See Noctuary in action
on a real microservices stack.
The demo runs the OpenTelemetry Astronomy Shop — a 20-service e-commerce application generating continuous real traffic, failures, and alerts. Noctuary monitors it live. No synthetic data. No hand-crafted scenarios.
Three capabilities, one live environment
AI incident context
Watch Noctuary receive Grafana alerts and return a specific root cause hypothesis with evidence — service name, failure type, confidence score — in under 10 seconds.
Signal Map
Explore the live service dependency graph built from OpenTelemetry traces. See which services are degraded, which are healthy, and where failures cascade.
Continuous monitoring
The Astronomy Shop generates real load 24 hours a day via a built-in load generator. Incidents surface as services fail — the dashboard updates in real time.
Your incident feed, with hypotheses already attached
The Noctuary dashboard shows every incident Grafana has sent, enriched with an AI-generated root cause hypothesis and recommended action. No raw alert noise — just actionable context.
Each incident card shows the service affected, the alert that fired, the confidence score, and what Noctuary thinks caused it. Click through to see the full evidence trail.
- Live incident feed, split into active and resolved
- Hypothesis with confidence score from 0–100%
- Grafana alert labels and annotations surfaced inline
- Engineer feedback loop — confirm or mark wrong
- Signal Map with live service topology
- Ask Noctuary — conversational AI over your environment
OpenTelemetry Astronomy Shop
The demo runs the CNCF OpenTelemetry Astronomy Shop — an officially maintained reference microservices application used to validate observability tooling. It's an e-commerce store selling astronomy equipment, built across 20 services in Go, Python, Java, Node, .NET, and more.
A built-in load generator fires continuous purchase transactions, creating realistic traffic patterns, error spikes, and latency variations for Noctuary to monitor.
- 20 microservices across multiple languages and runtimes
- Continuous load generator — real traffic, real failures
- Full OpenTelemetry instrumentation out of the box
- Feature flags to trigger specific failure scenarios
Grafana — dashboards and alert routing
The demo Grafana instance collects all Prometheus metrics from the Astronomy Shop. It's configured to route alerts to Noctuary via webhook — every alert that fires here triggers an enriched incident in the Noctuary dashboard.
Explore pre-built dashboards for service latency, error rates, pod memory usage, and the OpenTelemetry Collector pipeline. You can also browse the alert rules to see exactly which conditions trigger Noctuary incidents.
- Pre-built dashboards for all 20 services
- Alert rules wired to Noctuary via webhook contact point
- Prometheus, OpenSearch, and Jaeger data sources
- Feature flag dashboard via flagd integration
Jaeger — trace every request across services
Jaeger captures distributed traces for every transaction flowing through the Astronomy Shop. When Noctuary surfaces an incident, you can jump to Jaeger to see the exact spans that slowed down or errored — from the frontend request all the way to the database.
Use Jaeger to confirm or challenge Noctuary's hypothesis. If Noctuary says the recommendation service caused a degradation, the traces will show you which span failed and what the error was.
- End-to-end traces across all 20 services
- Span-level error details and timing breakdowns
- Service dependency graph from real trace data
- Searchable by service, operation, or trace ID
Log in and explore
All four environments are publicly accessible. The Noctuary demo account has read-only access to the demo environment — you can explore all dashboards, incidents, and the signal map.
Dashboard & incident feed
Log in with the demo credentials to see live incidents, the signal map, and conversational AI over the environment.
Grafana, Jaeger & the shop
The Grafana instance, Jaeger UI, and the Astronomy Shop itself are all open — no authentication needed.
Noctuary on your own stack
The demo shows Noctuary on a single stack. In production it connects to your existing Grafana, PagerDuty, or any OTLP source — no rip and replace required.