Core capabilities
Voice
0.5s transcription. Every word stored as meaning, not just text.
Vision
2MP OV2640 with on-command capture. Processes scenes in under 2s.
Memory
Vector-based retrieval. Search your life by context and meaning.
Dual-core processing
Continuous streaming
Wearable all day
3D printable case
Wireless streaming & charging
Open Architecture
Aura is a manifesto for the open-source hardware movement. Firmware, schematics, and the entire AI pipeline are completely public. MIT licensed.
import AuraCore let device = Aura(hardware: "ESP32-S3") await device.configure { config in config.wakeWord = "Hey Aura" config.visionEnabled = true } await device.startListening()
Ask anything. Aura surfaces what you've heard, seen, and said — with full context, instantly.
Optimized for high-contrast indoor and outdoor visual recognition.
Direct integration with advanced vision models for near-instant analysis.
Find that document you looked at 3 hours ago by simply describing it.
80 grams. Worn all day.
Power for hours in a radically smaller footprint. A high-density Li-Po cell optimized for minimal volume. Paired with our ultra-low-power ESP32-S3 architecture, it delivers all-day ambient awareness.
Up to
4 hours
of active listening time
on a single charge
Capacity
350 mAh
high-density Li-Po
battery cell
Charging
USB-C
fast charging everywhere
you go
Beyond laptop. Beyond phone.
Not a computer
Not a distraction
A presence
Conversation history.
Memory timeline.
Device control.
Low power architecture optimized for Wi-Fi and Bluetooth LE connectivity with integrated power management.
3.7V 350mAh cell designed for maximum energy density in minimal physical footprint.
Open source
Inexpensive
Custom app
Self-hostable
Great for students
Fork the repo, order the components, and build your own aware wearable. Full guides included.
Aura's 2MP wide-angle camera and dual MEMS microphones capture your conversations, environment, and visual context continuously — processing everything on-device with no raw data ever leaving your pocket.
Everything is processed locally on the XIAO ESP32-S3. Audio is transcribed on-device via Whisper. Only anonymised semantic embeddings sync to the cloud. Your raw conversations are never uploaded or stored.
Aura achieves sub-200ms response latency for context retrieval. Queries are processed via a RAG pipeline combining Pinecone vector search and Redis caching — designed to feel instant.
Yes. Aura is fully open-source. The hardware BOM costs ~$50, firmware runs on FreeRTOS, and the full build guide is in the Docs. No soldering experience required for the base build.
Aura pairs with the companion Flutter app on both iOS and Android. The app handles voice queries, memory browsing, and settings sync via BLE. No proprietary ecosystem required.
Aura runs all day on a single charge — designed for 8+ hours of active use. It charges via USB-C in under 45 minutes and enters a low-power standby mode when not actively capturing.