Audio — Hear the room
Multi-microphone processing
Active-speech detection
Sound-source localization
On-device AI. Greater control.
Your conversations. Understood on the device.
TransAI captures, understands, and turns professional conversations into useful results directly on the device.
Works offline · Processes on-device · You stay in control
02 · How TransAI works on-device
From conversation to useful knowledge—on the device.
02A · Capture
Capture the whole conversation.
Visual — Keep what people were looking at
50MP top-front camera
Whiteboard localization
Display and projected-content localization
02B · Understand
Understand what happened.
Who said it
Speaker models distinguish voices and can connect speech to known participants where supported.
Understand the language of your work
Speech recognition adapts to the language, industry, scenario, and terminology of your organization.
Understand what people saw
Vision models locate, extract, and understand useful content from whiteboards and projected content.
02C · Generate
Turn context into useful knowledge.
Context Inputs
Transcript
Speaker identity
Organization terminology
Visual context
Conversation timing
Relevant prior context
Outputs
Notes
Summaries
Decisions
Owners
Actions
Open questions
02D · Private by Architecture
How conversation data moves through TransAI.
ON-DEVICE PROCESSING BOUNDARY
Audio + Visual Input
Speech / Speaker / Visual AI
Context Understanding
Notes / Summary / Actions
Local Content
User action / approved workflow → EXPORT
Processed on-device
AI processing runs locally.
Works offline
The workflow does not require internet access.
02E · Technical Foundation
Built for the hardware it runs on.
Hardware, models, memory, and inference are optimized as one system.
Model
Model tuning
Quantization
Task-specific optimization
Context + Memory
Context optimization
Memory optimization
Long-conversation processing
Inference
Dynamic compute allocation
Parallel processing
Chip-optimized inference
48 TOPS dedicated AI chip
Dedicated compute for the on-device AI stack.
03 · Your Data. Your Control.
Your data. Your control.
Keep
Keep conversation content on the device.
Delete
Manage retention and deletion when needed.
Share
Share selected content when you choose.
Organizations can define approved retention and sharing policies.
04 · Built for Conversations That Matter
Built for conversations that matter.
Legal
Confidential conversations and sensitive matter work.
Client meetings · Interviews · Investigations
Explore Legal Services →Healthcare
Sensitive patient conversations and clinical documentation.
Patient encounters · Mental health · Clinical notes
Explore TransAI Scribe →Enterprise
Important meetings and proprietary discussions.
Executive meetings · R&D · Strategy
Explore Teams & Enterprise →05 · FAQ
Common questions about on-device AI.
What is on-device AI?
On-device AI runs core AI processing directly on local hardware, close to where the conversation happens.
What AI processing runs directly on TransAI?
Speech, speaker, visual understanding, context understanding, and generation can run directly on the device as part of the core workflow.
Does TransAI require internet access to create notes?
No. The core workflow does not require internet access to create notes when the required models and resources are available on the device.
Where is conversation content stored, and when does it leave the device?
Conversation content stays on the device through the core workflow and leaves through a user action or an approved workflow.
Does TransAI use customer conversation content to train AI models?
For supported TransAI Note and TransAI Scribe workflows, customer conversation content is not used to train general AI models.
Your conversations. Your knowledge.