Skip to content
PawOS Docs

Wake Word

Push-to-talk is the real, primary input method.

Not verified

This documentation could not confirm always-listening wake-word activation as a currently shipped, verified capability. Push-to-talk (a keyboard shortcut, configurable in Settings) is the confirmed, primary voice-input method — use that unless you have separately verified wake-word support in your build.

Product disclosure

Implemented

This page documents the current PawOS behavior for Wake Word. It is intended to disclose what users can see and use, what permissions are required, what evidence is recorded, and what limitations remain.

This disclosure is part of the PawOS documentation for Wake Word. It is written to make the product behavior understandable before a user relies on it, pays for it, connects an account, approves a plan, or authorizes a machine-affecting action.

What users see

  • The companion appears as a desktop presence with voice/text input, optional speech output, avatar/profile settings, and task/task-card integration.
  • Users can configure personality, voice behavior, companion package data, and in supported paths upload a model asset rather than generate one.
  • The companion should make active listening, draft review, speech output, and task execution state understandable.

What PawOS does

  • The companion is a renderer-side experience that submits user requests into the same task/action pipeline as typed input.
  • Speech recognition keeps recognized text reviewable before send; speech output should read user-facing summaries rather than raw command logs.
  • Profile/package data can describe the companion appearance and behavior, but execution remains governed by the same runtime and permission model.

Permissions and boundaries

  • The companion is not a separate permission system and cannot bypass action authorization.
  • A custom avatar or personality does not change billing, connector, filesystem, command, or security behavior.
  • Wake-word and ambient behavior must be documented according to what is actually implemented, not implied by branding.

Evidence and records

  • Tasks started through the companion should leave the same Work Records as tasks started through typing.
  • If a voice or avatar capability is not implemented or not verified, the docs should say so directly.

Limitations and user responsibility

  • AI-generated avatars from a photo are not implemented; uploading an existing compatible model is the documented working path.
  • Push-to-talk/review-before-send is the reliable input model unless a specific wake-word implementation is verified.

Warning

PawOS can assist with planning, coding, automation, connectors, billing flows, and system operations, but the user remains responsible for reviewing plans, confirmations, diffs, command effects, connector side effects, billing actions, and final outputs before relying on them.