Incident Lake

TRUST AND SECURITY

Your incident data is among the most sensitive you hold. Here is how we handle it

An incident record can hold root causes, unpatched weaknesses, customer impact and sometimes personal data. Before you connect a new vendor to that, your security review will want to know who runs the service, what controls exist, and exactly what the AI agent can and cannot touch. This page sets out the facts.

Independent assurance

ISO/IEC 27001

SIGQ, the company that operates Incident Lake, holds ISO/IEC 27001 certification.

SOC 2 Type 1

Incident Lake has received a SOC 2 Type 1 report. Contact us to request access as part of your vendor review.

Identity and access controls

Least privilege applies to incident data as much as to production. These controls let you scope who gets in and what they can see.

SAML SSO

Sign in through your existing identity provider.

IP restrictions

Limit access to approved source IP addresses.

Per-feature permission groups

Grant read, write or manage rights feature by feature, rather than one role for everything.

RBAC tags

Apply access control at the tag level.

Feature toggles

Switch off any feature you don't use, and keep the surface area small.

Audit logs

See who did what, and when.

Permission groups are included from the Team plan. Audit logs and SAML SSO are included with annual plans.

What the agent can do in your other systems

Integration secrets are encrypted per tenant. Monitoring, code and documentation connections are read-only. The only write-back to a work tracker is task sync, and it stays off until an admin enables it.

Integration
What the agent does
Writes back?
Notes
Monitoring: Datadog, Sentry, Grafana, Zabbix, Instana, Rollbar
Receives alerts, reads context
No
Read-only
Code and docs: GitHub, Notion, Google Drive
Reads for investigation
No
Read-only
Work tracking: Jira, Backlog, ServiceNow
Syncs tasks, auto-links URLs
Only if an admin enables it
Creates issues and transitions status
Chat: Slack, Microsoft Teams
Joins the thread, asks questions
Posts in the incident thread
Starts on @mention

Human in the loop, by design

The agent operates under AI guidelines that keep a human in the loop. Confirming severity, declaring resolution, containment, and submitting anything to an external party stay with your people; the agent prepares the facts and the drafts. You can choose which model it uses.

When data cannot leave your environment

For organizations that cannot send incident data to an external LLM, Incident Lake Self-Hosted runs alongside a self-hosted LLM. We will walk through architecture and requirements during onboarding.

About Self-Hosted →

For any certification or framework not listed here, please ask us directly rather than assuming coverage.

Running a vendor security review?

Request the SOC 2 Type 1 report and the rest of your review pack from us. Meanwhile, your engineers can evaluate the product on the free plan.

Request security documentation →

Start free (2 min)