Interactive Demo Guide

What to Do

Navigate to the main application and open the mapping editor. Select a source connector and target connector, then trigger a field mapping suggestion.

What to Notice

Each mapping row shows a confidence score with a breakdown into three components: semantic similarity, pattern matching, and historical accuracy. High-confidence mappings are auto-accepted; low-confidence ones are flagged for human review.

What It Proves

AI accuracy of 95-99% across 5 refinement phases. Explainable scores (not a black box) with human-in-the-loop approval.

Links / Commands

What to Do

Call the compliance summary endpoint to see reasoning trace statistics pulled from the database.

What to Notice

• totalSessions — number of AI sessions tracked
• totalTraces — individual reasoning steps recorded
• source: "database" — confirms data is DB-backed, not in-memory

What It Proves

Tier 1 #1 — Reasoning traces persist to the database and survive server restarts. Every AI decision is auditable.

Links / Commands

API call (requires JWT):

curl -H "Authorization: Bearer $TOKEN" \
  http://localhost:3000/api/compliance/summary

What to Do

Navigate to the SOC 2 Compliance Dashboard. Review each Trust Services category and its implementation status.

What to Notice

• CC6 Security — Implemented
• A1 Availability — Implemented
• PI1 Processing Integrity — Implemented
• C1 Confidentiality — Implemented
• P1 Privacy — Implemented

What It Proves

Tier 1 #4 — SOC 2 Trust Services aligned with live evidence. All five categories show "Implemented" with traceable evidence links.

Links / Commands

What to Do

Call the compliance export endpoint with format: "json" to generate the full auditor evidence bundle.

What to Notice

The JSON response contains four sections:

• reasoningTraces — all AI decision traces
• costTracking — per-provider cost records
• auditLog — system audit events
• soc2Mapping — Trust Services evidence mapping

What It Proves

Tier 1 #3 — Complete auditor evidence package exportable on demand. Meets SOC 2 Type II evidence collection requirements.

Links / Commands

Create dev JWT (required for compliance APIs):

TOKEN=$(node -e "const jwt=require('jsonwebtoken');console.log(jwt.sign({sub:'demo-user',username:'demo',roles:['admin'],permissions:['*']},process.env.JWT_SECRET||'test-jwt-secret-for-development-only',{expiresIn:'1h'}));")

curl -X POST http://localhost:3000/api/compliance/export \
  -H "Authorization: Bearer $TOKEN" \
  -H 'Content-Type: application/json' \
  -d '{"format":"json"}'

What to Do

Navigate to the Oracle Comparison page. Review the 8-row feature matrix comparing Oracle NSIP against SuiteCentral 2.0.

What to Notice

Oracle shows red X on all governance features: reasoning traces, SOC 2 evidence export, cost transparency, governance pacing, DLP/PII protection, approve-to-apply workflow, MCP native API, and NL Action Gate. SuiteCentral shows green check on all eight.

What It Proves

Tier 1 #2 — Clear competitive differentiation. Oracle's integration tool has no AI governance layer.

Links / Commands

What to Do

Call the Context Sidecar API with a vendor record. Try different record types: vendor, customer, invoice, purchaseorder.

What to Notice

• dataSource field — shows "service" when backed by real services, or "fallback" / "demo" when using deterministic demo data
• Insurance expiration checks with 30-day expiring-soon alerts
• Input validation: only whitelisted systems and valid record IDs accepted

What It Proves

Tier 2 #5 — Context Sidecar provides zero-click intelligence inside the ERP. Service-backed path tries real data first, falls back gracefully to deterministic demo data.

Links / Commands

curl "http://localhost:3000/api/context/netsuite/vendor/V-001"

What to Do

First, list available tools. Then call suitecentral.integration_status. If gateway is enabled, call suitecentral.mcp_discover and review tenant policy snapshot at /api/mcp/policies.

What to Notice

• Base tools are always listed: suitecentral.field_mapping_suggest, suitecentral.integration_status, suitecentral.governance_check
• Gateway tools appear when enabled: suitecentral.mcp_discover and suitecentral.mcp_call
• Standard JSON-RPC 2.0 response format with jsonrpc, result, id fields
• Policy + governance controls are enforced before external tool proxy calls

What It Proves

Tier 2 #6 + MCP Tier 3 — MCP-native API enables AI interoperability with governance depth: discovery + proxy patterns, tenant-scoped policy controls, and auditable gate checks.

Links / Commands

List tools:

curl -X POST http://localhost:3000/api/ai/proxy/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'

Call suitecentral.integration_status:

curl -X POST http://localhost:3000/api/ai/proxy/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"suitecentral.integration_status","arguments":{}},"id":2}'

Optional (gateway enabled): discover external tools:

curl -X POST http://localhost:3000/api/ai/proxy/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"suitecentral.mcp_discover","arguments":{}},"id":3}'

Tenant policy snapshot (auth required):

curl -H "Authorization: Bearer $TOKEN" \
  http://localhost:3000/api/mcp/policies

What to Do

Run the governed flow: parse natural language, propose the action, approve, then execute.

What to Notice

• cancel dispatches to SyncCentralService.cancelSubscription() — scoped to subscriptions only, validates active/pending status
• create-purchase-order dispatches to SupplierCentralService.createPurchaseOrder() — fuzzy vendor matching via normalizeName()
• Structured error codes: 404 (not found), 409 (conflict), 501 (not implemented), 400 (bad request)

What It Proves

Tier 2 #8 — 6 out of 6 NL Action Gate actions are operational. Real service dispatch (not simulated). Includes LLM intent parsing with regex fast-path and Claude Haiku fallback.

Links / Commands

Parse cancel intent:

curl -X POST http://localhost:3000/api/nl-action-gate/parse \
  -H 'Content-Type: application/json' \
  -d '{"input":"cancel subscription SUB-001"}'

Parse create-purchase-order intent:

curl -X POST http://localhost:3000/api/nl-action-gate/parse \
  -H 'Content-Type: application/json' \
  -d '{"input":"create purchase order for Acme Corp"}'

Then call POST /api/nl-action-gate/propose, POST /api/nl-action-gate/actions/:id/approve, and POST /api/nl-action-gate/actions/:id/execute.

What to Do

Navigate to the Executive Demo Center. Review the AI usage metrics, system health indicators, and module overview panels.

What to Notice

• AI usage metrics — request counts, provider distribution, cost per mapping
• System health — uptime, active connectors, database status
• Module overview — all production modules with status indicators

What It Proves

Production-ready monitoring and management — executives can see real-time system state, AI usage costs, and module health at a glance.

Links / Commands