "X happened — hear the 23-second evidence" is proof. Every result keeps source file, speaker, timestamp, transcript and playable original together. That provenance is the whole point.

Corpus Search Evidence Timestamp Original audio

Who this is for

People whose audio is evidence, not content.

What's under it

Live today. Not a deck.

Upload or point us at a folder — everything below the Early-access chips works on your recordings right now.

Ask across every recording LIVE

Semantic search over your whole library — meaning, not just keywords. Every hit carries its timestamp.

Speaker-attributed transcripts LIVE

Diarized, timestamped, editable. The edited version is what search reads.

Jump to the second LIVE

Every result opens the original recording at the quoted moment. Hear it, don't trust a summary of it.

Evidence clips LIVE

Cut the 30 seconds that matter and export it — captioned, shareable.

Citations & exports LIVE

Copy quote + file + timestamp in one click. TXT, SRT, VTT, DOCX, bulk ZIP per folder, export to Drive.

Cross-file clustering & contradiction timelines EARLY ACCESS

Themes counted across hundreds of files; conflicting statements laid out on a timeline. Rolling out with design partners — ask us.

🔒 Private cloud — recordings never train any model 🗑 Delete anytime — you own every file 🧾 Provenance on every result — file, speaker, second 🌐 100+ languages, auto-detected

Drop in your recordings.

Ask the question you'd ask a colleague who heard every minute — the first files are free.

Upload your first recordings