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

# AI Agents overview

> What an AI agent is, how the list page is organised, and what its four headline numbers count.

export const Screenshot = ({id, alt, caption, frame = 'browser', url, marks = [], lang = 'en', workspace = 'northwind-outbound'}) => {
  const src = `/images/screenshots/${lang}/${id}.png`;
  const chrome = url ? `app.pipelime.ai/${workspace}${url}` : 'app.pipelime.ai';
  return <figure className={`pl-shot pl-shot--${frame} not-prose`}>
      <div className="pl-shot__frame">
        {frame !== 'bare' && <div className="pl-shot__bar">
            <span className="pl-shot__dots">
              <span className="pl-shot__dot" />
              <span className="pl-shot__dot" />
              <span className="pl-shot__dot" />
            </span>
            <span className="pl-shot__url">{chrome}</span>
          </div>}
        <div className="pl-shot__media">
          <img src={src} alt={alt} loading="lazy" />
          {marks.map(mark => <span key={mark.n} className="pl-shot__mark" style={{
    left: `${mark.x}%`,
    top: `${mark.y}%`
  }} aria-hidden="true">
              {mark.n}
            </span>)}
        </div>
      </div>
      {caption && <figcaption className="pl-shot__caption">{caption}</figcaption>}
    </figure>;
};

AI Agents is the list of your agents: how many companies each one has analysed, how many leads it has written to, and the controls to create, pause or delete one without opening it.

<div className="pl-availability">
  <div className="pl-availability__row">
    <div className="pl-availability__label">Where</div>
    <div className="pl-availability__value"><span className="pl-path">Outreach<span className="pl-path__sep">→</span>AI Agents</span></div>
  </div>

  <div className="pl-availability__row">
    <div className="pl-availability__label">Your role needs</div>
    <div className="pl-availability__value">Read access to AI Agents <code>read-campaigns</code>. Admin, Member and Viewer have it by default.</div>
  </div>

  <div className="pl-availability__row">
    <div className="pl-availability__label">To create or change</div>
    <div className="pl-availability__value"><code>create-campaigns</code> to add one, <code>update-campaigns</code> to change one, on top of the permission above.</div>
  </div>

  <div className="pl-availability__note">If you cannot find this in your sidebar, your workspace may have a custom menu configuration. Contact support and we will check it for you.</div>
</div>

Nothing behind this page is stored as an agent on its own: each one is a campaign, and its lead sources are whichever ones share its [tag](/en/leads/tags). Two consequences are worth carrying with you. An agent with no tag, or a tag that no lead source shares, has nothing to work from, so it shows as **Needs setup** here whatever its own pause switches say. And because the list is the workspace's campaigns, a campaign you built on the older [AI Campaigns (Legacy)](/en/ai-agents/ai-campaigns-legacy) screen appears here too, as an agent. The same tag join limits the [agent dashboard](/en/ai-agents/agent-dashboard): reply rate, open rate and deliverability are only available for an agent carrying exactly one tag.

## The banner and the KPI strip

Above the list sit a banner (it repeats the **Create AI Agent** button) and four numbers.

**Active agents** is a ratio: how many agents have outreach switched on, against how many exist. It has no time window (it counts the workspace as it stands), and its caption says how many of them are paused.

The other three cover the last 30 days, and they say so in their captions. **Emails sent** counts every email the workspace sent to a lead in that window, not only the ones an agent sent, and is captioned with how many unique leads received one. **Reply rate** is captioned with how many of those sent emails came back with a reply, against the same sent figure: one email that drew three replies still counts once. **Meetings booked** counts leads whose meeting was booked inside the window, again across the whole workspace rather than per agent.

None of the four narrows when you search or pick a status chip below. They describe the whole workspace, not the rows on screen.

## Finding and reading an agent

Four chips filter the list: **All**, which carries the number of matching agents next to it while it is the chip selected, then **Active**, **Paused** and **Needs setup**. **Active** and **Paused** sort agents by the outreach switch alone, not by the pill on the row: an agent showing **Sourcing paused** or **Needs setup** still sits under **Active** if its outreach is on. A **Search by name** box sits beside the chips, along with a **Campaign analytics** icon that opens the campaigns report, and a **Create AI Agent** button.

<Screenshot id="ai-agents/overview--list" url="/ai-agents" alt="The AI Agents page: a banner, a four-card KPI strip, status filter chips, a search box, and a list of agent rows each showing a status pill, a goal pill, companies and contacted counts, and action buttons" caption="The KPI strip counts the whole workspace, even while a chip or a search narrows the list underneath it." marks={[{ n: 1, x: 18, y: 52 }, { n: 2, x: 58, y: 52 }]} />

Each row is one agent: its name, a status pill (1), then, when the agent has anything wrong with it, a pill counting its [open issues](/en/ai-agents/agent-alerts-and-issues), then a small coloured pill naming its goal. See [agent statuses and pause options](/en/ai-agents/agent-statuses-and-pausing) for what each of the six statuses means. Two stat columns follow (2): **Companies**, how many it has analysed against its sourcing limit, summed across all of its sources; and **Contacted**, unique leads it has written to, over how many leads carry its tags. Then come a pause-or-resume icon, an overflow menu (**Analytics**, **Edit settings**, **Delete**), and an **Open** button to its dashboard.

Pausing from that icon stops outreach and every source still running. Resuming restarts outreach and the sources this control could have paused itself: a source stopped because it ran out of credits, or because it reached its company limit, stays stopped until you fix the cause.

<Note>
  A role that lacks the permission behind one of these controls does not see that control at all: it is hidden, not greyed out. Pausing and **Edit settings** share one permission, `update-campaigns`. **Delete** needs `delete-campaigns`, and the checkbox inside its dialog needs a third, `delete-lead-generations`. The built-in Viewer role holds none of them, so a viewer gets the list, the numbers, **Analytics** and **Open**.
</Note>

**Delete** always stops the agent's lead sourcing first, pausing whatever is still running. If the agent has any sources, the dialog also offers a checkbox that deletes them outright instead of pausing them. Leads it already found are kept either way.

## Lead sources without an agent

A lead source only belongs to an agent because they share a tag. When one doesn't (its tag matches no campaign's tag, or it was never tagged at all), it becomes an orphan, and a dashed strip appears below the list naming up to four of them, with a "+N more" line for the rest. The strip is only worked out for the first page of the unfiltered list, so it disappears while a search is running, or while any chip other than **All** is selected.

<Screenshot id="ai-agents/overview--orphan-strip" url="/ai-agents" alt="A dashed warning strip below the agent list, listing lead sources not connected to any agent, each with a status dot and a Finish setup button" caption="Finish setup tags the search you already have rather than starting a second one, so the companies it has analysed come with it." marks={[{ n: 1, x: 8, y: 30 }, { n: 2, x: 88, y: 30 }]} />

Each entry shows a status dot (1), carrying the source's own status, and a **Finish setup** button (2) that opens the creation wizard already pointed at that lead source. A source that has analysed companies also shows how many.

## When there are no agents yet

With no agents at all, and no search or chip narrowing the list, the banner, the KPI strip and the list are replaced by a single invitation.

<Screenshot id="ai-agents/overview--empty" url="/ai-agents" alt="The AI Agents page with no agents: an illustration, the heading Hire your first AI sales agent, and two buttons, Create AI Agent and Suggest with AI" caption="A representative example: no seeded workspace has zero agents, so the list behind this screen is simulated. Suggest with AI hands the creation wizard a starting point instead of a blank form." />

**Hire your first AI sales agent** reads the heading, with a **Create AI Agent** button and a **Suggest with AI** button beside it. Most workspaces never meet this screen. Onboarding builds a first agent for you and lands you here: while that agent is being built the page shows a placeholder in place of the list, and once it exists you get its row.

## Related

<CardGroup cols={2}>
  <Card title="Create an AI agent" icon="sparkles" href="/en/ai-agents/create-an-ai-agent">
    The four-step wizard behind **Create AI Agent**, and what each field in it does.
  </Card>

  <Card title="Reading the agent dashboard" icon="layout-dashboard" href="/en/ai-agents/agent-dashboard">
    What the **Open** button on a row leads to: the KPIs, alerts, funnel and activity feed.
  </Card>

  <Card title="Agent statuses and pause options" icon="toggle-left" href="/en/ai-agents/agent-statuses-and-pausing">
    What each status pill means, and the difference between pausing sourcing, outreach, or both.
  </Card>

  <Card title="Tags: how agents, sources and leads connect" icon="tag" href="/en/leads/tags">
    The join behind everything on this page: what happens when a lead source and an agent share no tag.
  </Card>
</CardGroup>