01
Principle
Local-first
Inference, storage and computation default to the device. The cloud is opt-in, never required.
On-Device AI · Privacy by DesignIntelligence that lives
Intelligence that lives
on your device.
OmniFlux AI builds a family of mobile applications powered by a unified on-device inference engine. Zero servers. Zero tracking. Zero compromises.
- 0 byteData Uploaded
- Offline-firstNetwork Required
- 5+Products Roadmap
- Flux EngineUnified Engine
Product Lineup
A family of focused apps, powered by one private engine.
Each app is purpose-built for a specific human moment, sharing the same uncompromising privacy guarantee.
Flagship
WhisperFlux
On-device voice transcription, speaker diarization & smart summaries.
- Boardroom meetings
- Private conversations
- No-network discussions
- Confidential briefings
Learn More/whisperflux
In Development
VisionFlux
Multimodal visual understanding — entirely on your device.
Learn More/visionflux
Flagship · Coming Soon
WhisperFlux
Speech-to-text, speaker diarization & smart summaries — entirely on your phone.
- Boardroom meetings
- Private conversations
- No-network discussions
- Confidential briefings
9:41
WhisperFlux
New Recording
Live · On-device
00:42
Realtime transcript3 speakers
Alex
So if we shipped this offline first...
Mei
The recorder runs in the background.
Sam
And the model never leaves the device
All three stages happen on-device — your audio never leaves the phone.
Technology
Flux Engine the runtime that powers it all.
A high-performance on-device inference runtime, optimized across CPU / GPU / NPU. One engine, every product.
Layer 5
Apps
Layer 4
Modules
Layer 3
Flux Engine
Layer 2
Runtimes
Layer 1
Silicon
Local Inference
Quantized models run on CPU, GPU and NPU with adaptive scheduling, delivering desktop-class speed on mobile silicon.
Zero Telemetry
No analytics, no crash beacons, no account system. The app physically cannot phone home.
Open Models
GGUF, MLC, MNN — pick from leading open-source models or bring your own.
Modular SDK
Speech, Vision, NLP modules share a unified pipeline — features ship faster across products.
Compatible models
Whisper.cppGGUFMLC-LLMMNNONNX RuntimeCoreMLTFLitePaddleNLPLlama-classQwen-classPhi-classGemma-class
Our Vision
Personal AI should belong to the person.
We believe the next generation of AI should empower individuals without surveilling them. We're building it, one focused app at a time.
02
Principle
Privacy by design
We architect away the temptation to collect. No accounts. No telemetry. No backdoors.
03
Principle
Open by default
We embrace open weights, open formats and transparent claims. Audit us, anytime.
Journal
View all postsNotes from the lab
Engineering notes, product stories and ideas from the team.
2026-04-228 min read
Why we bet the company on on-device AI
Cloud AI is a transitional architecture. Here's why we believe the next decade of useful AI lives in your pocket — and what it means for the products we are building.
Read more
2026-04-089 min read
Designing Flux Engine: one runtime for every product
A peek at the architecture choices behind our shared on-device inference runtime — and the constraints that shaped them.
Read more
2026-03-257 min read
WhisperFlux preview: speech, speakers, summaries — all local
Our flagship enters internal beta. A walkthrough of the streaming pipeline and what we learned shipping ASR on real phones.
Read more
FAQ
Frequently asked questions
Cloud AI requires sending sensitive content (voice, photos, conversations) to remote servers. On-device AI removes that risk entirely — your data physically stays on the hardware you own.
Be the first to try what ships next.
Join our launch list for product releases, technical deep-dives and early access.