Metric threshold
An Agent Serve Metric threshold compares a governed metric with a value and fires when it crosses the line.
Pipeline coverage < 3.0
Agent Serve · Monitors
Know when the picture changes.
Monitors watch the metrics that matter. When one crosses its line, the alert reaches your team and a Relay can start the follow-up.
Agent Serve Monitors watch important metrics and conditions in Agent Serve, the AI workspace where teams build, deploy, and manage AI agents. Monitors check a governed metric against a threshold, count the rows that match a rule, or follow a dashboard KPI card, and a Relay can start work when a monitor alerts.
How it works
Set the line once, then stop checking.
An Agent Serve monitor is a condition, a check frequency and a severity. It runs on its own until you pause it.
- 1
Choose what to watch
Pick a governed metric for a Metric threshold, or a rule and a table for a Rule match.
- 2
Draw the line
Compare the metric with a value using >, ≥, <, ≤, = or ≠, or set how many rows must match the rule.
- 3
Set frequency and severity
Check every 5 minutes, every 15 minutes, hourly, every 6 hours or daily, and mark the alert Info, Warning or Critical.
- 4
Alert and hand off
The alert lands on the Alerts tab, and a Relay set to run when a monitor alerts picks up the follow-up.
Three kinds
A metric, a rule or a KPI card.
Rule match
An Agent Serve Rule match fires when at least the number of rows you set in a table match a rule.
≥ 5 rows match “AML threshold”
KPI-card alerts
Choose New KPI alert on an Agent Serve dashboard card and alert when it goes above, drops below or changes by a percentage.
NPS drops below 40
Severity that sorts the queue
Every Agent Serve monitor raises its alerts as Info, Warning or Critical, and the Alerts tab filters by severity.
Critical · Warning · Info
Status at a glance
The Agent Serve Monitors tab shows each monitor as Active, Firing or Paused, with its target and owner.
12 monitors · 2 firing
Open until resolved
Agent Serve alerts stay Open until they are resolved, so the Alerts tab doubles as the follow-up list.
Open · Resolved · All alerts
In motion
From a changing number to a next step.
- 01
Watch a rule, not a hunch
A Rule match uses the same rule your dashboards use, so the alert and the analysis agree on what “at risk” means.
- 02
Hand off the follow-up
A Relay with ongoing work set to When a monitor alerts reads the records behind the alert and drafts an explanation.
- 03
Alerts on the cards people read
KPI-card alerts are created on the dashboard itself, and editing one opens that dashboard.
Under the hood
Agent Serve Monitors, in detail.
| Monitor kinds | Metric threshold, Rule match, and KPI-card alerts created from a dashboard. |
|---|---|
| Comparisons | >, ≥, <, ≤, = and ≠ against a value you set. |
| KPI-card conditions | Goes above, Drops below, or Changes by a percentage since the last check. |
| Check frequency | Every 5 minutes, Every 15 minutes, Hourly, Every 6 hours or Daily. |
| Severity | Info, Warning or Critical. |
| Notifications | In-app on the Alerts tab, always. Email me for KPI-card alerts. Slack through a workspace alert webhook. |
| Limit | Up to 200 monitors per workspace. |
| Hand-off | Relays can run ongoing work When a monitor alerts. |
Examples
What banks and lenders keep a watch on.
Three monitors financial services teams set up in Agent Serve, each one a check someone used to run by hand.
Deposit outflow watch
An Agent Serve Metric threshold checks seven-day deposit outflows every hour and raises a Critical alert above the limit.
- Governed metric
- Hourly check
- Critical alert
- Relay brief
AML threshold flags
An Agent Serve Rule match counts transactions that meet the AML threshold rule and alerts once enough rows match.
- Transactions table
- AML rule
- Rows counted
- Alert to analysts
Branch NPS drop
A regional head sets an Agent Serve KPI-card alert on branch NPS and gets an email when it drops below the target.
- KPI card
- Drops below
- Email me
- Dashboard review
Examples show what teams can build in Agent Serve. They describe patterns, not named customer deployments.
By the numbers
Enough monitors for every desk.
- Monitor kinds: metric, rule and KPI card
3
Monitor kinds: metric, rule and KPI card
- Severities: Info, Warning and Critical
3
Severities: Info, Warning and Critical
- Shortest check frequency
5 min
Shortest check frequency
- Monitors per workspace
200
Monitors per workspace
Before and after
Fewer morning checks, faster follow-up.
| Work | Before | With Agent Serve |
|---|---|---|
| Spotting a change | BeforeSomeone opens a report each morning and hopes to notice. | With Agent ServeAn Agent Serve monitor checks on schedule and raises an alert. |
| Definitions | BeforeThe alert and the report disagree on what “breached” means. | With Agent ServeMonitors read the same governed metrics and rules as dashboards. |
| The follow-up | BeforeAn email thread asking who will look into it. | With Agent ServeA Relay starts the investigation when the monitor alerts. |
| Knowing what is open | BeforeAlerts scattered across inboxes and chat channels. | With Agent ServeOne Alerts tab, filtered by severity, Open or Resolved. |
Governance
Alerts that respect who may see what.
Agent Serve evaluates KPI alerts with the owner’s data access and keeps every alert inside the workspace first.
Security and governanceOwner’s data access
Agent Serve evaluates a KPI-card alert with its owner’s data access, so an alert never reads rows its owner cannot see.
In-app first
Every Agent Serve alert is recorded on the Alerts tab. Slack delivery only happens when a workspace alert webhook is set.
Row-level access
Row-level policies narrow what each person sees in tables, dashboards, exports and scheduled reports.
Roles and permission groups
Admin, write and read roles per workspace, plus permission groups that limit which models, blocks and integrations a team can use.
SSO and SCIM
Sign in through your identity provider with SAML 2.0 or OIDC, and provision people with SCIM.
Self-hosting
Run Agent Serve in your own cloud or data centre, next to the systems it reads.
Works with
What a monitor watches, and who it calls.
- Monitors & alertsDashboards & BI
Create a KPI-card alert straight from a dashboard card.
- Monitors & alertsRules
Use a rule in a Rule match to count the rows that need attention.
- Monitors & alertsSemantic model
Watch formula metrics that resolve measures from the semantic model.
- Monitors & alertsTables
Run a Rule match over any workspace table.
- Monitors & alertsRelays
Start a Relay’s ongoing work When a monitor alerts.
Further reading
Solutions built on Agent Serve Monitors.
- Solution deep dive
Reconcile failed card disputes before the chargeback deadline
Disputes that fail reconciliation tend to sit in a side queue until the chargeback window has closed. Here is an end-to-end setup that finds them early, ranks them by deadline and drafts the pack, with an analyst signing every filing.
Cards and payments operations · 12 min read - Solution deep dive
Clear the KYC periodic review backlog without adding analysts
Periodic reviews pile up because every file starts from scratch, even when nothing has changed. Here is a setup that refreshes the data, drafts the file and routes it by risk, so analysts spend their time on the reviews that need judgement.
Retail and business banking · 12 min read - Solution deep dive
Catch missing risk warnings on recorded advice calls within 24 hours
Sampled call reviews find a missing risk warning weeks after the client acted on the advice. Here is a setup that checks every recorded advice call within a day and puts the quote in front of a compliance reviewer.
Wealth and advice · 12 min read
Questions
About Monitors
Where can I find dashboard metric alerts?
Dashboard KPI-card alerts live in Agent Serve Monitors beside metric and rule monitors. Editing one opens the dashboard it was created on.
Can an alert start agent work?
Yes. An Agent Serve Relay can run ongoing work When a monitor alerts, and its result lands in the Relay’s task list for review.
Where do alerts go?
Every Agent Serve alert appears in-app on the Alerts tab. KPI-card alerts can also email you, and alerts can go to Slack through a workspace alert webhook.
How often does a monitor check?
Agent Serve monitors check every 5 minutes, every 15 minutes, hourly, every 6 hours or daily, chosen per monitor.
Does Agent Serve detect anomalies on its own?
Not today. Agent Serve monitors fire on a threshold you set, a rule match count, or a KPI-card condition, so every alert has a stated reason.
Last updated