AI Features
Waqti exposes a suite of AI-assisted procurement capabilities through dedicated endpoints and UI entry points: search in plain language, get vendor matches from descriptions, classify documents, extract invoice data, analyze POs for anomalies, and more.
Overview
The platform includes 14 AI-related endpoints (capabilities may be grouped in the product UI). A central AI dashboard typically shows:
- Feature flags — which AI modules are enabled for your tenant
- Usage statistics — how much you have consumed against plan limits
Plan tier
AI capabilities described here generally require Professional+ (or higher). AI chat is often limited to Professional and Enterprise tiers. Confirm your subscription with your administrator.
Natural language search
Ask questions or describe what you need in everyday language:
| Input | Behavior |
|---|---|
| Query | Your natural language question or search phrase |
| Entity type | Scope (for example POs, vendors) so results stay relevant |
Use this when you know what you want but not which filters to pick.
Vendor suggestions
Provide a short description of what you need to buy or the service context; the system returns vendor matches ranked for fit. Useful when onboarding new categories or exploring alternatives.
Document classification
Upload a file; the service returns a predicted document type (for example invoice vs contract). Helps route files into the right workflow without manual tagging.
Invoice extraction
Upload an invoice file; the system extracts structured invoice fields (amounts, dates, line hints, etc.) to reduce manual entry. Always verify extracted values before posting.
PO analysis (anomaly detection)
Run anomaly detection on purchase order data to surface unusual patterns (amounts, timing, vendors, or combinations). Use findings as signals for review, not automatic decisions.
Anomaly summary
A consolidated summary of detected anomalies helps managers and finance see risk areas without drilling into every alert.
AI chat
Conversational assistance for procurement questions and guided tasks (Professional and Enterprise plans where enabled). Availability depends on tenant configuration and limits.
Semantic search
Search by meaning, not only keywords—helpful for policies, clauses, or long descriptions where exact terms vary.
Contract analysis
Upload or select contracts to highlight key terms, obligations, or risk areas (exact outputs depend on model and configuration).
Spend forecast
Model future spend based on historical patterns to support budgeting and procurement planning.
Department spend forecast
Similar to spend forecast but scoped by department so budget owners see projections for their area.
Auto PO line generation
Generate suggested PO line items from a description or context (for example requisition text), speeding up data entry while keeping human review in the loop.
Human in the loop
AI outputs can be wrong or incomplete. Treat suggestions and extractions as drafts until a qualified user confirms them against source documents and policy.