Skip to main content

Dashboard Overview

The WorldFlow AI dashboard is a browser-based control plane for your gateway deployment. It provides real-time visibility into cache performance, cost savings, model routing, MCP server health, audit trails, user management, and GDPR compliance -- all without writing API calls.

This guide describes the major sections, what each page shows, and how to access them.

Accessing the Dashboard​

The dashboard is served at the root URL of your WorldFlow AI deployment (e.g., https://dashboard.worldflowai.com). Authentication is handled through the login page, which supports both password-based login and SSO redirect (if configured). After login, a JWT token is stored in localStorage and sent automatically with every API call.

Main Analytics​

Dashboard (Home)​

The landing page after login. It shows a high-level summary of gateway activity:

  • Total requests processed in the selected time window
  • Cache hit rate with a trend sparkline
  • Cost savings in dollars, calculated from cache hits that avoided upstream LLM calls
  • Latency percentiles (P50, P95, P99) comparing cached vs. non-cached responses
  • Top models by request volume and cost

Cost Optimization​

A dedicated analytics page focused on cost reduction. It displays:

  • Cost savings over time (daily/weekly/monthly aggregation)
  • Savings breakdown by model and provider
  • Projected annualized savings based on current hit rates
  • Cache efficiency metrics per tenant

Routing Analytics​

Visualizes how the multi-model routing optimizer distributes traffic:

  • Request distribution across providers and models
  • Cost-per-request comparison between routed and default model selections
  • Quality score tracking over time
  • Fallback and retry rates

Model & Provider Management​

Model Catalog​

A browsable registry of all LLM models available in your workspace. Each model card displays:

  • Model name, provider, and family (e.g., GPT-4, Claude 3, Gemini)
  • Pricing (input/output cost per 1K tokens)
  • Capabilities (chat, tools, vision, streaming, JSON mode)
  • Context window size
  • Status (available, deprecated, custom)

You can filter by provider type, model type (chat, embedding), family, and status. Custom models can be created, edited, and deleted. The Sync Models button fetches the latest model definitions from all integrated providers.

Integrations​

Managed from the Model Catalog page. Each integration stores encrypted API credentials for one provider account. The page lets you:

  • Add new integrations (OpenAI, Anthropic, Google, Cohere, custom)
  • Verify that an API key is valid
  • View masked key values and last-verified timestamps
  • Delete unused integrations

Providers​

Provider cards show operational configuration layered on top of integrations:

  • Budget limits (monthly cap in cents) and current spend
  • Rate limits (requests per minute)
  • Priority ranking for routing
  • Health status and latency (P50/P99)
  • Enable/disable toggle

Routing Configuration​

Configure the multi-model routing strategy:

  • Routing mode selection (cost-optimized, quality-optimized, balanced, manual)
  • Per-model overrides for quality scores
  • Fallback chain ordering
  • Budget guardrails

Quality Profiles​

Define quality evaluation profiles that the routing optimizer uses to score model outputs. Profiles include evaluation criteria, scoring weights, and threshold settings.

MCP Management​

MCP Servers​

Register and monitor MCP (Model Context Protocol) servers. The page displays:

  • Server name, URL, transport type (HTTP, STDIO, SSE), and authentication method
  • Health status (PENDING, HEALTHY, DEGRADED, DOWN, DISABLED)
  • Latency metrics (P50/P99)
  • Last health check timestamp
  • Tool count and discovery status

You can add new servers, trigger health checks, run tool discovery, and configure per-tool caching.

MCP Entities​

Browse entity definitions used for cache invalidation. Entities have parent-child hierarchies, and writing to a parent entity invalidates all child caches.

MCP Analytics​

Charts and tables covering:

  • Cache hit rate per MCP tool
  • Tool invocation volume
  • Cache savings from tool result caching
  • Invalidation event frequency

Memory & Intelligence​

Memory Explorer​

A visual interface for browsing GCC (Global Context Cache) memory entries. Shows stored context segments with their embeddings, similarity scores, and expiration times. Useful for debugging why a cache hit did or did not occur.

Memory Projects​

Manage memory project configurations including branching, milestones, and contributor access.

Memory Analytics​

Charts showing memory usage trends, storage consumption, and retrieval performance.

Memory Logs​

Detailed log of memory operations (reads, writes, evictions) with timing information.

Intelligence Overview​

A summary view of the intelligence subsystem covering content indexing, source management, and search quality metrics.

Intelligence Sources, Content, Chat, Team, Settings​

Pages for managing intelligence data sources, viewing indexed content, interacting through a chat interface, managing team access, and configuring intelligence settings.

Guardrails & Safety​

Guardrails​

Configure content filtering and safety controls:

  • PII detection rules (email, phone, SSN, credit card, custom patterns)
  • Content classification policies
  • Blocked topic lists
  • Token limit enforcement
  • Custom filter rules with regex support

Each guardrail rule can be enabled/disabled individually and tested against sample inputs.

Logs & Observability​

Request Logs​

The primary log viewer for gateway traffic. Each row shows:

  • Timestamp, request ID, and model
  • Cache status (HIT, MISS, BYPASS, PARTIAL)
  • Latency (total, cache lookup, LLM call)
  • Token counts (prompt, completion)
  • Cost in cents

You can filter by date range, status, tier, model, and free-text search. Clicking a row opens the detail view with the full request/response bodies, similarity scores, and metadata.

The Similar Logs feature finds semantically related requests for a given log entry, which helps understand cache behavior.

Audit Logs​

A filterable timeline of all administrative actions:

  • Event type, actor, timestamp, outcome
  • Classification level
  • Resource affected
  • Full-text search with highlighting

Supports CSV and PDF export for compliance reporting. The detail view includes hash-chain integrity information.

Alerts​

Configure and view alerting rules based on anomaly detection:

  • Spend spikes, request volume spikes, error rate increases
  • Threshold-based triggers
  • Notification channel configuration

Export & Compliance​

Export​

Create and manage data exports:

  • On-demand log and metrics exports (CSV, JSON, Parquet)
  • Scheduled recurring exports (daily, weekly, monthly)
  • S3 destination configuration with connection testing
  • Export job history with status tracking (pending, running, completed, failed, cancelled)
  • Direct file download for completed exports

Privacy Settings​

Configure workspace privacy controls:

  • Data retention period (days)
  • PII redaction toggle and rule management
  • Encryption status overview (algorithm, key rotation info)

Data Subject Requests​

Manage GDPR data subject requests through a dedicated queue:

  • Create new requests (ACCESS, ERASURE, PORTABILITY)
  • Track request status (PENDING, IN_PROGRESS, COMPLETED, CANCELLED)
  • View processing notes and audit trail
  • Filter by status, type, and search term

Encryption Demo​

An interactive page demonstrating how WorldFlow AI encrypts data at rest. Shows sample plaintext alongside its encrypted form with algorithm metadata (IV, auth tag, key ID).

User & Access Management​

Users​

List, search, and manage workspace users:

  • Status filtering (ACTIVE, SUSPENDED)
  • Email and name search
  • User detail view with SSO integration fields
  • Suspend and delete actions

Roles​

View and manage RBAC roles:

  • System roles (read-only) and custom roles
  • Permission list per role
  • Create, edit, and delete custom roles

Permission Audit​

Inspect the effective permissions for any user:

  • Permission tree view showing all roles and their granted permissions
  • Single permission check tool (user + resource + action)
  • Visual inheritance display

RBAC Demo​

An interactive page for testing RBAC behavior against the live permission system.

SSO Configuration​

Configure and manage SSO connections:

  • Create OIDC or SAML connections
  • Domain management with verification status
  • Test and activate/deactivate connections
  • JIT provisioning settings

SCIM Dashboard​

Monitor SCIM provisioning:

  • Token management (create, revoke, delete)
  • Activity log with operation/resource/status filtering
  • Sync statistics (users, groups, error rates)

Configuration​

Gateway Configs​

Version-controlled backend configuration management:

  • View current active configuration
  • Create new config versions
  • Validate configuration before applying
  • Activate a specific version
  • View version history
  • Rollback to a previous version

Thresholds​

Cache similarity threshold tuning:

  • Global threshold settings (cache hit, partial hit)
  • Per-tenant overrides
  • Threshold simulator showing projected impact on hit rate and savings

Virtual Keys​

Manage Synapse-issued virtual keys for usage tracking and rate limiting:

  • Create, update, rotate, and revoke keys
  • Usage statistics with date range and granularity controls
  • Per-key analytics with anomaly detection
  • Team assignment

API Keys​

Manage encrypted provider API keys:

  • Add, rotate, and delete provider keys
  • View masked key values
  • Key validation status

Rate Limits​

Configure rate limiting tiers:

  • Tier definitions (requests per minute/hour, token limits)
  • Tenant-to-tier mapping
  • Current usage vs. limit visualization

Additional Pages​

Playground​

Test cache behavior interactively:

  • Send prompts and see whether they hit the cache
  • Preview similarity scores against existing cache entries
  • Test query variations to understand threshold sensitivity

Prompts​

Manage prompt templates with version control:

  • Create and edit templates with variable extraction
  • Version history and rollback
  • Template execution testing

Content Bucket & Editor​

Manage content collections and edit individual content items. Used with the intelligence and memory subsystems.

Fine-tune​

Configure and monitor model fine-tuning jobs.

Modules​

Enable and disable optional gateway modules (feature flags for subsystems like guardrails, routing optimizer, SCIM, etc.).

  • The left sidebar organizes pages into logical groups: Analytics, Models, MCP, Memory, Intelligence, Security, Compliance, and Settings
  • Most list pages support URL query parameters for deep-linking to filtered views
  • The top bar shows the current workspace name and the logged-in user
  • Real-time data (audit stream, health checks) refreshes automatically without page reload

Error Handling​

Dashboard API errors are displayed as toast notifications with the error type and message. All API calls use the same error format as the REST API:

{
  "error": {
    "type": "forbidden",
    "message": "You do not have permission to access this resource"
  }
}

If your session expires, you are redirected to the login page automatically.