The Four Pillars
Legend


An Intelligent Container Journey
One Container. A Smarter Network.

The Four Pillars of AI: LLM, Agentic, Applied , Analytical
How to Read this Journey
Orientation

Welcome. This journey is designed to be read as an interactive decision lens, not a presentation deck.
How to use it: Explore in order, or jump to the sections most relevant to your role or questions.

What this journey does:

→ A single container is the thread. Following it shows how the network responds under pressure - how costs ripple through decisions, how one upstream change forces more downstream, how execution holds together or breaks.

→ Each section focuses on how the network behaves, not on technology for its own sake. AI appears where it changes outcomes, not where it simply exists.

→ This is a living reference, not a snapshot. The content will evolve as market dynamics, service conditions and operating realities change. You can come back to it as the environment shifts and find it current.

→ It's context, not a prescription. Use this to pressure-test your own assumptions and comparisons - not as a one-size-fits-all answer for every lane or operation.

Foundation: A Rosetta Stone for This AI Journey

A shared reference to how common industry AI terms map to the Four Pillars for this journey. Symbol Industry Term Our Pillar In this journey, think… LLM The Interpreter Reads and reconciles messy data into a trusted source of truth Applied AI The Optimizer Improves physical flow through vision, sensors, and automation Agentic AI The Conductor Executes routine decisions using your rules and guardrails Analytical AI The Sentinel Predicts risk and imbalance before they become failures Back to top

The Container Lifecycle Players
Four AI Pillars. Four Distinct Roles in the Journey.

Before a container ever moves, four forms of intelligence are in place. Each one plays a different role, acting at a different moment, but together they guide the journey end to end.

LLM Cleans Data

"The Interpreter"

Language models translate messy bookings, contracts, emails and messages into a single trusted source of truth, before the container moves.


Applied Shapes the Flow

"The Optimizer"

Continuously rebalances what happyens on the ground - yard flow, crane assignments, and terminal ops - to keep physical freight moving.


Agents Orchestrate the Handoff

"The Conductor"

Autonomous agents coordinate gate windows, driver instructions and handoffs in real time. The journey continues without human micro-management.

Analytics Anticipate Risk

"The Sentinel"

Looks ahead to predict ETAs, dwell risk, chassis shortages, and network imbalance so teams can act early instead of reacting late to failures.


Together, the Four Pillars of AI form a continuous intelligence layer that shifts execution from reactive problem-solving to predictive control at every handoff.
The Intelligent Journey in Context:
From Forecasting to Analysis

This presentation focuses on Goods Movement, the phase where AI's impact on cost, visibility and execution is most immediate and measurable.

1
Pre-Goods

Forecasting

2
Goods Movement

Planning | Execution | Reconciliation

3
Post-Goods

Analysis

Pre- and post-goods movements aren't covered here, but they could be. Should Part 2 of the The Intelligent Container Journey cover that? Let us know.

AI value emerges when observation, interpretation, orchestration and interaction work together across the entire process.


The Intelligent Container Journey
AI Through the Lens of the Intermodal Operating System

Follow a single container from origin to final delivery and see how the Four Pillars of AI for Intermodal operate together as a unified execution layer, removing friction, anticipating disruption and improving reliability and efficiencies across the network.

Operations Strategy
From Static Intermodal Plans to Adaptive Control
AI replaces fixed plans with a living operational mission by aligning decisions before the container ever moves.

• AI senses demand, capacity and constraints in real time. This enables the network to anticipate conflicts and shortages before they surface and become operational delays.
• AI aligns every mode to a shared operational goal. Rail, port, yard and final mile execution operate from the same continuously refreshed understanding of priority and risk.
• AI adjusts trade-offs as the journey unfolds. Decisions rebalance cost, speed and reliability dynamically as conditions change, rather than locking them in upfront.

End-to-End Execution
One Container. One Intelligence Layer. Zero Blind Spots.
A single container is guided across every handoff from ports to rail, yard and final mile by shared, predictive intelligence.
  • AI tracks and predicts the container across booking, port, rail, yard and final mile. Each handoff inherits full contextual awareness instead of starting from partial or stale information.
  • AI generates predictive ETAs at every node and transition. Downstream partners receive foresight, not just status updates, enabling proactive coordination.
  • AI detects exceptions early and resolves them automatically. Most disruptions are mitigated at the beginning, before they devolve into missed windows or service failures.
Network Velocity
Velocity becomes a network property, not a metric
AI preserves flow across the intermodal network by resolving friction before delays ever happen.
  • AI predicts early warning signals across the intermodal network. Leading indicators reveal emerging risks at ports, yards, rail and drayage before impact is felt downstream.
  • AI resolves friction before delays compound. Small disruptions are addressed at the start, preventing local issues from becoming systemic delays.
  • AI preserves flow across every mode. The network maintains momentum end-to-end, even as conditions change across individual segments.
As an operating system, AI aligns strategy, execution, and flow - replacing static plans with controlled, adaptive intermodal operations.
Four Pillars, One Intelligent Layer

These aren't experimental technologies. They are four practical capabilities that work together to turn fragmented intermodal execution into a coordinated, predictive operation. Here's how each player contributes behind the scenes.

LLM:
The Interpreter

Get Everyone on the Same Page.
LLMs read and reconcile messy operational information like emails, bookings, contracts, and instructions, and turn it into a single source of truth before the container moves.

LLMs get you:
  • Fewer booking errors
  • Fewer downstream surprises
  • Less manual checking


Examples: Chat GPT, Claude, Copilot

Agentic AI:
The Conductor

Coordinate Decisions at Speed.
Agents execute routine operational decisions automatically by dispatching drivers, adjusting appointments, and rerouting around disruptions using your rules.

Agentic AI results:
  • Dispatch doesn't wait on inboxes
  • Appointments adjust in real time
  • Exceptions are handled before customers notice


Agents turn insight into execution

Applied AI:
The Optimizer

Improve Physical Flow. Applied, also known as physical AI, optimizes what happens on the ground with gates, telematics, cranes, and inspections using cameras, sensors, drones and computer vision.

Applied AI provides:
  • Fewer mis-spots and rehandles
  • Better stack density
  • Faster gate and train builds

Cameras, scans, mobile and gate captures, inspections

Analytical AI:
The Sentinel

See Problems Before They Materialize.
This capability looks ahead to predict ETAs, dwell risk, chassis shortages, and imbalances so teams can act early instead of reacting late.

Analytical AI offers:
  • Higher on-time departures
  • Better chassis positioning
  • More credible customer commitments

Dashboards, SLAs, KPIs

Individually, each pillar improves local efficiency.
Together, they increase network velocity.
The Container Wakes Up

Before dawn, a single steel box waits in a quiet yard.

The paperwork is largely in place. Bills of lading filed, release messages sent, pickup numbers issued. Instructions live across familiar systems: TMS platforms, booking portals, email chains, yard management tools, customs filings. Each is doing its job though not always in perfect sync. A forklift hums past. A dispatcher checks a screen and prepares for the next move.

Downstream partners refresh their dashboards, trusting the information is current and ready to adapt when conditions change.

The container hasn't moved an inch, but the journey already carries complexity: coordination across five or more partners, equipment availability, yard conditions and handoffs between modes.

Each transition requires alignment. Small delays compound downstream. What begins as a physical journey becomes an information challenge.

Missing Documents

A single missing field becomes a customs delay.

Misaligned Appointments

A missed gate window cascades downstream and the entire lane absorbs the cost.

Yard Bottlenecks

A small congestion point becomes a service failure. The container sits while the meter runs.

Chassis Uncertainty

Availability, positioning, and release timing: visibility lags across all three.

None of these failures are dramatic. All of them are expensive.
The Intelligent Journey at a Glance
From first touch to final repositioning. Intelligence at every stage.
1
Origin

Container is prepared and released for the journey.

2

Gate & Terminal

Moves through check-in, staging, and terminal handling.

3
Rail Linehaul

Long-haul rail transport carries the container efficiently.

4

End Ramp

Arrives at the destination terminal for transfer.

5
Final Mile

Completes delivery to the final consignee.

6
Empty Return

Empty container repositions for the next journey.

The container moves. Intelligence acts at every handoff.
Origin:
Cleansing Data Before the Journey Starts
Booking & Order Intake: Shipper → Carrier

Problem this stage solves: Most downstream execution failures originate here.

LLM (Interpretation and Standardization)
• Converts unstructured emails, EDI variants, PDFs and portals into clean, structured order.
• Resolves ambiguity in equipment type, locations, and customer terminology
• Flags low‑confidence fields for human review

Agentic AI (Execution & Communication)
• Automatically generates and sends rate confirmations, BOLs, and contract summaries
• Acknowledges bookings and exceptions in natural language within minutes
• Learns customer‑specific booking patterns over time

Analytical AI (Foresight & Preparation)
• Forecasts lane‑level demand before orders are placed
• Enables early equipment positioning and rail capacity alignment
• Improves yield and reduces last‑minute premium moves

AI turns messy, unstructured intent into a single executable truth
before movement begins.
Origin Pickup & Loading:
Translating Intent into Motion
Dispatching the Trucker to Shipper

Problem this stage solves: This is where clean intent meets physical reality. Delays, missed rail cuts, and detention typically originate from poor coordination and limited visibility at the dock.

Agentic AI (Dispatch & Coordination)
• Automatically matches available drivers to loads using real‑time HOS, proximity, appointments, and cut‑off constraints
• Re-optimizes dispatch as conditions change (late releases, traffic, driver availability)
• Communicates dynamic instructions and exceptions to drivers and operations teams

Applied AI (Perception at the Dock)
• Predicts dwell time and flags containers at risk of missing rail cut‑offs
• Enables proactive intervention before service failures occur
• Improves on‑time rail tender and downstream velocity

Analytical AI (Execution Foresight & Prioritization)
• Predicts dwell time and missed rail cut risk before service failure occurs
• Quantifies detention, demurrage, and downstream cost exposure by load
• Directs proactive intervention to protect service and margin

Gate, Yard & Terminal:
Coordinating Before Congestion
Truck Delivery to Intermodal Facility

Problem this stage solves: Terminal congestion compounds fast. Poor gate sequencing, mis-spots, and suboptimal stacking increase dwell, delay train builds, and erode network velocity.

Agentic AI (Gate Execution & Control)
• Automated gate‑in agent reads OCR plate and container numbers
• Validates load, appointment, and equipment data in real time
• Assigns optimal stack position and routing without manual intervention
• Reduces gate queues and administrative friction

Applied AI (Physical Flow Optimization)
• Computer vision–guided crane operations optimize stack density
• Reduces mis-spots, re-handles, and unproductive crane moves
• Improves yard throughput without expanding physical footprint

Analytical AI (Terminal Foresight & Optimization)
• Optimizes slotting and container sequencing to minimize train build time
• Predicts congestion, crane contention, and yard imbalance before they occur
• Aligns yard decisions with downstream train schedules and service priorities
• Directs proactive re-sequencing to protect velocity and reliability

AI coordinates thousands of micro-decisions to preserve flow
before congestion compounds.
Rail Linehaul:
Optimizing the Invisible
Container Moves Cross-Country

Problem this stage solves: Once a container is on the train, visibility drops. Weather, crew availability, congestion, and interchange delays create uncertainty that’s often discovered too late.

Analytical AI (Predictive Foresight)
• Generates continuously updated ETAs factoring weather, track congestion, crew availability and interchange delays.
• Identifies dwell and delay risk across the linehaul before service failures occur
• Aligns downstream planning with realistic arrival windows

Agentic AI (Exception Handling & Coordination)
• Automatically executes pre-approved reroutes around embargoes or disruptions
• Coordinates responses across railroads and partners without manual dispatch escalation
• Preserves service commitments as conditions change

LLM (Communication & Alignment)
• Drafts and sends proactive customer notifications when ETAs shift beyond thresholds
• Explains delays in clear, contextual language using live operational data

Applied AI (Condition Monitoring)
• Detects equipment issues en route (hot boxes, shifted loads, reefer failures) using IoT and sensor fusion
• Flags risks early to prevent downstream disruption

Humans set intent. AI maintains controls across the invisible middle.
Destination Ramp:
Making the Future Visible
Train Arrives, Container Staged

Problem this stage solves: Arrivals are often "known too late." Chassis mismatches, yard congestion and static appointments create avoidable dwell and driver friction.

Analytical AI (Foresight & Preparation)
• Predicts container availability and dwell risk before train arrives
• Forecasts chassis demand by type and timing
• Aligns ramp priorities with downstream delivery commitments

Agentic AI (Outgate Coordination)
• Dynamically schedules truck appointments as containers are confirmed available
• Adjusts gate windows in real time as conditions change
• Communicates readiness and exceptions automatically

Applied AI (Condition & Claims Protection)
• Automates damage inspection at outgate using vision and sensors
• Captures pre-existing damage to prevent downstream disputes

AI makes arrivals actionable before the train ever stops.
Final Mile Delivery:
Keeping the Promise Intact
Trucker Delivers to Consignee

Problem this stage solves: Final mile execution is vulnerable to traffic, appointment drift and poor documentation, often erasing gains made upstream.

Agentic AI (Execution & Exception Handling)
• Adjusts routes and appointments dynamically based on live conditions
• Handles delivery exceptions without manual escalation
• Keeps drivers and operations aligned in real time

Analytical AI (Route & Cost Optimization)
• Optimizes multi-stop routing to reduce empty miles and fuel spend
• Quantifies detention and service-failure risk before delivery

LLM (Closure & Documentation)
• Automates POD capture and reconciliation against BOLs
• Closes orders cleanly in the TMS without manual follow-up

AI ensures the final mile keeps the promise made upstream.
Empty Return & Repositioning:
Treating Empties as Assets
Container Returned or Repositioned

Problem this stage solves: Empty containers are managed reactively, creating imbalance, wasted miles, and unnecessary repositioning cost.

Analytical AI (Imbalance Foresight)
• Forecasts empty demand and imbalance days ahead
• Identifies where assets will be constrained before shortages occur

Agentic AI (Repositioning Decisions)
• Executes optimal repositioning using backhaul, pools, and rail options
• Balances cost, timing, and service priorities automatically

LLM (Dispute & Exception Resolution)
• Automates EIR discrepancy review and dispute drafting
• Resolves documentation issues without manual back and forth

AI anticipates imbalance and repositions assets before the shortages appear.
From Journey to Network: How Individual Decisions Become Collective Intelligence
Every container generates data. Network intelligence is what happens when data works together.

The problem at this scale: Dwell compounds. Imbalances build. Discrepancies become disputes. None of it is visible from inside a single journey.

Analytical AI (Network-Wide Insight)
• Surfaces end to end dwell, transit and on-time performance in real-time, across lanes, terminals and partners.
• Turns fragmented data into a single, readable picture of network health.

LLM (Anomaly Detection)
• Flags revenue leakage, billing discrepancies and SLA breach risk before they escalate
• Identifies issues humans don't have the bandwidth to spot

Agentic AI (Network Execution)
• Executes pre-approved capacity rebalancing actions and coordinates moves across the network
• Requires minimal human oversight, using guardrails humans have set

Humans decide locally. AI ensures those decisions compound
into network intelligence.
The Network's Nervous System

As the container moves, an intelligent layer senses, checks and learns alongside it.

1
Monitor

Real-time SLA dashboards and operational indicators track progress and surface problems as signals, not surprises.

2
Assure

Human-in-the-loop checkpoints sit in front of high-impact decisions. Automation executes; humans govern.

3

Iterate
Every exception teaches the system something. Bad data is caught early. Recurring patterns become training signals. The network grows sharper with every container.

The network doesn't just move freight. It learns from it, one movement at a time.
Risks, Reality & Mitigation

AI introduces new failure modes - hallucinations, ontology errors, quiet drift - if deployed carelessly. Each of the six patterns below has a warning sign, a typical point of failure, and a way to design around it.

Isolated Local AI Wins

Risk: AI is applied to individual functions in isolation, optimizing locally without shared context across the journey.

Why it Matters: Local gains fail to translate into network velocity, shifting friction downstream instead of adding value.

Mitigation: Integrate all four AI pillars across the lifecycle with shared data, intent, and feedback loops.

Overreliance on Automation

Risk: Automation executes decisions beyond appropriate boundaries, reducing human oversight and allowing actions to drift from operational intent.

Why it Matters: Errors compound across the journey as conditions change, and trust breaks only after impact is felt.

Mitigation: Maintain human‑in‑the‑loop governance for high‑impact decisions and regularly review outcomes to ensure alignment with intent.

Poor Origin Data Quality

Risk: Predictions are built on noisy or inconsistent origin data, amplifying errors before the journey even begins.

Why it Matters: Bad inputs propagate downstream, degrading predictions, compounding mistakes, and eroding trust in the system.

Mitigation: Use LLMs and validation rules to clean, reconcile, and confirm data before movement, and feed corrections back into learning loops.

Untested Logic

Risk: Rules, prompts, or agent behaviors are deployed without being checked against in real data or SME constraints.

Why it Matters: Outputs drift from operational reality over time, allowing confident recommendations to violate domain rules silently.

Mitigation: Test logic against data and SME rules before deployment and build in continuous check points, before any damage is caused.

Trust Gaps

Risk: AI recommendations are ignored or overridden when operators do not understand or trust the system’s decisions.

Why it Matters: Low adoption prevents learning from compounding and keeps intelligence trapped at the local level.

Mitigation: Start in advisory mode, expose rationale clearly, and measure adoption alongside performance outcomes.

Siloed Data

Risk: AI decisions are informed by partial or siloed data rather than a unified network view.

Why it Matters: Optimizations appear correct locally but fail to compound into true network‑level intelligence.

Mitigation: Adopt shared data models, common identifiers, and cross‑domain feedback loops across the lifecycle.

AI doesn't fail all at once. It fails quietly, at the seams. Build your checkpoints there.
What the Journey Revealed
One container's journey shows us where alignment and intelligence matter most.

This journey followed a single container to make one idea clear:
intermodal performance is shaped less by individual moves and more by how well decisions are aligned across the network.

Across the journey, we saw that:

  • Most intermodal challenges are coordination and information problems, not physical ones.
  • Small gaps at handoffs can compound, while early alignment can preserve flow end‑to‑end.
  • Value emerges when strategy, execution, and velocity are connected through shared signals.
  • AI’s role is not to replace operations, but to support better decisions at the moments they matter most.

Taken together, the opportunity is not a new operating model but a clearer way to see where intelligence can reduce friction, anticipate disruption, and improve reliability within the intermodal system as it exists today.

Small moments of alignment shape the entire journey.
Open Invitation: Where Does the Journey Go Next? You Decide.
This isn't a conclusion. It's only Part One.

Part One followed a single container to clarify how decisions, signals and coordination shape intermodal performance during the core movement of freight.

From here, the journey can expand in several powerful directions. Click the arrows below to learn more:

Beyond the Core Move

Planning, readiness and feedback for pre- and post-goods movements

Deeper into Execution

Use intelligence to build trust, manage exceptions and base decisions for handoffs

Risk/Reward of Moving First

Early adopters absorb the risk but the compounding business rewards prove worthwhile.


Beyond the Core Move

The journey begins long before it moves and continues long after it arrives. Pre- and post-goods movements where planning, readiness, learning and feedback loops are where outcomes are won or lost. Perhaps Volume 2 explores how intelligence applied at these segments compounds the value of everything in between.




Deeper into Execution

Day-to-day intermodal performance is shaped less by systems than by decisions — who owns them, when they're made, and how exceptions get resolved across handoffs. This volume examines how trust, decision authority, and real-time intelligence can transform execution from reactive to genuinely coordinated.




Risk & Reward of Moving First

Early AI adoption in intermodal isn't without friction. It demands new thinking, new workflows, and tolerance for imperfect starts. But for those who move first, operational risk becomes a proving ground and the compounding returns in efficiency, resilience and competitive positioning make the case that waiting carries the greater cost.

Or something else. What would make the next part of the journey valuable to your work?

Help us shape Part Two.

Where AI lands next in intermodal will be decided by people running the freight, not the people writing about it.

Should we go beyond the core move, to pre and post-goods movement, deeper into execution or explore the risk/rewards - or somewhere we haven't named yet?

This work would not exist without the generosity of the intermodal community and these seasoned executives and innovators who gave their time, hard-won experience, and candor to help shape these ideas.

Our sincere thanks to the subject matter experts who contributed their knowledge and perspective:

This is a reflection of your collective wisdom. The next chapter will be too.

A Note on Creating This Journey: This presentation is a demonstration of human-AI collaboration in practice. AI helped shape how the story was told; humans determined what story was worth telling. We believe in being transparent about how we work, and this presentation is what a thoughtful partnership between people and AI can achieve.