STAGE 01 // INGRESS
Multi-Agent & MCP Gateway
Receives plain-English requests, voice notes, or BOM spreadsheets directly from the AI interface your team already uses.
- • OpenAI Actions / MCP Server
- • Gemini & Claude Tool Calling
- • WhatsApp Cloud Webhook
- • PDF / CSV BOM OCR Parser
STAGE 02 // NORMALIZATION
Deterministic Spec Compiler
Resolves ambiguous item descriptions into strict industry grading, units of measure, compliance standards, and dock deadlines.
- • SKU & UOM Canonicalization
- • ASTM / ISO / FDA Spec Tagging
- • Freight & Incoterm Resolution
- • Budget & Spend-Cap Validation
STAGE 03 // FAN-OUT
Zero-Portal Supplier Dispatch
Fans out parallel RFQ dispatches to approved roster vendors and vetted regional wholesalers with 1-click response links.
- • Signed Tokenized Quote Links
- • Email / SMS / EDI 840 & 843
- • Real-Time Stock & Lead Check
- • Automated Follow-Up Chasers
STAGE 04 // RANKING & GATE
Multi-Objective Scoring Engine
Ranks bids across landed cost, lead-time buffer, and historical supplier OTD (On-Time Delivery), then pauses for human sign-off.
- • Landed Cost Normalization
- • Deadline Risk Penalty Model
- • Role-Based Approval Routing
- • Immutable PO Audit Trail
// Compiled Kandy AI RFQ Payload (Generated from 1-sentence prompt)
{
"rfq_id": "rfq_mk_94820a",
"origin_channel": "chatgpt_mcp_v2",
"delivery_target": {
"zip": "08512", "city": "Cranbury, NJ",
"deadline_utc": "2026-11-20T17:00:00Z",
"dock_requirements": "palletized_48x40_standard"
},
"line_items": [
{ "item": "RSC Corrugated Box", "spec": "ECT-32 Kraft", "qty": 2000, "uom": "EA" },
{ "item": "Cast Stretch Film", "spec": "80ga 18in x 1500ft", "qty": 100, "uom": "RL" },
{ "item": "Nitrile Gloves", "spec": "4mil Powder-Free", "qty": 40, "uom": "CS" }
],
"governance": { "budget_cap_usd": 9000, "require_human_signoff": true }
}
Decision Intelligence
How Kandy Ranks Competing Bids
Unlike basic lowest-bid tools, Kandy evaluates the Total Cost of Risk—weighing unit price against your hard delivery deadline, partial-shipment penalties, and vendor track record.
Score = w₁(LandedPrice) + w₂(LeadTimeBuffer) + w₃(SpecCompliance) + w₄(VendorOTD)
Every recommendation returned to ChatGPT or WhatsApp includes the full trade-off rationale (e.g., paying 2.6% more to eliminate a 48-hour late-delivery risk).