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

# How do I hand the agent a task that takes days?

> Turn a request into a long-running goal. The agent keeps working, follows up, and reports back when the outcome is done.

A goal is an outcome the agent owns until it is finished, even when finishing depends on other people replying. Use one for things like "get a time on the calendar with Maya" or "chase this vendor until the contract is signed."

Goals work in the agent chat (`Cmd+/`) and in iMessage. To hand off a goal by email instead, see [Assistant email](/features/ai-agent/assistant-email).

## Start a goal

<Steps>
  <Step title="Ask for the outcome">
    Describe what you want in plain language, for example "Handle scheduling with Maya for next week."
  </Step>

  <Step title="Confirm the goal">
    When the request needs follow-through over time, the agent proposes making it a goal and asks you first. Reply **yes**.
  </Step>

  <Step title="Or start one directly">
    Type `/goal` followed by the objective, for example `/goal Get the signed contract back from Acme`. That counts as your confirmation.
  </Step>
</Steps>

The agent never creates a goal from your first message alone unless you use `/goal`.

## What the agent does with a goal

Once you confirm, the goal runs on its own in the background:

* It reads the relevant threads and your calendar, drafts and sends emails, creates or updates calendar events, and sets reminders for itself.
* It wakes up when a reply arrives on a thread it is tracking, or when one of its reminders fires.
* If it needs a private fact only you know, it asks you one clear question and waits.
* When the outcome is achieved, it tells you what happened.

<Warning>
  Goals do not ask before each email. Once you confirm a goal, the agent is authorized to send email from your connected inbox and change calendar events to reach the outcome. Confirm only what you are comfortable with it doing on its own. Normal chat and iMessage requests still ask before sending.
</Warning>

Read Mode still applies. With it on, the agent cannot send email to other people or change your calendar. See [Agent autonomy](/features/ai-agent/autonomy).

## Check on a goal

Ask the agent: "What's the status of my goals?" It lists recent goals with what each is doing now and how it turned out. To add information or change direction, tell the agent in the same conversation, for example "Maya can only do mornings." The agent passes it to the goal. A finished goal reopens if the same outcome changes.

## Goal, reminder, or automation?

| Use | When |
| - | - |
| **Goal** | The outcome depends on future replies or events, and you want the agent to decide what to do at each step |
| **[Reminder](/features/automations/scheduled-reminders)** | One nudge at a set time |
| **[Email trigger](/features/automations/email-triggers)** | The same action every time a certain email arrives |

Slashy can also suggest goals on its own through the [Daily Admin Scout](/features/automations/scheduled-reminders#daily-admin-scout).

<CardGroup cols={2}>
  <Card title="Assistant email" href="/features/ai-agent/assistant-email">CC the assistant on a thread to hand it off.</Card>
  <Card title="Agent autonomy" href="/features/ai-agent/autonomy">What the agent does on its own and when it asks.</Card>
  <Card title="Scheduled reminders" href="/features/automations/scheduled-reminders">Schedule the agent for one-off tasks.</Card>
  <Card title="Does Slashy send emails on its own?" href="/getting-started/ai-sending-control">How sending is controlled.</Card>
</CardGroup>