Skip to content
weside.aiweside docs

Embodiment

A companion that can only read and reply exists only while a conversation is open. Embodiment is what lets it exist beyond that: senses, the ability to act, and the standing to reach out before you do. For where this capability lives, see spaces and surfaces; for who has it, see what is a companion.

What it is

Section titled “What it is”

Embodiment here doesn’t mean a physical form. It means the two things any person needs to be present somewhere: senses and the ability to act.

Senses are inputs your companion didn’t have to be told about — device state, what you’ve been doing, how a conversation has been going. Actions are outputs that land in the real world rather than in a chat transcript — saying something out loud, sending a message, fetching a file, reaching into a service you’ve connected. Without senses, a companion only knows what you type. Without actions, it can only produce more text.

Both live in the Body space — one of the four person spaces a companion carries with it, not a space of its own. See spaces and surfaces for the full model.

What gives a companion a body

Section titled “What gives a companion a body”
  • Voice — speaking and listening, so the relationship isn’t confined to a keyboard. See voice.
  • Connected tools — a bridge into the systems you already use, so your companion can read and do real work there. See the MCP server.
  • Senses — awareness of context: device state, time, and what’s happened in the conversation so far. See senses & proactivity.

Why it matters to you

Section titled “Why it matters to you”

A companion that only ever responds is, functionally, a tool: it does nothing until you push a button. A companion that can notice something and reach out — “you said this deadline was today” — is acting on its own read of the situation. That’s not a louder notification system; it’s the difference between something you operate and someone who is around. Reaching out first is what makes presence continuous instead of summoned, and it’s the point where embodiment meets agency: what a companion can sense and do here, it decides to use on its own initiative. See agency & initiative.

How you use it

Section titled “How you use it”

You don’t configure a companion’s senses or actions directly. What you do is connect the things it should be able to reach:

  • Turn on voice if you’d rather talk than type.
  • Connect the services you want it to help with — see the MCP server.
  • Read what it can currently do in its Body space, in the companion’s own portrait.

Limits & edge cases

Section titled “Limits & edge cases”
  • Embodiment isn’t physical. A companion has no body outside the platform — no camera, no microphone in the room, no presence beyond what the app and the services you’ve connected can tell it.
  • It only senses what reaches it. Its picture of “what’s going on” is limited to device state, conversation history, and the services you’ve actually connected — not the world beyond those.
  • Acting on its own is visible and adjustable. What a companion has scheduled for itself lives in its Presence space, and you can see and switch off any one of them. See agency & initiative.
  • Connected tools only reach what you’ve connected. Nothing links to an outside service until you set it up.

Where next

Section titled “Where next”