A privacy-first AI system that transforms video recordings into interactive digital avatars — processing locally on-device, then enriching with cloud AI for personality, voice, and face profiles that family members can converse with.
When someone passes away or becomes unavailable, all that remains are static photos and videos — no way to interact, ask questions, or hear their perspective on new events.
The user records themselves via the browser. The app captures video and audio, showing a live webcam preview with recording controls.
The key differentiator: all initial AI processing runs locally via Ollama. No raw video ever leaves the user's machine.
Only lightweight JSON analysis results go to the cloud. Google Cloud APIs add deeper analysis: transcription, facial landmarks, sentiment, and voice synthesis.
Local and cloud analysis merge into a comprehensive avatar profile — personality, voice, face, and memory — stored in Firestore.
Family members chat with the avatar. Responses match the person's authentic style and tone, with multi-provider LLM fallback and optional voice synthesis.
A privacy-first AI platform combining on-device processing with cloud enrichment to create interactive digital avatars.
On-device AI processing meets cloud enrichment — creating interactive avatars that preserve personality, voice, and memory.