Escalations are currently in live beta. We're rolling this out gradually, so it may not be available in your dashboard yet. If you'd like to take part, contact us at [email protected]. As a beta feature, some details described here may change as we refine it.
The Risk Escalation System (RES) uses AI to monitor incoming help requests and voice messages for potential risks to health, safety, or property. When something looks serious, ZippyAssist raises an escalation — and can call your team directly so someone responds straight away.
This article covers turning escalations on, tuning how sensitive they are, deciding who gets contacted, and working an escalation through to resolution.
Turning escalations on
Escalations are off by default.
Go to Settings → Escalation.
Read the activation notice. Enabling escalations means request content is sent to OpenAI for assessment, and you're accepting OpenAI's Privacy Policy and Terms of Use.
Click I agree and activate.
Once active, every incoming help request and voice message is assessed automatically.
How an escalation gets triggered
There are three ways an escalation can start:
Trigger | What happens |
Risk score threshold | The AI scores each request for health/safety risk and property risk. If either score reaches your threshold, an escalation is raised. |
Equipment rules | If the request's equipment matches rules you've defined, an escalation is raised regardless of score. |
Manual | You create an escalation yourself from any request. |
Requests are grouped. If a new qualifying request comes in for equipment that already has an open escalation, it's attached to that escalation rather than creating a second one. This keeps one incident in one place instead of scattering it across several escalations.
Tuning sensitivity (thresholds)
Go to Settings → Escalation → Thresholds.
You have two independent sliders, each scored from 1 to 100:
Threshold | Default | Covers |
Health or Safety Risks | 30 | Risks to people — illness, injury, contamination |
Property Risks | 40 | Risks to property — damage, loss |
Requests scoring at or above your threshold are escalated. Scores below it are not. Lower threshold = more escalations — a threshold of 20 catches more than a threshold of 60.
Previewing the impact before you save
Underneath each slider, ZippyAssist shows how many of your actual past requests would have been escalated at the threshold you're currently setting. You can switch the window between 7 days, 30 days, 90 days, and 1 year.
If your draft setting differs from what's saved, a badge shows the difference — for example, "+14 vs saved" means this change would have escalated 14 more requests over that period.
Click View Requests to open the full list and read the actual requests that would have been caught. This is the best way to sanity-check a threshold before committing to it.
Recommendations
Start with the defaults. Watch how the system handles your own traffic for a while before adjusting anything.
Move in small steps. Large jumps produce surprising results.
Use the preview and the request list rather than guessing.
Equipment rules
Sometimes a piece of equipment warrants escalation no matter what the AI scores. A machine in a hospital, a school, or a secure site might need a human response every time.
Go to Settings → Escalation → Equipment Rules.
Turn on Enable equipment rules.
Build your rules using the same filter builder used for Equipment Sync Rules — choose a column, a condition, and a value, and combine filters with And / Or or nested groups.
A live count shows how many equipment items match. Click View to see the list.
Click Update to save.
Any request from matching equipment is escalated automatically, bypassing threshold scoring entirely.
Choosing who responds
Go to Settings → Escalation → Roles.
Responders and observers
Role | In the call pool? | Can claim? | Gets notifications? |
Responder | Yes | Yes | Yes |
Observer | No | No | Yes |
Responders are the people who will actually be contacted when an escalation is raised, and who can claim it. Observers are kept informed — they receive notifications when escalations are created, claimed, and resolved — but they're never called.
Add users to either list, and drag responders into your preferred order.
Note: Responders and observers must have a valid phone number on their user profile. If any don't, ZippyAssist shows a warning — voice and SMS notifications to those users will fail. Update their profile or remove them from the list.
Call settings
Setting | Default | What it does |
Automatic call routing | On | Automatically calls responders when a new escalation is created. Turn off to create escalations without calls. |
Randomize call order | Off | Randomizes which responder is called first. When off, responders are called in the order listed. |
Responder roster
If a responder is only available on certain days, click the roster option on their row and select the days they're on call (Sunday through Saturday). Leave the roster blank to make them available every day.
A cover summary under the responder table shows which days of the week currently have someone rostered — useful for spotting gaps before they matter.
Working an escalation
Escalations appear in the Escalations tab and move through three statuses:
Status | Meaning |
Unclaimed | Raised, nobody has taken it yet |
Claimed | A responder has taken ownership |
Resolved | The situation has been dealt with |
Open an escalation to see the risk assessment (the health/safety and property scores, with the AI's explanation for each), a plain-language summary that updates as more requests attach, every attached request forming part of the incident, and the Responder Log.
You can also add Notes to an escalation to record what your team did.
Reading the Responder Log
The log shows each contact attempt, grouped by attempt number, so you can see exactly how the system worked through your responder list.
Status | Meaning |
Initiated | The call has been started |
Ringing | The responder's phone is ringing |
Acknowledged | The responder answered and acknowledged |
No Answer | Nobody picked up |
Voicemail | The call reached voicemail |
Declined | The responder declined the call |
Failed | The call couldn't be placed |
Creating an escalation manually
If you spot something that should be escalated but wasn't, you can raise one yourself.
Open the request from the help request or voice message list.
Choose Create Escalation from the actions menu.
Set the initial Status and optionally assign it to a user.
Choose whether to trigger call routing — this defaults to your organization's automatic call routing setting.
Confirm.
Permissions
Escalation access is controlled by user role. Admin and Owner roles can view, create, update, and delete escalations and change escalation configuration. Users with the standard User role can work escalations but do not see the escalation configuration screens. See Are there different roles for users? for more on roles.
Related articles
See also About Help Requests, Voice Messages, and Notifications and Notification Center.
If you have any questions, use the chat widget at the bottom of the screen or email us at [email protected]!