Skip to main content
The fastest way to understand sureops is to watch it handle a real incident end-to-end. This guide walks you through the demo environment path — a pre-wired environment that comes with a running sample application, connected observability stack, and the ability to trigger incidents on demand.
The demo environment uses a real running application with real telemetry. The AI agents diagnose from actual logs, metrics, and traces — the same way they would on your production stack. There’s no simulation.
1

Get your invite and log in

sureops is in private beta — apply for an invite and we’ll review your application by hand. Approved teams get a one-tap invite link by email — there’s no separate signup step. Open the link to sign in directly.After logging in, you’ll land on a guided setup tour that walks you through onboarding step by step.
2

Provision your sample environment

Early in the tour, click Provision sample environment. sureops spins up a demo environment for your account automatically. This environment includes:
  • A sample microservices application (OpenTelemetry-instrumented)
  • Grafana (with Prometheus, Loki, and Tempo datasources) and Tempo pre-connected via MCP
  • A Kubernetes cluster with ArgoCD managing deployments
  • GitHub connected for repository context and fix PRs
Provisioning takes a few minutes; the tour continues automatically once it’s done. The remaining steps walk you through choosing your AI model, connecting telemetry, alerts, and Slack, connecting GitHub, generating your contract and landscape, drafting a starter runbook, and setting up routing and escalation.
3

Trigger a demo incident

The last step of the tour is Trigger your first incident. Pick a failure scenario from the picker — a crashing pod (recommendation-cache OOM), an ad-service CPU saturation spike, a payment charge failure — and sureops injects it into your sample environment for real, then drops you on the Command Center dashboard to watch.Detection runs on live telemetry, so the incident takes roughly 5–12 minutes to surface, depending on the scenario — exactly as a real alert would. You don’t need to do anything else — watch the incident card appear and the stage indicator advance in real time.You can trigger additional scenarios later too: go to Settings → Environments → [your environment] and click Inject demo failure to open the same scenario picker. Each scenario can only be injected once per environment — re-triggering the same one is blocked until you terminate and re-provision; pick a different scenario instead.
4

Follow the incident through the lifecycle

Click the incident to open the Incident Hub. You’ll see:
  • The triggering alerts that created the incident
  • The triage assessment — severity, affected services, estimated impact
  • The diagnosis agent running its investigation (live progress in the agent activity feed)
  • Once diagnosis completes: ranked recommendations with evidence and confidence scores
The demo environment starts in Supervised mode, so the AI pauses after diagnosis and waits for your approval before proceeding.
5

Approve a HITL action

When the diagnosis completes, a pulsing Review Approval button appears on the Incident Hub, and the Command Center’s approvals panel shows an Action Required badge with a {N} pending count.Review the recommendations — each one shows the proposed action, the risk level, the agent’s reasoning, and the evidence it used. When you’re ready:
  • Click the approve button to let sureops proceed — its exact label depends on what the recommendation does: Approve & let the agent execute, Approve & open a fix-PR, Approve & hand to a human, or Approve & escalate to a human
  • Click Reject to decline and either provide feedback or take manual control
After approval, the resolution agent executes the fix and verification checks that it actually worked. If recovery is confirmed, the incident closes automatically with an AI-authored stage summary — and, if your incident policy opens a Problem Record for this severity, a deeper RCA document too. If verification can’t confirm recovery, sureops escalates and keeps the incident open rather than closing it silently.
This approval step is the core of Human-in-the-Loop (HITL). In Self-driving mode, this gate can be configured to auto-approve or auto-execute up to a risk ceiling — but two carve-outs always require it regardless of settings: P1/P2 severity incidents, and any recommendation that opens a fix PR. See HITL gates for the full policy model.

​
What to explore next

After your first incident, the demo environment gives you a few more things worth trying:
  • Take Control mid-incident — click “Take Control” while the diagnosis agent is running. Watch it finish its current operation, hand off findings, and pause. Use the stage chat to ask the agent questions, then click “Release to AI” to hand the incident back — the AI picks up with all context intact. Note this only hands control back and forth; it doesn’t change the incident’s underlying autonomy mode.
  • Problem Records — after the incident closes, navigate to Problems. Depending on your org’s incident policy (Settings → Incident Policy), closing this severity may have opened a Problem Record automatically. This is where longer-lived pattern tracking and post-closure investigation live.
  • SAGE — open SAGE from the incident header. It has full context on this incident’s stages, timeline, and recommendations, and can answer questions about what happened, why the AI made the decisions it did, and what to watch for next.

​
Ready to connect your own stack?

The demo environment is designed for evaluation. When you’re ready to run sureops against your own services and infrastructure, see: