Trigger and glance
Press, speak, done. A job starts and comes back as a summary short enough to read on your wrist.
Ceviz — Apple Watch & iPhone
Voice tasks for OpenClaw, from your wrist.
Start a job by speaking to your watch, see a short summary at a glance, and continue on your phone when the work needs depth. Ceviz is a local-first, Watch-first layer for this focused workflow—not a replacement for the full OpenClaw mobile client.
Status: open beta — anyone with an iPhone and an Apple Watch can join through TestFlight.
How it works
Each surface keeps to what it is good at.
Depth and approval
The full report: status, analysis, the subagents and tools that ran, and the next action worth taking.
The actual work
OpenClaw keeps executing where it already lives — on your own computer, over your own private network.
Short on the watch, deep on the phone, real work in the background. Long logs and recovery steps never get pushed onto a screen too small for them.
Install the backend
Run Ceviz alongside OpenClaw on your own machine.
Ceviz is open source. Install the bridge on the machine where OpenClaw runs, connect privately with Tailscale, then pair your iPhone by scanning a QR code.
New installation only · Python 3.11+
git clone --single-branch \
--branch ceviz-helper-v2026.9.12-beta.1 \
https://github.com/MertBasar0/ceviz.git
cd ceviz
bash deploy/install.sh- InstallRun the commands on your OpenClaw machine.
- ConnectChoose Tailscale — recommended — or another private connection.
- PairScan the generated QR code in Ceviz Settings and test the connection.
- DiagnoseRun
bash deploy/doctor.shif setup or delivery fails; its shareable report hides credentials and command contents.
Already using Ceviz? Updating TestFlight does not update your backend. Follow the existing-backend guide for a checked Linux/WSL user-service update that keeps pairing, settings and job records. Do not rerun the installer over a working setup; macOS and custom service installations need assisted updates.
Good to know: the first voice command may take longer while the local transcription model loads. The pairing QR contains a credential; keep it private and regenerate it if exposed.
Use cases
Built for the moments you are away from the keyboard.
Deploy / CI check
Ask whether the release is healthy, see the verdict on your wrist, and open the phone when logs or a rollback decision are needed.
Pull request snapshot
Get a compact review state on the watch, then read the structured report — files, findings, verification — on your phone.
Incident triage
Check severity and what your agent is already doing about it, then continue the diagnosis with the full report in hand.
Where your data lives
Commands run on your machine.
Ceviz has no hosted command-processing backend or analytics SDK. Your voice command goes from the watch to your phone and on to a backend you run yourself, reachable only over your own private network. Speech is transcribed locally on that machine by default. A minimal hosted relay forwards concise completion notifications to Apple Push Notification service; it never receives voice audio, the full transcript/report, or your backend token. Read the privacy policy.