​

Intelligence API

ProBeya’s intelligence layer combines statistical analysis, AI-powered predictions, cross-site benchmarking, and organizational knowledge management. All intelligence features enforce multi-tenant isolation and use pure statistical methods where possible (no external AI API calls required for core analytics).

​

AI Insights (KPI Analytics)

Statistical anomaly detection and trend prediction for KPI time-series data using z-score analysis, moving averages, and linear regression.

​

Endpoints

​

GET /api/v1/intelligence/kpi-anomalies

Run anomaly detection on a single KPI’s historical data. Uses both z-score and moving average methods to flag unusual values.

tRPC: aiInsights.detectKpiAnomalies Auth: Bearer token required (scope: read:items) or session cookie Org context: Required

Parameters:

Response:

{
  "anomalies": [
    {
      "index": 42,
      "date": "2026-03-15",
      "value": 98.5,
      "expected": 72.3,
      "deviation": 36.2,
      "method": "zscore",
      "severity": "high"
    }
  ],
  "totalDataPoints": 90,
  "anomalyCount": 3
}

​

GET /api/v1/intelligence/kpi-trend

Run linear regression trend prediction on a KPI to forecast future values.

tRPC: aiInsights.predictKpiTrend Auth: Bearer token required (scope: read:items) or session cookie Org context: Required

Parameters:

Response:

{
  "slope": 0.42,
  "intercept": 65.1,
  "rSquared": 0.87,
  "trend": "improving",
  "predictedValues": [
    { "date": "2026-04-15", "value": 78.2 },
    { "date": "2026-04-22", "value": 78.6 }
  ]
}

​

GET /api/v1/intelligence/board-insights

Combined anomalies and trend analysis for all KPIs on a board. Returns a comprehensive AI-generated insight report.

tRPC: aiInsights.getInsights Auth: Bearer token required (scope: read:items) or session cookie Org context: Required

Parameters:

​

AI Project Intelligence

AI-powered project bootstrap templates and schedule prediction. Templates contain pre-built WBS structures, risk registers, and schedule frameworks for rapid project setup.

​

Endpoints

​

POST /api/v1/intelligence/bootstrap-templates

Create an industry-specific project template with pre-built WBS, risks, and schedule phases.

tRPC: aiIntelligence.createBootstrapTemplate Auth: Bearer token required (scope: write:items) or session cookie Org context: Required

Parameters:

WBS element structure:

{
  "elements": [
    {
      "code": "1.1.2",
      "name": "Protocol Development",
      "level": 1,
      "durationDays": 60,
      "dependencies": ["1.1.1"],
      "resources": ["Clinical Lead", "Biostatistician"]
    }
  ]
}

​

GET /api/v1/intelligence/bootstrap-templates

List bootstrap templates with optional industry and project type filters.

tRPC: aiIntelligence.listBootstrapTemplates

​

tRPC-Only Intelligence Procedures

​

Benchmarking

Cross-site KPI comparison for organizations with multiple sites. Enables executive-level comparisons across all level-1 workspaces (physical sites).

Data flow: workspaces (level=1) → projects → boards → kpiDefinitions → kpiValues

​

Endpoints

​

GET /api/v1/benchmarking/site

Compare a specific KPI across all sites using averaged values.

tRPC: benchmarking.getSiteBenchmark Auth: Bearer token required (scope: read:items) or session cookie Org context: Required

Parameters:

​

GET /api/v1/benchmarking/category

SQCDP scorecard per site showing a traffic-light matrix (red/amber/green) based on KPI thresholds.

tRPC: benchmarking.getCategoryBenchmark

​

GET /api/v1/benchmarking/trend

Time-series overlay of a KPI across all sites for trend comparison.

tRPC: benchmarking.getTrendComparison

​

GET /api/v1/benchmarking/ranking

Rank all sites by a specific KPI from best to worst.

tRPC: benchmarking.getRanking

​

Knowledge Forge

Internal knowledge base with articles, a knowledge graph of people/assets/processes, and expertise tracking per user per domain.

​

tRPC-Only Knowledge Forge Procedures

Node types: person, asset, product, process, problem, procedure

Edge relations: solved_by, applies_to, requires, supersedes, related_to

Article statuses: draft, published, archived

​

Error Codes