TOMS™ Navigator
The Orchestration
Maturity Spectrum™
Five levels. From manual to autonomous. Where does your operation sit?
The Orchestration Maturity Spectrum (TOMS™) is a five-level practitioner framework for assessing where your supply chain operations sit on the journey from manual coordination to autonomous intelligence - and what it takes to move up.
Most operations today sit between L2 and L3. The dashboards are live. The AI pilots are running. But the coordination function - the routing of information and decisions across functions - is still human. TOMS™ maps the full journey from spreadsheets to self-orchestrating value chains, with practical signals at each level so you can assess honestly where you are today.
This framework is part of the Intelligent Orchestration Manifesto - the structural argument for why operations coordination is moving from people to systems.
Quick Assessment
Find your level in 4 questions
How are most operational decisions made?
At a Glance
| Level | Name | Decision Mode | Data Flow | Nerve Centre | Automation |
|---|---|---|---|---|---|
| L1 | Manual | Reactive, human-driven | Email, copy-paste | None | ~8% |
| L2 | Scripted | Rule-based if/then | Scheduled reports | Dashboard | ~28% |
| L3 | Integrated | AI-assisted, human-approved | Real-time APIs | Connected | ~56% |
| L4 | Orchestrated | AI-driven, human exceptions | Event-driven agents | Intelligent | ~78% |
| L5 | Autonomous | Self-optimizing, closed-loop | Execution feeds learning | Autonomous | ~95% |
1
Level 01
Manual
Human execution. Zero automation. Maximum friction.
Everything runs on spreadsheets, email threads, and tribal knowledge. Data lives in silos. Decisions depend on who has the right file open. The work gets done - but the cost is invisible: re-keying, version conflicts, delays that nobody tracks until something breaks.
Decision mode
Reactive. Human-driven on every step.
Data flow
Manual export, copy-paste, email attachments.
Visibility
Fragmented. Depends on who you ask.
Risk profile
High. Single points of failure everywhere.
SC in Practice
What this looks like in supply chain
Weekly status Excel sent by emailShipment ETA tracked in a shared docSupplier scorecard updated manually each quarterInventory counted on-site, entered by handDelay reporting via phone call
Nerve Centre
Visibility maturity
No nerve centre. Visibility is personal - whoever owns the spreadsheet owns the truth. Status updates happen in meetings, not systems. Firefighting is the default operating mode.
Tech signals
Typical stack
ExcelEmailShared DriveWhatsAppPhoneSAP GUI (manual entry)
Moving to L2: Identify the top 5 manual tasks by volume. Script the first one. A single Power Automate flow or Python script that eliminates one copy-paste process is the entry point.
2
Level 02
Scripted
Rules-based automation. Repeatable, but brittle.
Someone wrote the script - a macro, a scheduled query, a Power Automate flow. Predictable tasks run without human touch. But the logic is hardcoded. When the world changes, the script doesn't know. It executes what it was told, nothing more.
Decision mode
Rule-based. If/then logic predefined by humans.
Data flow
Scheduled pulls, automated reports, basic alerts.
Visibility
Consistent within defined scope. Blind spots remain.
Risk profile
Medium. Breaks silently when inputs change.
SC in Practice
What this looks like in supply chain
Automated daily KPI report via emailReorder triggered when stock < thresholdERP-to-Excel scheduled extractPower Automate approval flow for POsAlert when shipment delayed > 2 days
Nerve Centre
Visibility maturity
Dashboard nerve centre. Static reports and scheduled extracts. Power BI or Tableau pulling from ERP on a timer. You can see what happened yesterday - but not what's happening now. Alerts are threshold-based, not intelligent.
Tech signals
Typical stack
Power AutomatePython scriptsSQL scheduled jobsSAP ABAP reportsD365 workflowsPower BI
Moving to L3: Connect your first real-time API. Replace one scheduled extract with a live data feed. Add an AI layer that interprets - not just reports - what the data means.
3
Level 03
Integrated
Systems talk. AI assists. Humans still decide.
The silos start to dissolve. Systems connect through APIs, agents query live data, and AI surfaces patterns a human would miss or catch too late. The operator still holds the wheel - but now they have a dashboard, not a rear-view mirror.
Decision mode
AI-assisted. Human validates and approves.
Data flow
Real-time API calls, multi-source aggregation.
Visibility
Near-complete across core SC nodes.
Risk profile
Low-medium. Anomalies surface before failure.
SC in Practice
What this looks like in supply chain
Chat-driven SQL queries on live shipment DBAI-generated delay root cause summaryMulti-source KPI dashboard with alertsSupplier risk scored against news signalsNatural language reporting for ops teams
Nerve Centre
Visibility maturity
Connected nerve centre. Live data feeds from ERP, TMS, carrier APIs. Real-time shipment tracking. AI-powered anomaly detection flags exceptions before they escalate. Natural language queries replace manual report building.
Tech signals
Typical stack
n8n / MakeGemini / Claude APISAP BTPPostgreSQLREST APIsPower AppsWebhooks
Moving to L4: Deploy your first autonomous agent. Not AI that recommends - AI that acts within guardrails. Start with a low-risk domain: a procurement agent that auto-triggers an RFQ when inventory signals warrant it.
4
Level 04
Orchestrated
Multi-agent coordination. Self-adjusting. Humans set the intent.
No single agent. A network of specialized agents coordinates across functions - procurement, logistics, risk, finance - each watching its domain, handing off context, acting without being asked. Humans define the objectives and review exceptions. The system figures out how.
Decision mode
AI-driven within policy guardrails. Human for exceptions.
Data flow
Event-driven. Agents trigger on signals, not schedules.
Visibility
End-to-end, with forward-looking scenario modeling.
Risk profile
Low. System anticipates and routes around disruption.
SC in Practice
What this looks like in supply chain
Proactive rerouting when port risk spikesProcurement agent auto-triggers RFQ on stockout signalFreight agent compares carriers and recommends switchDaily briefing authored end-to-end by agent networkCross-function exception triage without human initiation
Nerve Centre
Visibility maturity
Intelligent nerve centre. The nerve centre doesn't just display - it acts. Agents monitor feeds, detect disruptions, propose alternatives, and execute approved responses. Scenario modeling runs continuously. The nerve centre becomes the orchestration layer itself.
Tech signals
Typical stack
Multi-agent n8nCopilot StudioClaude / GPT-4RAG + QdrantSAP S/4HANAD365 + DataverseEvent bus
Moving to L5: Close the loop. Connect execution outcomes back to the model. Let the system learn from every cycle - retraining demand models on actual outcomes, adjusting safety stock from real consumption, optimizing routes from delivered performance.
5
Level 05
Autonomous
Self-optimizing. Continuous learning. The art of the possible.
The supply chain thinks for itself. Agents learn from every cycle, re-train on new patterns, rewrite their own decision logic. Humans set strategic intent and audit outcomes. Every execution is informed by the full history of what worked - and what didn't.
Decision mode
Fully autonomous with continuous self-optimization.
Data flow
Closed loop. Execution feeds learning. Learning refines execution.
Visibility
Predictive and prescriptive. Future-state modeled continuously.
Risk profile
Minimal. Self-healing with human strategic oversight only.
SC in Practice
What this looks like in supply chain
Demand sensing with real-time model retrainingAutonomous contract negotiation within policy boundsSelf-adjusting safety stock across 1000+ SKUsCarbon-optimal routing updated every executionZero-touch supplier onboarding and qualification
Nerve Centre
Visibility maturity
Autonomous nerve centre. The nerve centre is the nervous system. It predicts disruptions before signals appear, self-heals routing and inventory, continuously optimizes cost vs. service vs. sustainability trade-offs. Human role: strategic oversight and governance only.
Tech signals
Typical stack
Reinforcement LearningFine-tuned LLMsDigital TwinSAP IBPReal-time streamsMLOps pipelineVertex AI / Azure ML
At the Edge: Even at L5, humans set strategic intent, govern ethical boundaries, manage supplier relationships built on trust, and make the calls where the cost of being wrong is existential. Read about The Edge in the Manifesto.
Frequently Asked Questions
What is TOMS?
The Orchestration Maturity Spectrum (TOMS) is a five-level practitioner framework for assessing how mature your supply chain operations are in terms of automation and intelligent orchestration. The five levels are: L1 Manual, L2 Scripted, L3 Integrated, L4 Orchestrated, L5 Autonomous.
How do I assess which level my operation is at?
Ask four questions: How are decisions made? How does data flow? What does your nerve centre look like? What percentage of operations run without human intervention? Use the quick assessment above or review each level in detail below.
What is the difference between L3 and L4?
At L3 Integrated, AI assists humans who still make the final decision. At L4 Orchestrated, a network of AI agents coordinates across functions autonomously within policy guardrails - humans only handle exceptions. The shift is from human-approved to AI-driven with human oversight.
What is intelligent orchestration?
Intelligent orchestration is the operating model at TOMS Level 4 and above - the shift from human-routed coordination to system-composed decisions across value chains. Read the full Intelligent Orchestration Manifesto.
Where does your operation sit?
The hierarchy served us well. It is time to move to the edge.