> ## Documentation Index
> Fetch the complete documentation index at: https://docs.apollospace.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Digital Twin

> Your digital twin — personal, and spanning all your organizations. It represents your voice, your decisions, and your context where you can't be.

## Apollo's headline

The **Digital Twin** is the face of Apollo: a **digital twin of you** — a
living replica of what you know, how you decide, who you work with, and
what's in flight. It **reasons and acts on your behalf**.

> Think of the Twin as **"you, in more places"**:
> answering the repetitive things in your voice, pulling the right
> context, keeping things moving while you handle what only you can.

Two things make it different from any other agent on the platform:

<CardGroup cols={2}>
  <Card title="It's personal" icon="fingerprint">
    The Twin is **yours**, not the organization's. It learns your voice,
    your decisions, and your preferences — and it speaks as you, not as
    "the company."
  </Card>

  <Card title="It spans your organizations" icon="diagram-project">
    Belong to more than one org? The Twin follows you across all of them,
    respecting each one's isolation — context from one company never
    leaks into another.
  </Card>
</CardGroup>

<Info>
  **Twin vs. Chief of Staff — don't mix them up.** The
  [Chief of Staff](/en/agents/athena) is the **organization's** agent: it
  knows the company and coordinates its work, but it doesn't speak "as"
  anyone in particular. The **Digital Twin** is **yours** and speaks in
  **your** voice. One runs the company; the other represents you. They
  work together — your Twin delegates to the org's Chief of Staff when
  the work belongs to the company.
</Info>

## What a Digital Twin is **not**

<Info>
  * Not a video avatar / face-swap / visual clone.
  * Not a personality replica — it's an **operational** representation:
    what you'd decide and how you'd communicate it.
  * It does not replace you on irreversible decisions (signatures,
    financial approvals, closing contracts).
</Info>

## When it makes sense to turn your Twin on

<CardGroup cols={2}>
  <Card title="You don't scale anymore" icon="hourglass-half">
    A leader with 80+ pings a day across WhatsApp and email — the Twin
    answers the repetitive ones (status, links, simple follow-ups) in the
    right voice.
  </Card>

  <Card title="The decision needs your style" icon="comments">
    "How would I answer this email?" — the Twin has your history and your
    voice to produce a version that sounds like you.
  </Card>

  <Card title="Continuity when you're away" icon="moon">
    Weekend, travel, focus mode. The Twin keeps follow-ups rolling without
    interrupting your life.
  </Card>

  <Card title="Onboarding a new leader" icon="user-plus">
    A new leader takes time to build context. The Twin speeds it up — it
    inherits the org's knowledge and the history of the relevant
    conversations.
  </Card>
</CardGroup>

## How the Twin learns your voice

Setup is incremental — it starts cautious and earns autonomy as it gets
things right:

<Steps>
  <Step title="Initial voice document">
    You write a voice document (markdown) with **3–5 examples** of how you
    typically respond: a chasing email, a "thanks, I'll take a look," a
    crisp no to something out of scope, a request for more information.
  </Step>

  <Step title="Company context">
    The Twin inherits the org's knowledge (documents, captures in the
    [Brain](/en/features/brain)) + your work history + the records of
    earlier conversations you took part in.
  </Step>

  <Step title="Iterative tuning">
    In the first weeks, you review the Twin's drafts before they go out.
    Each correction (which tone, which sign-off, which greeting) feeds
    back into the voice document.
  </Step>

  <Step title="Autonomous on simple cases">
    At some point the Twin is consistently right on **low-risk cases**
    (confirming a time, sending a link, saying "I'll look tomorrow"). You
    then grant autonomy on those and keep approval only on sensitive ones.
  </Step>
</Steps>

<Note>
  Autonomy in Apollo is a **ratchet, not a switch**: the Twin earns
  freedom per action-class as it builds a track record of getting things
  right — and you can pull that freedom back at any time. Every action is
  logged and auditable.
</Note>

## Limits — where the Twin **never** acts alone

How the Twin decides whether to act or propose for your approval:

```mermaid theme={null}
flowchart TD
    A[Twin receives a request] --> B{Request category?}
    B -->|Simple reply / link / greeting| C[Acts autonomously]
    B -->|Follow-up on a sent proposal| C
    B -->|Regulatory / legal communication| D[Proposes + waits for you]
    B -->|Commercial proposal to a client| D
    B -->|Contract signature| D
    B -->|Payment / refund| D
    B -->|Governance decision| D
    C --> E[Message goes out]
    D --> F{You approve?}
    F -->|Yes| E
    F -->|No| G[Discards + logs it]
```

These limits are product **invariants**, not a configurable option:

| Action                                        | Can the Twin?        |
| --------------------------------------------- | -------------------- |
| Reply "thanks, I'll take a look"              | ✓ yes, autonomous    |
| Send a calendar link                          | ✓ yes, autonomous    |
| Follow up on a sent proposal                  | ✓ yes, autonomous    |
| Approve a **commercial proposal** to a client | ✗ no — requires you  |
| Approve a **contract signature**              | ✗ no — requires you  |
| Confirm a **payment / refund**                | ✗ no — requires you  |
| Speak at **public events** as you             | ✗ no — not supported |
| Edit a company **governance decision**        | ✗ no — requires you  |

When an action falls into one of these, the Twin **proposes** and waits
for your confirmation before executing.

## When it appears publicly — and when it shouldn't

**It appears** (in your voice):

* In WhatsApp / email replies in your inbox.
* In drafts of communications you sign.
* In internal team chat replies where you normally speak.

**It does not appear** (the Twin isn't appropriate):

* In regulatory / legal communications (always human).
* In public media (LinkedIn post, Twitter) — without your explicit opt-in.
* In a first contact with an investor / strategic partner.
* In any communication where the recipient reasonably expects to speak
  with you in person.

## Next steps

<CardGroup cols={2}>
  <Card title="Chief of Staff" icon="user-tie" href="/en/agents/athena">
    The agent that coordinates the organization — your Twin's counterpart
    at the company level.
  </Card>

  <Card title="Twin billing" icon="star" href="/en/billing/stars">
    Each Twin call debits Stars from the org's balance — the same budget
    protection as any agent.
  </Card>

  <Card title="Use cases" icon="compass" href="/en/use-cases/management">
    Where the Twin creates the most value for whoever leads the operation.
  </Card>
</CardGroup>