What we’re building, in order.

The flagship: enroll a voiceprint once, recognize the same speaker across every future recording. Consent-gated, deletable, BIPA-aligned.

"Hi, I’m Sarah from Pinterest" → cluster auto-renamed to Sarah. Moderator cues honored. No more renaming Speaker 3 thirty times per file.

Auto-delete in 30 days. Voiceprints opt-in. No model training on your audio. SOC 2 Type II audit kickoff in Q3.

Real Buy-now flow on /pricing, replacing today’s email-capture. Tiered plans live with the anti-bill-shock guarantee codified in /refunds.

Speaker-attributed exports preserved across formats. Round-trip-safe — names you trained survive the export.

English first, then Spanish, Portuguese, French, German, Japanese. Mid-sentence code-switching handled natively.

"Show me everything Sarah said about pricing across all 47 calls." The compounding moat once a customer has 100+ hours of voice IDs in the system.

Hybrid local + cloud. Sensitive files stay on your machine; collaboration files sync to cloud. Therapists and journalists especially.

For teams that want speaker-aware transcription as infrastructure. Compete with AssemblyAI / Deepgram on the diarization dimension specifically.

Vendor-side BAAs with every subprocessor that touches PHI. Once complete, BAA on Team plan moves from "on request" to default-included.

Full homepage with side-by-side transcript demo, comparison table, pricing, FAQ. Blog with 12 deep-dive guides on speaker ID, privacy, and competitor comparisons.

Terms, Privacy, Refund, AUP, Security, DPA, BAA, Trust Center, DMCA, Status, Changelog, Press, Roadmap. Cookie consent, JSON-LD, operating-entity disclosure.