Voice-to-Print Pipeline
Speak a part description and get a ready-to-slice 3D model through automated speech capture, geometry generation, and a direct handoff to the Bambu X1C.
The Flagship Builds
Two products, one thesis
Your context is the asset. Synapstack OS captures it. Kyara acts on it. Neither one asks you to hand your life to someone else's server.
Live · iPhone + Web + MCP
Synapstack OS
A personal journaling system with an API. Capture on iPhone, review on the web, and let an AI assistant read and write the same journal through a remote MCP server — no copy-paste, no export step.
- Native iPhone capture — Expo/React Native app wrapping Apple's Journaling Suggestions picker
- Edge backend — Cloudflare Workers on D1 and R2, sync engine built in
- Remote MCP server — OAuth over Streamable HTTP; Claude reads and writes entries directly
- No delete tool, by design — a deliberate blast-radius limit on anything an agent can touch
Synapstack OS — timeline
assets/images/synapstack-hero.png
Beta · Windows desktop · iOS in progress
Kyara
An always-on desktop AI companion. One hotkey brings her up over whatever you're doing — voice in, voice out, with a 3D avatar, persistent memory, and hands on your actual tools.
- Realtime voice — push-to-talk pipeline with barge-in, not a chat box with a mic button
- VRM avatar — three-vrm rendering with viseme lipsync, custom outfits and motions
- Persistent memory — local SQLite, rolling summarization, three swappable personas
- Real integrations — Gmail, Omi wearable feed, screen capture, web and X search
Kyara — desktop overlay
assets/images/kyara-hero.png