Skip to content

Dashboard User Guide

A comprehensive guide to using the Fulcrum dashboard for managing AI governance.

Overview

The Fulcrum Dashboard is a high-performance web administration interface built with Next.js for your AI governance platform. From here you can:

  • Monitor AI agent activity in real-time
  • Create and manage governance policies
  • Track budgets and spending
  • Review and approve pending actions
  • Investigate execution traces
  • Configure organization settings

Access the dashboard at: fulcrumlayer.io

Getting Started

1. Sign In

Sign in using your organization's SSO or email/password via Clerk authentication.

2. Organization Setup

If you're a new organization: 1. Create your organization profile 2. Invite team members 3. Set up your first agent connection

3. Connect Your First Agent

Use the SDK to connect your AI agent:

from fulcrum_governance import FulcrumClient

client = FulcrumClient(
    api_key="your-api-key",
    tenant_id="your-org-id"
)

# Your agent is now governed by Fulcrum policies

Dashboard Sections

Overview (/app/overview)

The Overview page provides a high-level view of your AI operations.

Key Metrics: - Active Agents - Number of connected agents currently running - Total Evaluations - Policy evaluations processed today - Violations - Policy violations detected - Pending Approvals - Actions awaiting human review

Quick Actions: - Deploy new policy - View recent alerts - Check budget status

Activity Feed: Real-time stream of agent events, policy triggers, and system alerts via NATS WebSocket.

Policies (/app/policies)

The heart of your governance strategy.

Features: - View all active policies in card format - Filter by type (Cost, Rate, Content, Approval, Model) - Search policies by name or ID - Sort by violations, priority, or date

Creating Policies: 1. Click Deploy New Policy 2. Choose a template or create custom 3. Configure rules and conditions 4. Set priority and scope 5. Click Deploy

Policy Cards Display: - Policy name and ID - Type badge with color coding - Active/Disabled status - Violation count - Last updated date

Templates: Quickly deploy common policies including: - Rate Limits - Cost Caps - PII Protection - Content Filters

See Policy Authoring Guide for detailed policy configuration.

Approvals (/app/approvals)

Human-in-the-loop workflow management.

Pending Queue: Actions flagged by policies requiring manual intervention: - High-value transactions - Sensitive data access - External communications

Approval Actions: - Approve - Allow the action to proceed - Deny - Block the action with optional feedback

Approval History: Track past decisions with: - Decision timestamp - Approver identity - Full action context - Policy that triggered review

Budgets (/app/budgets)

Control costs at the source.

Budget Overview: - Daily/Monthly spend graphs - Budget utilization percentage - Cost breakdown by agent/model

Thresholds: Set warning and critical alerts: - Notify at 80% - Block at 100%

Allocation: Assign budgets to: - Specific tenants - Departments - Individual workflows

Period Reset: Automatic resets for daily, monthly, or quarterly windows.

Traces (/app/traces)

Deep observability into execution history.

Event Timeline: Scrutinize every LLM call, tool invocation, and checkpoint.

Trace View: Each trace shows: - Request ID and timestamp - Agent and model used - Input/output content - Policy evaluation results - Cost and token usage

Filtering Traces: - By agent - By time range - By policy violation - By cost threshold

Decision Logs: See exactly which policy rule allowed or blocked an action.

Agents (/app/agents)

Manage your AI agent deployments.

Agent Cards: - View status (Running, Idle, Paused) - Connection health - Last activity timestamp

Group Management: Organize agents by workflow or department.

Agent Details: - Configuration settings - Assigned policies - Activity history - Performance metrics

Quick Links: Jump directly to traces for a specific agent.

Analytics (/app/analytics)

Data-driven insights for optimization.

Spend Trends: Visualize cost accumulation over time.

Model Efficiency: Compare performance vs. cost across different LLM providers.

Anomaly Detection: Spot statistical outliers in token usage.

Custom Reports: - Configurable date ranges - Multiple chart types - Export to CSV/PDF

Settings (/app/settings)

Configuration and security.

API Keys: - Issue new keys - Rotate existing keys - Revoke compromised keys

Tenant Config: Manage multi-tenant isolation settings.

Integrations: - Slack/Email notifications - LLM provider configuration - Webhook setup

Team Management: - Invite members - Assign roles (Admin, Editor, Viewer) - Remove access

Common Tasks

Investigating a Policy Violation

  1. Go to Overview and find the violation alert
  2. Click the violation to view trace details
  3. Review the input that triggered the policy
  4. Check if the policy needs adjustment
  5. If legitimate, update policy scope or conditions

Setting Up Cost Alerts

  1. Go to Budgets
  2. Click Set Budget or edit existing
  3. Configure alert thresholds (e.g., 50%, 80%, 100%)
  4. Choose notification method (email, Slack)
  5. Save configuration

Creating a Content Filter Policy

  1. Go to Policies → Templates
  2. Select a security template (e.g., PII Protection)
  3. Customize blocked patterns if needed
  4. Set scope (all agents or specific ones)
  5. Deploy the policy

Approving a Pending Action

  1. Go to Approvals
  2. Review the pending item
  3. Click to expand full context
  4. Make decision:
  5. Approve if action is legitimate
  6. Deny if action violates guidelines
  7. Add optional comment for audit trail

Debugging Agent Issues

  1. Go to Traces
  2. Filter by agent and time range
  3. Look for failed or denied requests
  4. Click trace for details
  5. Review policy evaluation results
  6. Adjust policies or agent configuration as needed

Demo Mode

The dashboard includes a demo mode for exploration without connecting live agents.

Enabling Demo Mode: 1. Sign in to the dashboard 2. Toggle "Demo Mode" in settings 3. Sample data will populate all views

Demo mode includes: - Sample agents and policies - Realistic trace data - Mock approval workflows

Design Language

The dashboard uses a "Swiss Industrial" aesthetic prioritizing information density and clarity.

Typography: - Monospace for data values - High-weight grotesque sans-serif for headings

Colors: - Deep black background (#050505) - Pure white text - Signal red (#FF3B30) for errors/alerts - Teal (#14B8A6) for success states

Troubleshooting

Dashboard Not Loading

  1. Clear browser cache
  2. Check your internet connection
  3. Try incognito/private mode
  4. Check status.fulcrumlayer.io

Data Not Updating

  1. Check WebSocket connection (green indicator in header)
  2. Refresh the page
  3. Verify agent is connected
  4. Check NATS server status

Can't Create Policies

  1. Verify you have Editor or Admin role
  2. Check organization permissions
  3. Ensure valid policy configuration

Approval Workflow Issues

  1. Verify policy is active
  2. Check approval conditions
  3. Review trace for evaluation details

Last Updated: January 20, 2026