AI Insights

The AI Insights feature analyzes your business data and surfaces actionable observations automatically. It combines LLM-powered analysis with rule-based fallback to ensure reliability.

How It Works


KPI Data  -->  Rule Engine  -->  Insights
              |                    ^
              v                    |
           LLM Analysis  ---------+
           (when available)

Data collection — The system aggregates KPIs from all active modules (revenue, pipeline, inventory levels, HR metrics)
Rule-based analysis — A set of predefined rules evaluates the data for common patterns (thresholds, trends, anomalies)
LLM analysis — When an AI provider is configured, the data is sent to the LLM for deeper, contextual analysis
Merge and deduplicate — Results from both engines are combined, deduplicated, and ranked by relevance
Display — Insights appear on the Dashboard and in module-specific views

Rule-Based Fallback

The rule engine runs independently of any AI provider, ensuring insights are always available:

Rule	Trigger	Example Insight
Revenue trend	MoM change > 10%	“Revenue increased 23% compared to last month”
Overdue invoices	Count > 0	”5 invoices totaling SAR 120,000 are past due”
Pipeline stagnation	Deal unchanged > 30 days	”3 deals have not progressed in over 30 days”
Low stock	Below reorder point	”4 products are below minimum stock levels”
Payroll variance	Change > 5% from prior run	”Payroll cost increased 8% this month”
Attendance anomaly	Absence rate > threshold	”Absence rate is 12%, above the 5% threshold”

LLM-Powered Analysis

When configured, the LLM provides richer, contextual insights:

Correlation detection — “Revenue growth correlates with the 3 new deals closed this quarter”
Forecasting — “At the current burn rate, Project X will exceed budget by week 6”
Recommendations — “Consider increasing reorder quantity for SKU-1234 based on seasonal demand”
Narrative summaries — Natural language summaries of financial reports

Configuration

Set the AI provider in Settings > AI Configuration:

Provider — OpenAI, Azure OpenAI, or Anthropic
API key — Your provider API key (stored encrypted)
Model — Model selection (e.g., gpt-4o, claude-sonnet)

If no provider is configured, the system falls back to rule-based insights only.

KPI Analysis

The system tracks and analyzes these KPI categories:

Category	Metrics
Financial	Revenue, expenses, profit margin, cash flow
Sales	Pipeline value, conversion rate, average deal size
Operations	Inventory turnover, order fulfillment rate
HR	Headcount, turnover rate, absence rate
Projects	Velocity, burn rate, on-time delivery

Privacy and Data Handling

KPI data sent to the LLM is aggregated and anonymized — no customer PII is included
Raw transactional data is never sent to external providers
All AI requests are logged in the audit trail
AI features can be fully disabled per company

API Endpoint


GET /api/v1/ai/insights?companyId=COMPANY_UUID

Returns an array of insight objects:


{
  "insights": [
    {
      "id": "ins_abc123",
      "category": "financial",
      "message": "Revenue increased 23% month-over-month",
      "severity": "info",
      "source": "rule",
      "createdAt": "2026-04-13T10:00:00Z"
    }
  ]
}