EU Horizon Infraestructure Defense

AI Agent Demo for ERP / CRM Integration

Interactive HTML demonstration for a spare parts management environment. This simulated prototype shows how an AI agent can query availability, pricing, compatibility, order status, and preliminary diagnostics through a unified conversational layer connected to mock ERP, CRM, and knowledge services.

2Native Agent Modes

5Mock API Services

12Sample Parts Records

1Unified UI Simulation

Interactive Demo

Enter a request such as: “Do you have pump PX-100 in stock?”, “Price for valve kit VK-210 for customer C100?”, “Order SO-1004 status”, or “Machine overheats and pressure drops”.

Agent ecosystem mode Microsoft Native Agent Style SAP Native Agent Style Customer ID C100 - Standard Account C200 - Priority Account C300 - Distributor Account Request Check stock and price for PX-100. Also tell me whether it may match overheating and pressure loss symptoms.

Availability Pricing Compatibility Order Tracking Preliminary Diagnosis

Orchestration Console

System ready.

Reference Architecture

User / Customer / Employee
↓
Conversational AI Interface
Web portal, CRM workspace, service desk, internal assistant
↓
Agent Orchestrator
Intent detection, business rules, prompt shaping, tool selection
↓
Mock Enterprise APIs
Inventory API · Pricing API · Order API · Compatibility API · Knowledge API
↓
Enterprise Data Domains
ERP master data · CRM account context · Documents · Historic cases · Commercial rules
↓
Audit, security, logging, human escalation

Mock Data Snapshot

Part	Description	Stock	Base Price

Native Agent Script Layer — Microsoft Style

AGENT: CommercialTechnicalAssistant INPUT: user_query, customer_context STEP 1: detect_intent(user_query) STEP 2: extract_entities(part_id, order_id, symptom_keywords) STEP 3: call tools through connectors - inventory.lookup(part_id) - pricing.get(part_id, customer_id) - orders.status(order_id) - kb.search(symptoms) STEP 4: assemble grounded answer STEP 5: recommend next action STEP 6: log interaction to Dataverse / CRM timeline OUTPUT: - contextual response - escalation proposal - audit trace

Native Agent Script Layer — SAP Style

AGENT: JouleLikeServiceAssistant ON REQUEST: classify_intent() parse business entities trigger governed API call through integration layer IF availability_request: query inventory service IF pricing_request: retrieve price + customer condition IF order_request: fetch sales order status IF symptom_request: retrieve ranked matches from knowledge and compatibility layer RETURN: structured answer action proposal optional workflow handoff logged event

Why this demo matters

This visual prototype shows how an enterprise AI agent can sit above ERP and CRM platforms without replacing them. The agent acts as an orchestration and decision-support layer, transforming transactional systems, customer context, and technical documentation into a single operational conversation interface.

In a real deployment, the mock APIs shown here would be replaced by secure connectors, governed services, event-based integrations, and enterprise-grade identity, logging, and approval mechanisms.

Confidential draft. Demonstration only. This HTML includes simulated business logic and sample data for presentation purposes.