Skip to main content

Quick Start Guide

This guide walks you through your first steps with LangGuard. By the end, you'll have connected an integration and explored your first traces.

Step 1: Sign In​

  1. Go to your organization's LangGuard URL
  2. Click Sign in with Google (or use your SSO provider if configured)
  3. Authorize LangGuard to access your account
First Time?

If your organization uses LangGuard, ask your admin for an invitation. If you're starting fresh, signing in creates your workspace automatically.

Step 2: Explore the Dashboard​

After signing in, you'll see the main dashboard:

  • Discovery - Discover and inventory AI assets
  • Monitoring - Overview metrics and recent activity
  • Trace Explorer - Search and analyze traces
  • Policies - Manage governance rules
  • Data Catalog - Browse discovered entities
  • Integrations - Configure data sources (Admin only)
  • Settings - Workspace configuration, users, SSO, and more (gear icon)

Dashboard Overview​

The dashboard shows key metrics at a glance:

  • Active Integrations - Connected data sources
  • Recent Traces - Latest agent activity
  • Policy Violations - Recent governance issues
  • Key Metrics - Success rates, latency, costs

Step 3: Connect Your First Integration​

To see your AI agent data in LangGuard, connect an integration.

  1. Click Integrations in the sidebar
  2. Click Add Integration
  3. Browse categories and select your platform (e.g., Databricks)

Enter Credentials​

For Databricks as an example:

  1. Name: Give it a friendly name (e.g., "Production Databricks")
  2. Host URL: Your workspace URL (e.g., https://dbc-xxx.cloud.databricks.com)
  3. Access Token: Your personal access token (starts with dapi)
Where to Find Databricks Tokens
  1. Log in to your Databricks workspace
  2. Click your profile → User Settings
  3. Go to Access Tokens → Generate New Token

Test and Save​

  1. Click Test Connection to verify credentials
  2. If successful, click Save Integration
  3. The integration appears in your list with status "Connected"

See Connecting Integrations for detailed guides for each platform.

Step 4: View Your Data​

How traces arrive in LangGuard depends on the type of integration you connected:

  • OTLP-based integrations (LangChain, CrewAI, Claude Code, Cursor, OpenCode) — Traces are pushed to LangGuard in real time as your agents run. No sync step is needed.
  • Platform integrations (Databricks, Azure AI Foundry, AWS Bedrock) — Traces are delivered via webhook in real time through LangGuard's OpenCITE connector.
  • Polling integrations — Some integrations periodically fetch new traces. The default interval is 15 minutes, configurable from 1 minute to 1 hour. When no users are active, the interval automatically backs off to conserve resources.

For polling integrations, you can see the Last Sync time and Sync Status on the integration card. Click the Sync button (refresh icon) to trigger an immediate sync.

Step 5: Explore Your Traces​

Open Trace Explorer​

  1. Click Trace Explorer in the sidebar
  2. You'll see a table of all synced traces

Use the controls to find specific traces:

  • Time Range: Last hour, 24 hours, 7 days, etc.
  • Status: Success, Error, Warning
  • Agent: Filter by agent name
  • Search: Full-text search across trace content

View Trace Details​

  1. Click any trace row to open the detail drawer
  2. Explore the tabs:
    • Overview - Key metrics and metadata
    • Timeline - Span breakdown
    • Input/Output - Request and response data
    • Violations - Any triggered policies

Step 6: Enable Policies (Optional)​

Policies automatically detect issues in your traces.

  1. Click Policies in the sidebar
  2. Browse available policies:
    • PII Detection
    • Credential Surface Discovery
    • Unapproved Tool Use
    • And more...
  3. Toggle Enabled for policies you want active
  4. View violations in Policies > Violations

You're All Set!​

You've successfully:

  • Signed in to LangGuard
  • Connected an integration
  • Explored traces and agents
  • Learned about policies

What's Next?​

Need help? Contact info@langguard.ai or check our FAQ.