Deep Dive

Built for data protection.
Engineered for insurance accuracy.

Every feature is designed to keep claimant data on your device while delivering accurate insurance documentation.

Data protection architecture

Claimant data stays on your device at every stage of processing. No cloud. No exposure. No risk.

Microphone Local capture
Insurance AI Engine 7,000+ terms + on-device
Claims Document Your application
Local DB On-device storage
Insurance AI Engine

7,000+ insurance terms. Advanced AI. Advanced hallucination removal.

Our on-device processing stack is optimized for insurance accuracy with voice activity detection, model inference, hallucination stripping, insurance dictionary injection, snippet expansion, and AI grammar polish.

  • 7,000+ built-in insurance terms — policy types, coverage terms, claims terminology, medical-legal crossover, regulatory terms
  • AI grammar polish — available on macOS with on-device AI
  • Hallucination stripping removes phantom text from silent segments
  • Optional filler word removal for cleaner output
  • Custom vocabulary biasing and post-processing corrections
  • Text shortcuts — say "slash claim" or type /claim to expand a full claim report template with 4 pre-loaded insurance shortcuts
  • Claims documentation accuracy — correctly transcribes policy references, claim numbers, and coverage terminology

On-Device Processing Stack

VAD Detection
Model Inference
Hallucination Strip
Insurance Dict Inject
Snippet Expand
AI Polish
Desktop App

Live-typing into every claims system

VoicePrivate — Insurance Edition types transcribed text directly into whatever app has focus — Guidewire, Duck Creek, Applied Epic, or any macOS application. No copy-paste needed.

  • Global hotkey — start transcribing from any claims app instantly
  • Intelligent silence detection — auto-stops when you pause for thought
  • Transcription history — searchable archive of all your dictation
  • File import — transcribe recorded statements, meeting audio, voice memos
0:00
Data Protection

Tamper-evident audit log. Proof policyholder data stayed local.

Zero-trust transcription, verified. Every transcription is recorded in a cryptographic audit trail showing that all policyholder audio was processed on-device with zero outbound network calls. Each entry is chained with SHA-256 checksums — if any record is modified or deleted, the chain breaks. Export the full log as JSON for internal compliance reviews.

  • SHA-256 hash chain — tamper-evident log proves no policyholder data left the device
  • Network assertion — every entry records "no_outbound" status, protecting PII and claims data
  • Compliance export — one-click JSON export for internal compliance reviews
  • No vendor risk — claims conversations never touch a cloud server or third-party processor
  • Chain verification — built-in integrity check confirms no entries were altered
  • Encrypted backup & restore — create encrypted backups of your transcription database and restore on any device, keeping policyholder data safe and portable

Audit Log Entry

timestamp: 2026-03-09T14:32:07Z
event: transcription_complete
engine: on-device-engine
app_version: 1.6.0
os_version: macOS 15.3
source: microphone
duration: 45.2s
words: 312
network: no_outbound
checksum: a1b2c3d4e5f6...
Chain verified — policyholder data stayed local

Experience the difference

Start transcribing with complete data protection today.