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

# IT help desk

> Build a complete IT help desk in Ravenna: one channel, the three forms that cover most volume, knowledge, an agent with gated tools, and SLAs.

export const Checkbox = ({items = []}) => {
  const [checked, setChecked] = useState(Array(items.length).fill(false));
  const toggle = index => {
    setChecked(prev => {
      const copy = [...prev];
      copy[index] = !copy[index];
      return copy;
    });
  };
  return <div className="custom-checkbox-list">
      {items.map((text, index) => <div key={index} className="custom-checkbox-item flex items-center mb-2">
          <input type="checkbox" checked={checked[index]} onChange={() => toggle(index)} className="mr-2" />
          <span>{text}</span>
        </div>)}
    </div>;
};

This is the desk most teams launch first, and the one every other playbook borrows from. The goal is not to model every IT process, it is to absorb the repetitive 70 percent so your team gets its afternoons back.

<Info icon="list-check">
  **Prerequisites:** the [Day one](/guides/day-one/overview) track, finished. This playbook assumes you already have a workspace, a channel, a triage channel, categories, knowledge, and one agent.
</Info>

## What good looks like

An IT desk is working when:

<Checkbox
  items={[
"Employees ask in one Slack channel and never wonder where to go",
"Common requests arrive as forms, so nobody chases details",
"The agent answers documented questions without a human touching them",
"Requests that need action are routed and assigned automatically",
"Anything risky waits for approval instead of being blocked outright"
]}
/>

<Steps>
  <Step title="Keep intake to one channel">
    One request channel, `#ask-it`. Resist the pull to split by topic. Every extra channel is another place a ticket can sit unnoticed, and Ravenna slices work with views instead.

    Add the other entry points to the same channel rather than creating new ones: the channel's inbound [email address](/guides/playbooks/email-support) for your old `it@` inbox, DMs to the Ravenna bot for anything sensitive, and the [Portal](/guides/playbooks/launch-the-portal) for staff who do not use Slack.

    <Callout icon="link" color="#6B7280">Learn more about [choosing between channels, views, and categories](/guides/best-practices/channels-views-categories)</Callout>
  </Step>

  <Step title="Build the three forms that cover most volume">
    Do not build twelve forms. Look at a month of requests and you will nearly always find the same three shapes:

    | Form                | What it captures                                                   | Why it earns its place                               |
    | ------------------- | ------------------------------------------------------------------ | ---------------------------------------------------- |
    | Account access      | Application, access level, business reason, duration               | Feeds an approval and, later, real provisioning      |
    | Device and hardware | Device type, location, urgency, existing asset                     | Turns "my laptop is broken" into something shippable |
    | Something is broken | Affected system, error text, when it started, who else is affected | Distinguishes an outage from a one-person problem    |

    Set defaults on each: priority, tag, and the channel to publish into. That is what saves your team the triage click.

    <Callout icon="link" color="#6B7280">Learn more about [forms](/documentation/tickets/forms/overview) and [custom fields](/documentation/tickets/forms/custom-fields)</Callout>
  </Step>

  <Step title="Give responders views, not more channels">
    Three saved views handle most of the working day:

    * **Unassigned** and not resolved. The safety net.
    * **Mine, active**. What each responder works from.
    * **Breaching soon**, once SLAs are on.

    <Callout icon="link" color="#6B7280">Learn more about [views](/documentation/tickets/organize/views)</Callout>
  </Step>

  <Step title="Index the runbooks people actually ask about">
    Start with VPN, laptop setup, MFA, software requests, and printing. That short list answers a surprising share of tickets.

    Test in the knowledge chat panel before deploying. If an answer only exists inside a screenshot, the agent cannot read it, and you will not discover that from a dashboard.

    <Callout icon="link" color="#6B7280">Learn more about [structuring knowledge for AI](/guides/best-practices/knowledge-for-ai)</Callout>
  </Step>

  <Step title="Give the agent tools, gated by risk">
    This is the step that separates a help desk from an answering machine. Attach tools to rules and set the execution policy per rule:

    * **Auto-execute** the lookups: find a user, check group membership, read a ticket.
    * **Requires confirmation** for reversible writes where you want a "shall I?" prompt.
    * **Requires approval** for anything that grants access or changes production, pointed at an approval template with your IT leads.

    A rule that looks up a group automatically and gates the membership change is the pattern to copy.

    <Callout icon="link" color="#6B7280">Learn more about [tool execution policies](/documentation/automate/agents/configure#tool-execution-policies)</Callout>
  </Step>

  <Step title="Automate the handoffs, not the judgment">
    Two workflows are usually enough to start:

    * **Form-driven routing.** Trigger on Form Submitted, branch on urgency, set priority and assignee, post to the triage channel.
    * **Nothing has happened.** Trigger on inactivity and nudge the assignee before the requester has to.

    Remember that the agent and your workflows do not call each other. The agent sets a form, category, or status, and the workflow triggers on that.

    <Callout icon="link" color="#6B7280">Learn more about [workflow triggers and actions](/documentation/automate/workflows/triggers-actions)</Callout>
  </Step>

  <Step title="Set response targets you can actually meet">
    Add one SLA before you add five. First response within four business hours, measured against a business schedule that matches your working week, with a pause status for tickets waiting on the requester.

    <Callout icon="link" color="#6B7280">Learn more in [SLAs and business hours](/guides/playbooks/slas-and-business-hours)</Callout>
  </Step>

  <Step title="Review it weekly for a month">
    Fifteen minutes, same slot every week:

    * Read ten agent conversations end to end
    * Clear the top two knowledge gaps
    * Check CSAT comments, not just the score
    * Look at AI Outcome, and expect it to move slowly

    <Callout icon="link" color="#6B7280">Learn more in [reporting and briefs](/guides/playbooks/reporting-and-briefs)</Callout>
  </Step>
</Steps>

## Where teams get this wrong

<AccordionGroup>
  <Accordion title="Too many channels on day one">
    Splitting `#ask-it-hardware` and `#ask-it-access` before you have volume creates dead channels and confused employees.
  </Accordion>

  <Accordion title="Forms that ask everything">
    Every optional field is a field requesters skip and responders read past.
  </Accordion>

  <Accordion title="An agent with no tools">
    Answering questions is table stakes. Resolving requests needs tools.
  </Accordion>

  <Accordion title="Ungated write tools">
    One bad automated group change costs more trust than a month of good answers.
  </Accordion>
</AccordionGroup>

## Next

<CardGroup cols={2}>
  <Card title="Employee onboarding" icon="user-plus" href="/guides/playbooks/hr/employee-onboarding" horizontal>
    The highest-value process to automate next.
  </Card>

  <Card title="Approval flows" icon="check-check" href="/guides/playbooks/approval-flows" horizontal>
    Formalize who signs off on access.
  </Card>
</CardGroup>


## Related topics

- [IT](/guides/playbooks/it/overview.md)
- [SLAs and business hours](/guides/playbooks/slas-and-business-hours.md)
- [Playbooks](/guides/playbooks/overview.md)
- [User handbook](/guides/user-onboarding/handbook/quirky-edition.md)
- [Channels, views, and categories](/guides/best-practices/channels-views-categories.md)