Investigations

When should I use this?

• You need to explain what happened, why, and what to do next with verifiable evidence.
• You are evaluating how AutoObserve differs from dashboard-only observability tools.
• You want to understand the investigation model before connecting telemetry.

Requirements

• Metrics, logs, and traces connected via OpenTelemetryRequired
• Deployment or change signals when availableOptional
• Known incident window or service to inspectOptional

Steps

1. Understand the workflow

AutoObserve’s core workflow is not “another dashboard.”

Incident
  → Investigation
  → Evidence
  → Causal chain
  → Root cause
  → Recommended action

2. Walk the workflow guides

Each stage has a dedicated guide — work through them in order or jump to the stage you need:

1. Evidence — inspectable metrics, logs, traces, and change signals
2. Timeline — shared incident clock and confidence updates
3. Hypotheses — competing explanations tested against evidence
4. Root cause — current best explanation with uncertainty visible

3. Inspect every claim

Every root-cause claim should be backed by evidence you can open:

• Temporal correlation
• Deployment or configuration change
• Resource saturation
• Dependency propagation
• Trace / log behaviour change
• Historical baseline deviation

AI answers without inspectable evidence are not the product model.

4. Explore the canonical demo

Explore the canonical checkout-latency investigation on the Demo page — timeline, causal graph, and Inspect on every evidence card.

Did it work?

• You can articulate the Incident → Investigation → Evidence flow
• Demo investigation opens with inspectable evidence cards
• Ready to continue the investigation workflow

Troubleshooting