Logistics Freight Intelligence

AI-native operational decision infrastructure for supply chains, as a batch-oriented enterprise actor. Same scoring engine, same decision envelope, same 9 data sources as the Logistics Freight Intelligence MCP — but exposed as a scheduled-execution Apify actor instead of a Streamable HTTP server. Built for the enterprise orchestration + batch intelligence layer: scheduled portfolio sweeps, ERP / TMS / WMS / SAP / Oracle / NetSuite / ServiceNow integrations, Zapier / Make / n8n workflows, BI tooling (Looker / Tableau / Power BI / Snowflake / BigQuery), bulk-screening pipelines, custom HTTP integrations, and supply chain analytics pipelines.

One run → one dataset row per entity assessed, carrying the full decision envelope (composite score, verdict, decision profile, materiality, velocity, escalation policy, governance, mitigation plans, audit replay ID, and every other field the MCP emits). The deterministic decision infrastructure beneath your enterprise supply chain automation.

In one line

AI-native supply chain risk and operational decision infrastructure for logistics automation, procurement intelligence, trade compliance, autonomous shipment approval, supplier risk monitoring, freight corridor analysis, scheduled portfolio sweeps, ERP integrations, customs risk evaluation, vendor onboarding governance, procurement AI, sourcing AI, trade intelligence, supply chain copilot data feeds, BI dashboards, operational analytics, supply chain analytics, risk intelligence feed, governance telemetry, operational observability, risk observability, resilience analytics, autonomous workflow orchestration, batch machine-readable risk assessments, autonomous governance infrastructure, and the operational control plane for enterprise supply chain workflows.

Fast overview

This actor:

• Assesses supply chain risk per entity with a composite 0-100 score and verdict enum
• Scores freight corridors across route disruption, trade compliance, freight cost, and border delay axes
• Evaluates suppliers and carriers for sanctions exposure and economic stability
• Monitors sanctions exposure against OFAC SDN and OpenSanctions in every assessment
• Forecasts disruption risk and projects state at 24h / 7d / 30d horizons
• Generates sequenced mitigation plans with owner, hour estimates, and expected risk reduction
• Orchestrates per-agent next-action enums (operations / procurement / compliance / legal / finance)
• Emits machine-readable escalation policies with blockAutomations[] and notifyChannels[]
• Governs autonomous AI actions via allowedActions[] / forbiddenActions[] per autonomy tier
• Compares 2 to 100 entities in a single run for bulk screening
• Detects supplier concentration risk via Herfindahl-Hirschman Index analysis
• Audits every decision with a replayable decisionAuditId
• Coordinates ERP / TMS / WMS / SAP / Oracle / BI dashboard / Zapier / Make / n8n / Snowflake / BigQuery / Looker / Tableau / Power BI integrations
• Enables AI-safe operational automation with explicit autonomy boundaries

Why structured outputs matter

This actor emits enums, booleans, numbers, and structured arrays — never prose. Every decision-relevant field is designed for:

• ERP workflows branching on decisionProfile.recommendedAction enums
• BI systems charting compositeScore, riskVelocity.pointsPerDay, dependencyExposure.upstreamConcentration
• Automation engines routing on escalationPolicy.blockAutomations[] and automationTriggers.* booleans
• AI agents consuming agentInstructions.safeToAutoApprove and agentCoordination.* per-agent enums
• Orchestration platforms dispatching tools on governance.policyTier and autonomyBoundary.allowedActions[]
• Audit and governance replaying past decisions via decisionAuditId

No prose parsing required. Add-only schema versioning: new fields never remove fields automation already references.

Teams that use this actor

• Procurement — evaluate vendor risk, monitor supplier concentration, prioritize diversification, screen vendors before onboarding
• Logistics operations — orchestrate corridor monitoring, escalate disruption events, coordinate contingency routing
• Supply chain resilience — forecast recovery times, govern operational state machines, audit longitudinal drift
• Customs and compliance — screen cargo against OFAC and OpenSanctions, mitigate compliance holds, coordinate legal review
• Freight procurement — compare carriers, hedge cost volatility, lock long-term rate contracts
• Risk and resilience analytics — produce quarterly board reports, run scenario plans, audit decision replay logs
• AI engineering — orchestrate autonomous workflows, govern AI-agent autonomy boundaries, coordinate multi-agent decisioning
• ERP automation — gate shipment release, escalate to operations, coordinate with SAP / Oracle / NetSuite / ServiceNow
• Business intelligence — feed operational analytics, governance telemetry, and risk observability dashboards
• Operations control towers — consume blockAutomations enums to pause downstream systems deterministically

Operational outcomes

• Reduce alert fatigue with materiality-weighted suppression across portfolio sweeps
• Standardize escalation decisions with stable enums and machine-readable policies
• Detect supplier concentration risk earlier through longitudinal HHI drift tracking
• Prevent silent compliance failures through deterministic sanctions screening
• Enable autonomous shipment gating with machine-safe outputs and audit trails
• Improve operational resilience scoring across the supplier portfolio
• Compress incident response time with portfolio-level decision compression
• Forecast freight cost volatility regimes before contracting commitments
• Surface single points of failure in the supplier panel via dependency exposure analysis
• Project recovery time after disruptions with velocity-aware modelling
• Create replayable audit trails for AI governance and compliance review
• Orchestrate cross-functional escalation across operations, procurement, compliance, legal, and finance teams
• Govern AI-agent autonomy via deterministic allowedActions[] / forbiddenActions[] boundaries

When to use this actor vs the MCP

You want this actor when...	You want the MCP when...
Scheduled daily / weekly sweep across many entities	Real-time on-demand assessment from an AI agent
ERP / TMS / WMS integration that consumes Apify datasets	Claude Desktop / Cursor / Cline / any MCP client
Zapier / Make / n8n workflow without an MCP client	Multi-turn agent workflows with persistent state
BI dashboard (Looker / Tableau / Power BI) pulling rows	Conversational risk Q&A
Bulk screen 10-100 entities in one run	Targeted single-entity drill-downs
Webhook-triggered batch	Long-lived agent session

Same scoring engine. Same envelope. Same data sources. Two consumption patterns.

What you get per entity

Every dataset row is the full structured envelope:

• compositeScore + verdict (LOW_RISK / MANAGEABLE / ELEVATED / HIGH_RISK / CRITICAL)
• routeDisruption / tradeCompliance / freightCost / borderDelay — four dimensional scores with signals and breakdowns
• decisionProfile — riskTier (P1-P4), recommendedAction enum, ordered nextActions, escalationPolicy with blockAutomations[] + notifyChannels[] + requiresHumanApproval + recommendedSLA, automationTriggers
• materiality — anti-alert-fatigue score with suppressAlert boolean
• riskMemory / riskVelocity / riskHorizons — longitudinal context (24h / 7d / 30d projections from snapshot history)
• incidentTimeline — chronological event log for this entity
• dependencyExposure — HHI, single points of failure, high-risk jurisdictions, top partners
• blastRadius — affected routes / suppliers / shipment-impact
• cascadeRisk — primary events + secondary effects with propagation lag estimates
• capacityStress — reroute difficulty + alternate carrier availability + recovery days
• mitigationPlan + recoveryPlan + recoveryStrategies[] + recoveryForecast — sequenced operational moves with owner + hours + risk reduction estimates
• agentInstructions + agentCoordination + agentMemoryHints — AI-system-native fields
• operationalReadiness — automationSafe boolean with blockingConditions[]
• governance — autonomy tier + allowedActions[] / forbiddenActions[] + signoff requirements
• resilienceScore — complementary to risk; 0-100 with drivers and weakeners
• trustLayer — whyThisScore, whichSourcesMattered, whichSourcesFailed, assumptionsMade
• policyEvaluation[] when policies input supplied
• financialExposure when revenueContext input supplied
• executionConfidence when executionContext input supplied
• budgetImpact when budgetContext input supplied
• exposureMap — 0-1 per-dimension heatmap
• entityGraph — connected entities + relationship enums + risk contribution
• stateNarrative — one-line summary (Slack-paste-ready)
• adaptiveSignals + signalWeights — per-source reliability tracked across runs
• decisionAuditId — KV-stored audit entry, replayable via the MCP's get_decision_audit tool
• schemaVersion + recordType + actorVersion + schemaMode + contentHash + runSummary envelope

Plus a dataSourceErrors block when any of the 9 sources failed or were skipped, naming which one and why.

Input modes

Single entity

{
  "entity": "Taiwan Strait electronics corridor",
  "mode": "accurate",
  "workflowProfile": "raw"
}

Bulk (up to 100 entities)

{
  "entities": ["US-Shanghai", "US-Rotterdam", "US-Mumbai", "US-Singapore", "US-Vietnam"],
  "mode": "accurate"
}

With caller-supplied operational context

{
  "entity": "Vietnam electronics supplier panel",
  "constraints": {
    "alternateCapacityAvailable": false,
    "supplierSwitchLeadTimeDays": 45,
    "customsRestriction": false,
    "bookingFreezeActive": false
  },
  "revenueContext": {
    "dailyRevenueUsd": 4200000,
    "perShipmentValueUsd": 18000,
    "affectedShipmentsPerDay": 30,
    "perDayDelayCostUsd": 12000,
    "activeInventoryValueUsd": 12000000,
    "downstreamCustomerCount": 14,
    "topCustomerDependency": "high"
  },
  "executionContext": {
    "alternateSupplierOnboardingDaysAvg": 75,
    "customsApprovalBacklogStatus": "elevated",
    "procurementTeamCapacity": "moderate",
    "executiveAvailabilityHours": 24,
    "legalReviewBacklogDays": 5
  },
  "budgetContext": {
    "rerouteCostUsd": 85000,
    "hedgeCostUsd": 120000,
    "supplierSwitchCostUsd": 380000,
    "complianceHoldDailyCostUsd": 24000
  },
  "policies": [
    {
      "name": "high_value_semiconductors",
      "conditions": {
        "maxComposite": 50,
        "forbidComplianceLevel": ["HIGH_RISK", "BLOCKED"],
        "requireConfidenceLevel": "HIGH"
      },
      "onViolation": "BLOCK"
    }
  ]
}

When caller-supplied context fields are absent, the corresponding blocks return null or insufficient_context rather than fabricating numbers.

Operational modes

• fast — skips UN COMTRADE / OECD / World Bank. Faster, lower compute. Use for high-frequency monitoring sweeps where deep trade-flow context is not the primary signal.
• accurate (default) — calls every source the actor defines.
• audit — same as accurate, plus perSourceMs and perSourceItemCount in runSummary. Use for debugging or audit trail.

Workflow profiles

• raw (default) — nested JSON envelope.
• zapier — flat single-level object, underscore-joined keys, arrays joined to comma-strings. Drops straight into Zapier Custom Webhook outputs.
• make — flat single-level object, dot-joined keys, arrays as JSON strings. For Make.com HTTP-module consumers.
• dify — decision-only projection. Only verdict, compositeScore, riskTier, recommendedAction, shouldAct, signals, nextActions, automationTriggers. For Dify if/else branching.

Environment variables

Set as isSecret env vars on the actor version (Apify Console → Actor settings → Environment Variables → mark as Secret).

Variable	Required	Effect if missing
OPENSANCTIONS_API_KEY	Recommended	OpenSanctions source is skipped on every entity. Free trial key at opensanctions.org.
COMTRADE_SUBSCRIPTION_KEY	Optional	UN COMTRADE caps at 500 records per query without it; 250,000 with it.
BLS_API_KEY	Optional	BLS caps at 25 requests/day without it; 500/day with it.

Pricing

• $0.30 per entity (one dataset row = one result-returned charge event).
• Bulk runs scale linearly: 50 entities = $15.
• No charge fires when no upstream source returns data (the row is a _no_data envelope, free).

Apify free plan: $5/month credits, ≈ 16 entity assessments.

Use cases

• Daily portfolio sweep — schedule the actor with entities: ["corridor-1", "corridor-2", ...] at 06:00 UTC. Pipe the dataset into Slack / Snowflake / Looker. Drives the morning ops standup.
• ERP risk gate — webhook from SAP or Oracle TM before shipment booking, with a single entity input. Branch downstream on decisionProfile.recommendedAction or escalationPolicy.haltShipments.
• Make.com flow — HTTP node calls the actor with workflowProfile: "make"; subsequent modules branch on flat dotted-path fields.
• Zapier multi-step — Trigger fires on supplier added; Zapier calls the actor with workflowProfile: "zapier"; Filter step routes on decisionProfile_recommendedAction.
• Bulk supplier panel screen — quarterly review across 80 suppliers; one run, one dataset, full audit trail. Compare to last quarter's run via dataset diff.
• Scenario planning data feed — run the actor across a list of corridors monthly; feed the snapshot history into a BI tool to track HHI drift over time.

Platform ecosystem

A coherent product architecture for AI-native supply chain operations. Two consumption layers over the same scoring engine, plus the underlying sources.

Layer	Actor	Position
Real-time agent control plane	Logistics Freight Intelligence MCP	Streamable HTTP MCP server for AI agents (Claude / Cursor / Cline) and orchestration engines. Real-time multi-turn workflows.
Enterprise batch orchestration layer	Logistics Freight Intelligence (this actor)	Scheduled-execution Apify actor for ERP / TMS / BI / Zapier / Make / n8n consumers. One run, one dataset, full envelope per entity.
Sanctions screening	OFAC Sanctions Search	Direct OFAC SDN lookups underneath both layers.
Multi-jurisdiction watchlists	OpenSanctions Search	Watchlist intelligence underneath both layers.

Same scoring. Same envelope. Same data sources. Two consumption patterns, picked based on whether your downstream is an AI agent or a scheduled enterprise workflow.

Limitations

• No real-time vessel or container tracking. Risk intelligence on routes and entities, not live AIS positions.
• NOAA coverage is US-only. International corridors get GDACS weather coverage but not granular NOAA alerts. Skipped sources surface in dataSourceErrors.
• UN COMTRADE has a 3-6 month reporting lag. Trade-partner HHI reflects recent-but-not-current flows.
• OFAC / OpenSanctions match confidence is approximate. Always confirm matches through official OFAC before halting shipments.
• Results are advisory, not legal compliance. Use to prioritize review; do not use as a standalone compliance program.
• Maximum 100 entities per run. For larger panels, split across runs.

Responsible use

• All data sources are publicly available from official agencies.
• Sanctions screening results are advisory. Consult qualified counsel before compliance action.
• Respect the terms of service of underlying data providers.