Governance Notifications
ANTS Platform sends governance findings to the tools your teams already use. Each kind of finding goes to the channels that fit it:
| Finding | Where you send it from | Channels | How it is sent |
|---|---|---|---|
| Unapproved model use | AI Agent Center, agent Risk tab, Unapproved Models | Slack, Jira, Linear | One click, per agent |
| Drift alerts | Drift Alerts, and the agent's Drift Alerts section | ServiceNow, PagerDuty | One click, per alert |
| AI policy detections | PII Detection policy, PagerDuty Alerting | PagerDuty | Automatic, per agent |
| Evidence requests | Requests & Review | Slack, Jira, Linear | When you create or resend a request |
Enterprise feature. Governance notifications are separate from the in-app notification bell, which covers exports, expiring keys and role changes. See Notifications.
Set up
Connect the integrations
Open Project Settings → Integrations and connect the tools you want: Slack, Jira, Linear, ServiceNow and PagerDuty. Connecting needs the Notification & analytics integrations permission (under Integrations & service accounts in the role builder) and permission to configure notification automations (the Governance module default at Admin). Among built-in roles, Owner has both.
Map agents to Slack, Jira and Linear
Slack, Jira and Linear messages go to the channel, project or team mapped to each agent. Once an integration is connected, its settings page shows a Mappings section with Manage Mappings. Map each agent there. Editing mappings needs AI Agent Center or Agent Profiles & Drift at Read & Write.
An agent with no mapping has no Slack, Jira or Linear destination, and its Integrations button points you back to the settings.
Turn on incident creation for ServiceNow and PagerDuty
ServiceNow and PagerDuty are set at project level. Turn on incident creation in each integration. PagerDuty also needs a synced service for the agent; with auto-create services on (the default), ANTS Platform creates one per agent.
Unapproved model alerts
When an agent has used an unapproved model in the selected period, the Unapproved Models section of its Risk tab shows an icon for each Slack channel, Jira project and Linear team mapped to it. Select one to open Send Compliance Notification, choose one or more destinations and select Send Notification.
| Channel | What is sent |
|---|---|
| Slack | "Compliance Alert": the agent, how many times it used unapproved models in the period, the list of models, a View Violation Traces button, and a reminder to approve the models in AI Inventory |
| Jira | A Task "Compliance Alert: agent using unapproved models", labelled compliance, ai-governance and auto-generated, with the model list and a link to the traces |
| Linear | An issue "Compliance Alert: agent" at Medium priority, with a table of agent, agent ID, violations and models, and an action checklist |
Sending needs the AI Agent Center permission. If the agent has no violations in the period, nothing is sent.
Drift alerts
In the Integrations column of a drift alert, select an integration icon to open Send Drift Alert Notification. Choose ServiceNow, PagerDuty or both.
The incident is titled "Drift Alert [severity]: agent - drift type" and includes the description, the declared and observed values, an action note, a link to Drift Alerts, and the alert, agent and project IDs. Severity maps to ServiceNow impact and urgency (Critical and High to 1, Medium to 2, Low to 3) and to PagerDuty severity (critical, error, warning, info). The alert then records when it was sent and to which channel.
Drift alerts are sent only when you choose to. They are not sent to Slack, Jira or Linear.
Detection incidents
Scheduled detection policies can raise PagerDuty incidents for an agent's PII, PCI, PHI, harmful content, bias and jailbreak detections. Turn on Enable PagerDuty Alerting in the agent's detection policy and choose a service. See PII Detection.
Evidence requests
Compliance evidence requests are delivered to the agent's mapped Slack channel, Jira project or Linear team, with a link to accept the request. Feedback from a review is posted back to the Slack thread or Jira issue. See Requests & Review.
Permissions
| Action | Permission needed | Built-in roles |
|---|---|---|
| Connect or disconnect integrations | Notification & analytics integrations, and Governance module default at Admin | Owner |
| Map agents to Slack, Jira and Linear | AI Agent Center or Agent Profiles & Drift, Read & Write | Owner, Department Admin, Security & Compliance, Cost Analyst |
| Send unapproved model alerts | AI Agent Center | Owner, Department Admin, Security & Compliance, Cost Analyst |
| Send drift alerts | Agent Profiles & Drift | Owner, Department Admin, Security & Compliance, Auditor |
Custom roles can grant each of these. See Custom Roles.
Next steps
- AI Agent Center: find agents with unapproved models.
- Drift Alerts: work the drift queue.
- Integrations: connect Slack, Jira, Linear, ServiceNow and PagerDuty.