Wang Jianjun
All projects

TRIX Companion · Across Devices

Use one conversation and focus companion across Web, iOS and desktop.

Three clients share account data and connect to agents through a native OpenClaw channel, with text, image and voice input.

View diagram (swipe sideways)
ONE COMPANION. THREE SCREENS.TRIX CompanionDesktop / WebiOSOpenClaw · Native channelARCHITECTURE ILLUSTRATIONWJ / SELECTED WORK
Project DevelopmentFeb – Sep 2026GitHub
ReactSwiftUIElectron
All 5 technologies
OpenClawSupabase

Key results

  • Web, iOS and desktop client implementations
  • Native OpenClaw channel with text, image and voice interaction

Product and contexts

TRIX combines conversation with focus spaces. The Web client supports quick access, iOS serves mobile contexts, and Electron provides a desktop interface. The public repository describes multimodal chat, focus timers and conversation history.

Connecting the clients

1

Device interfaces

React Web, SwiftUI iOS and Electron desktop handle interaction in their respective device environments.

2

Identity and shared data

Supabase provides authentication, database, realtime updates and storage for a shared service layer.

3

Agent access

REST / WebSocket pairing and communication connect clients to agents through a native OpenClaw Channel plugin.

4

Conversation and focus

Text, images and voice provide multimodal input, alongside focus rooms, timers and history.

Design considerations

A multi-client product must account for input differences, authentication state, pairing and message synchronization. Separating client responsibilities from the agent channel makes the repository architecture easier to follow.

Project references

This case is based on the public repository as of September 21, 2026. Dates begin with the public repository's creation month. Device illustrations show structure, not product screenshots or usage data. See TRIX_ap for code and setup instructions.

My work and design choices

I contributed to this cross-platform AI companion, connecting chat, focus sessions and agent access. The public repository contains Web, iOS, desktop and channel implementations; this case describes the system along those boundaries.

Adapt interactions to devices while sharing data services. Browsers, phones and desktops have different inputs and usage patterns. Each client keeps its interaction model, with shared authentication and data services. Cross-client consistency requires separate verification.

Give the agent connection its own channel. Pairing, connection state and messaging have a lifecycle of their own. Keeping this responsibility distinct from the interface makes it easier to investigate the client, shared data and channel in turn.

Record implementation and verification scope separately. The repository contains three client types and setup instructions. This case does not supply same-version regression results for cross-client message consistency, reconnection or physical devices. The device illustration explains structure; it is not evidence of those checks.