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

# Agent Inbox

> A ranked feed of what needs your attention today, with replies already drafted

The **Agent Inbox** is the main Home screen. It turns a noisy inbox into a short, ordered to-do list: the threads
worth your time, ranked with the most time-sensitive first, each with a drafted reply waiting.

## What you see

| Section             | What it is                                                                                                                                            |
| ------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Needs attention** | The threads Emanate thinks you should act on today, newest and most urgent first. Each card shows who it's from, what they want, and why it surfaced. |
| **Tasks**           | The specific things to do, staged from the feed: reply to this email, quote this request, follow up on that quiet thread. Clear them as you go.       |
| **Recent accounts** | A quick-glance list of the companies you've been emailing with, so context is one click away.                                                         |

Every card explains *why it's here*: "waiting 3 days for a reply," "looks like a quote request," "deal going quiet."
Nothing is a black box.

## Acting on a task

Open a task and you have everything in one place, no jumping to your email client.

<Steps>
  <Step title="Read the thread">
    The original email is right there, with the earlier back-and-forth collapsed so you're not scrolling.
  </Step>

  <Step title="Review the draft">
    For the emails most likely to need a reply, Emanate has already written one in your voice. It's grounded in the thread, your knowledge base, and how you've handled similar mail before.
  </Step>

  <Step title="Edit in place">
    Adjust the draft right in the panel. There's no separate editor to open.
  </Step>

  <Step title="Send, stage, or act">
    **Send now**, **stage** it as a draft in your real inbox for later, or take another action: schedule a follow-up, add the contact to a sequence, or open the full editor. When you're done, the task moves to **Done** (with an **Undo** if you clicked too fast).
  </Step>
</Steps>

### Dismissing

If a surfaced thread doesn't need you, **dismiss** it. It drops off the feed, and Emanate takes note, so it learns
what you don't want to see and stops resurfacing that kind of thing.

## How your digital twin works

Behind the Agent Inbox is your **digital twin**: an AI model of how *you* work. It's what makes the ranking feel
like your own judgment instead of a generic filter.

Each weekday morning, the twin runs through a simple loop:

<Steps>
  <Step title="Takes a fresh look at your inbox">
    It reads a snapshot of your most recent mail.
  </Step>

  <Step title="Reasons like you would">
    It weighs each thread against what it knows about your priorities, your past actions, and your writing style.
  </Step>

  <Step title="Ranks what's urgent">
    Every thread gets scored two ways: **how time-sensitive it is** (how long it's been waiting, whether someone's blocked on you) and **how likely you'd actually reply**. The two combine into the order you see.
  </Step>

  <Step title="Pre-drafts the top replies">
    For the handful of most time-critical threads that need a response, it writes the reply ahead of time so it's ready the moment you open the task.
  </Step>

  <Step title="Stages your tasks">
    The highest-value items become Tasks in your queue, so your list is short and specific rather than a wall of mail.
  </Step>
</Steps>

### It gets better every week

Once a week the twin reviews how it did (which items you acted on, which you dismissed, which you edited heavily)
and corrects itself if it's drifting from how you actually work. You don't train it manually; it learns from what you
already do.

<Tip>
  The fastest way to improve your feed is simply to use it: send the good drafts, edit the ones that miss, and dismiss
  what you don't care about. Every one of those is a signal the twin learns from.
</Tip>

## Next steps

<CardGroup cols={2}>
  <Card title="Tactics" icon="zap" href="/home/tactics">
    Hands-off plays that run on a schedule
  </Card>

  <Card title="Demand" icon="inbox" href="/demand/overview">
    The full email workspace behind your Home feed
  </Card>

  <Card title="Quotes" icon="file-text" href="/demand/quotes">
    Turn a surfaced RFQ into a finished quote
  </Card>

  <Card title="Channels" icon="bot" href="/channels/overview">
    Set how the AI email agent drafts and replies
  </Card>
</CardGroup>
