Eridian Ops
Incidents, runbooks, on-call, and post-mortems on the Ops vertical.
Eridian Ops is the SRE and IT vertical. Incident summaries, log analysis, on-call assistance, runbook generation, and post-mortems run through GPT and Gemini on the Eridian gateway. On-call owns the gateway and your services. The foundation model is a dependency, not the incident commander.
Platform procedures when Eridian itself is in the path are on Runbooks. This vertical is for your services.
What you get
| Workflow | What it does | Docs |
|---|---|---|
| Incidents | Open, summarize, timeline, close | Incidents |
| On-call | Shift context and assistant queries | On-Call |
| Logs | Natural-language queries over ingest | Log Analysis |
| Post-mortems | Reviewable records and action items | Post-Mortems |
| Services | Catalog, SLOs, error budgets | Services and SLOs |
| Integrations | PagerDuty, Slack, Datadog, ServiceNow | Ops Integrations |
| Webhooks | Incident and SLO events | Ops Webhooks |
Prerequisites
- Project
prj_ops_001(verticalops). - Key
eridian_sk_withops:write,inference:write,models:read. - At least one service in the catalog.
- An ingest source (OpenTelemetry, Datadog, or log drain). See Ops Integrations.
First summary
POST /v1/ops/incidents
Host: api.geteridian.com
Authorization: Bearer eridian_sk_...
Content-Type: application/json
Idempotency-Key: inc-2847-open
X-Eridian-Project: prj_ops_001
{
"service_id": "svc_payments_api",
"severity": "sev1",
"title": "payments-api 5xx > 5% for 10m",
"source": "pagerduty",
"external_id": "PD-20491",
"model": "auto",
"summarize": true
}
{
"id": "inc_2847",
"object": "eridian.ops.incident",
"status": "open",
"summary": {
"likely_cause": "Upstream ledger timeout; retries without backoff.",
"next_steps": ["Pin traffic to healthy AZ", "Confirm ledger p95"],
"model": "gemini"
},
"eridian": {
"request_id": "axm_req_ops_2847",
"route": "gemini",
"region": "eu-west-1",
"cost_usd": 0.058
}
}
Model policy
| ID | Typical Ops use |
|---|---|
auto | Policy selects GPT or Gemini |
gemini | Long log windows and timeline packs |
gpt | Post-mortem prose and customer-facing summaries |
Do not lower log-query filters during a SEV-1 to "get more context" without checking Ops Limits.
Next steps
- Wire PagerDuty
- On-call assistant
- Ops API
- Platform Runbooks when the model is in your callers' path
Contact: contact@geteridian.com for tenant escalations of the Eridian gateway. Your incident channel stays yours.
Production API credentials are issued with an institution workspace. Contact sales if you need access.