Set up the Assistant for a department
Turn on auto start for a department, add instructions for each phase, check which data sources and tools the Assistant can use, and find its output on escalated tickets
The Assistant helps resolve escalated tickets. It reads the ticket, looks up the requestor in your connected systems, posts a resolution plan, and then carries the plan out with your enabled tools. Its output is private to your team. The requestor never sees it.
The Assistant starts on a ticket in one of two ways:
- Auto start. When a department has Start automatically when a ticket escalates on, the Assistant starts on its own each time a ticket in that department escalates to a person.
- By hand. On any ticket that has a department, open the ticket's Team chat and send a message. Org Admins and Managers can do this, and Supervisors of the ticket's department. This works whether auto start is on or off.
This guide covers the department settings for auto start. For the Assistant itself (the docked panel, the Assistant page, approval cards, and approval mode), see the Assistant overview.
Prerequisites
- The Org Admin or Org Manager role, to change a department's Assistant settings.
Department Supervisors can use the Assistant on tickets, but cannot open these settings.
1. Turn on auto start for a department
There is no organization-wide switch: auto start is a per-department setting.
Open the department's settings
In the Risotto Dashboard, go to Settings. Under Departments, click the department.
Open the Assistant page
Click Assistant.
Turn on Start automatically when a ticket escalates
In the Auto-start card, turn on Start automatically when a ticket escalates. The change saves right away, and Risotto shows "Assistant settings saved".
2. Add your department's instructions for each phase
An automatic run has two phases. While auto start is on, the Auto-start card shows an Additional instructions (optional) field for each:
- Investigation. The Assistant reads the ticket and looks up what is already set up in your workspace that could resolve it, then posts a plan. Use this field to steer what it looks at and which tools it reaches for first.
- Execution. The Assistant carries out the plan and asks for your approval before any step that changes something. Use this field to steer how it acts and what it writes back, for example an internal note that summarizes what it changed.
Your instructions add department rules on top of the fixed auto start behavior. The Assistant's tone, its investigate-then-act flow, and its approval and safety rules stay the same. Refer to the ticket as this ticket and to the person behind it as the requestor, so the Assistant knows what it can act on.
Type your instructions
Type in the Additional instructions (optional) field under Investigation, Execution, or both. Each field holds up to 2,000 characters. The counter under the field shows how many you have used.
Save
Click Save at the top of the Auto-start card. Click Cancel to discard your edits instead.
The instructions steer only the automatic run that starts when a ticket escalates. When you start the Assistant by hand on a ticket, it doesn't use them.
3. Check what the Assistant can use
The What the Assistant can use card on the same page shows what the Assistant has to work with in your organization. It is the same for every department. There is nothing to configure here: connect more integrations or enable more tools to give the Assistant more to work with.
- Requestor context lists the three data sources the Assistant uses to look up the person behind a ticket: People (HR profile, manager, location, and department from your HRIS), Identity (identity profile and group memberships from your identity provider), and Devices (assigned devices and their status from your MDM). Each row shows Connected via and the provider, for example Connected via Okta, or Not connected.
- Tools lists the enabled tools the Assistant can propose when it carries out a plan. Each tool is marked Read-only or Requires approval. Manage tools opens the Tools page.
An automatic run always starts in Read only approval mode, so read-only lookups run on their own and a tool that changes something waits for your approval. See Approval mode.
4. View the Assistant's output on escalated tickets
Open the ticket
Open the escalated ticket from Tickets or your Inbox.
Open Team chat
Click Team chat at the right end of the top bar. The docked panel opens on the ticket's shared conversation. In the Inbox, the same conversation is the Assistant tab next to Ticket History above the selected ticket.
After an automatic run, the conversation shows:
- The Assistant's plan: one line for what the requestor needs, then the steps it will take. The plan is a message in this conversation, not a ticket comment.
- A Requestor context card with the details it found in your connected data sources.
- An approval card for the first step that changes something. Click Approve or Deny. See How tool execution works.
Each ticket has one Assistant conversation, shared by everyone who can open it. If the ticket escalates again, the Assistant continues in the same conversation. If the Assistant is already answering a message on the ticket when it escalates, the automatic run doesn't start.
If auto start is off for the department, the conversation is empty until you start the Assistant. Send a message, or pick a starting point such as Summarize this ticket.
Related
Something wrong or missing? Report an issue · View as Markdown