Start a Bits AI SRE investigation

You can launch a Bits AI SRE investigation from several entry points:

• Monitor alerts, which you can trigger in two ways:
  • Manual: Start from an individual monitor alert
  • Automatic: Configure monitors to automatically launch a Bits investigation whenever they enter an alert state
• APM latency graphs on service pages
• APM latency Watchdog stories
• General prompt

Manually start an investigation

Monitor alerts

You can invoke Bits on an individual monitor alert or warn event from several entry points:

Option 1: Bits AI SRE Monitors list

1. Go to Bits AI SRE > Monitors > Supported.
2. Click Investigate Recent Alerts and select an alert.

Option 2: Monitor status page

Navigate to the monitor status page of a Bits AI SRE-supported monitor and click Investigate with Bits AI SRE in the top-right corner.

Option 3: Monitor event side panel

In the monitor event side panel of a Bits AI SRE-supported monitor, click Investigate with Bits AI SRE.

Option 4: Slack

To use the Slack integration, connect your Slack workspace to Bits AI SRE.

In Slack, reply to a monitor notification with @Datadog Investigate this alert.

APM latency

APM latency graphs on service pages

1. In Datadog, navigate to APM and open the service or resource page you want to investigate. Next to the latency graph, click Investigate.
2. Click and drag your cursor over the point plot visualization to make a rectangular selection over a region that shows unusual latency to seed the analysis. Initial diagnostics on the latency issue appear, including the observed user impact, anomalous tags contributing to the issue, and recent changes. For more information, see APM Investigator.
3. Click Investigate with Bits AI SRE to run a deeper investigation.

APM latency Watchdog stories

On a Watchdog APM latency story, click Investigate with Bits AI SRE.

General prompt

Click on New Investigation and describe the issue you want to troubleshoot. Include as much relevant context as possible:

• Observed symptoms (e.g., errors, latency) including any links to Datadog telemetry that indicate this
• Relevant tags such as service, application, or environment that isolate the issue
• A time window (default is past 4 hours)

The more specific your prompt, the more accurate and useful the investigation will be.

Good examples:

• Investigate requests to search-service in prod. Some requests are returning empty results when they should not. The issue began around 10:00 UTC.
• Investigate what’s causing these errors: https://app.datadoghq.com/logs?query=service%3Asearch-service%20status%3Aerror They started appearing about 5 minutes ago.

Bad example:

• App is slow. What’s wrong?

You can also trigger an investigation from Slack. Mention Datadog in a message: @Datadog Investigate high CPU in ai-gateway in prod over the last 30 minutes. If invoked within a Slack thread, Bits AI SRE automatically uses the entire thread as investigation context.

Enable automatic investigations

In addition to manual investigations, you can configure Bits to run automatically when a monitor transitions to the alert state:

From the Bits AI SRE Monitors list

1. Go to Bits AI SRE > Monitors > Supported.
2. Toggle Auto-Investigate on for a single monitor, or bulk-edit multiple monitors by selecting multiple monitors, then clicking Auto-Investigate All.

For a single monitor

1. Open the monitor’s status page and click Edit.
2. Scroll to Configure notifications & automations and toggle Investigate with Bits AI SRE.

Supported monitors

Bits is able to run investigations on the following monitor types:

• Metric
• Anomaly
• Forecast
• Integration
• Outlier
• Logs
• APM (APM Metrics type only; Trace Analytics is not supported)
• Synthetics API and Browser tests (Preview)

How Bits AI SRE investigates

When Bits AI SRE investigates an issue, it operates in a continuous loop of observation, reasoning, and action. It begins by forming hypotheses about the potential root cause, then uses its tools to query telemetry data to validate or invalidate those hypotheses. Each step builds on prior findings. As new evidence emerges, Bits AI SRE updates its understanding, refines its reasoning, and chains together additional investigative steps—adapting and course-correcting until it converges on the most likely root cause.

At the end of an investigation, Bits AI SRE either presents a clear, evidence-backed conclusion or marks the investigation as inconclusive when the available data is insufficient to support a defensible conclusion.

Supported data sources

Bits uses the following data sources during investigations:

Datadog products

• Metrics
• APM traces
• Logs
• Dashboards
• Events
• Change Tracking
• Source code (GitHub only)
• Watchdog
• Real User Monitoring
• Network Path
• Database Monitoring
• Continuous Profiler

Third-party integrations

• Grafana
• Dynatrace
• Sentry
• Splunk
• ServiceNow
• Confluence

For best practices on maximizing the effectiveness of investigations, see Knowledge sources.

Investigation display modes

There are two display modes: Agent Trace and Investigation.

While an investigation is in progress, Bits captures every step it takes—including how it evaluates evidence and makes decisions—in the Agent Trace view. This provides a real-time, detailed record of the agent’s reasoning process.

Once the investigation is complete, you can switch to the Investigation view to explore a structured, tree-based visualization of the investigative path, making it easier to understand findings and conclusions at a glance.

Reports

The Reports tab enables you to track the number of investigations run over time by monitor, user, service, and team. You can also track the mean time to conclusion to assess the impact of Bits AI SRE on your on-call efficiency.