← All workflow demos

Say it. Send it. Watch your agent act.

Workflow film · 1:25

Say it. Send it. Watch your agent act.

Full instructions →

Real app recording · Original AI voice and presenter portrait · Captions included · Cuts and playback speed shown in the film. Narration uses Code Overlord’s theatrical English.

Read the transcript

Choose your agent on the phone. I am using Claude in Ember Launch. The phone is already paired; waits are cut.

In the shared notepad, add Launch checks with two tasks: check downloads and review captions.

The app turns speech into an editable draft. I review it and correct caption to captions before sending.

Here is the corrected instruction. Tap Send to deliver it to that agent on the desktop.

The instruction arrives in the same Claude conversation inside Code Overlord. The agent starts updating the shared notepad.

Watch Launch Checks appear with Check downloads and Review captions beneath it. The agent has made the actual change beside the conversation.

Open the notepad on the phone. Both tasks are already there.

Back at the desktop, focus the agent input and click the microphone.

In the shared notepad, mark check downloads complete. Leave review captions open.

Desktop dictation also gives you a transcript to review. Submit it to the same agent.

Check downloads is now completed. Review captions stays open. The result and the updated list are both visible inside Code Overlord.

The phone shows the same result. Expand completed tasks: Check downloads is done, and Review captions is still waiting.

Dictate on your phone, review the words and send them to your agent. Continue by voice on the desktop.

Try this workflow yourself

Pair your phone with the desktop workspace, allow microphone access and finish speech-model setup on both devices. Enable notepad sync and keep the selected agent reachable.

In the shared notepad, add Launch checks with two tasks: check downloads and review captions.

Dictation creates editable text. Review it before using Send on your phone or submitting it from the agent input on your desktop.