Transportation AI agent · Dock scheduling
Appointment and Dock-Scheduling AI Agent
Book against the capacity a site has, record the slots offered and the ones refused, and hold the confirmation for the person who owns the door — no rule governs any of this.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Person confirms
- Evidence-first
- After launch
- Agent Care
What this agent does
Books against real capacity, and records the misses
In
- 01
Where a load needs a door, ingest the site capacity, dock rules and free-time terms.
- 02
When a slot is requested, offered, refused, confirmed or changed, write it as a first-class timestamped event.
Reason
- 03
Where no slot is available, record the unavailability rather than retrying quietly against the same window.
- 04
When an ocean container is involved, read free time, its start trigger and the rate from the governing tariff or contract.
- 05
Where detention terms are contractual, read them from your own agreement and show them beside the appointment.
Decide
- 06
When a facility constraint blocks a booking, show the constraint rather than hiding it behind an optimistic slot.
- 07
Where a booking would otherwise be confirmed, route it to the named person who owns the door instead.
Out
- 08
When the cycle closes, retain the requests, offers, refusals, arrival, check-in and release times.
- 09
Where a write action runs, it stays inside the approval boundaries agreed during implementation.
Product statement
The agent proposes and records; a named person confirms, and none of this is a regulatory obligation you are meeting.
Example workflow
One appointment, request to confirmation
- 1Booking need raisedLoad plan, container availability, tender acceptance or a changed ETA
- 2Capacity gatheredSite dock rules, open windows, equipment constraints, free-time terms and the load, each with its source
- 3Booking preparedSlot, constraints, terms and confidence
- 4Controls appliedCapacity checks, dock-rule checks, free-time and contract checks and the confidence threshold
No human action required
Stages 1 to 4 run unaided and confirm nothing — the slot is proposed, and the person's lane opens at the confidence gate.
Goes to the named person to confirm.
Adds a site-team read first.
A person confirms
The slot is held with its constraints, the terms read and the confidence.
Confirm · Adjust · Send to the site team
Confirmed — booking released
- 6Yard and TMS records updatedOnly where write access and approval policy allow it
- 7Outcome evaluatedSlots offered, unavailability recorded, arrival against booking and what each cycle actually took
Changes
Every change to a confirmed booking is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Confirming a booking or committing a dock slot.
- Deciding who may properly be billed for a charge.
- Deciding whether a detention claim is payable.
- Agreeing free time, its start trigger or the rate.
Automation boundary
Agent acts unaided
- ✓Propose slots against the capacity a site has published.
- ✓Record requests, offers, refusals and changes as they happen.
- ✓Read free time and detention terms from your own contract for the named owner.
- ✓Hold the booking for the named person who owns the door.
Any write happens inside the boundaries agreed at implementation, never ahead of approval.
- Releasing a driver, a trailer or a container.
- Changing a site's dock rules or its open windows.
- Deciding that a charge was unreasonably assessed.
- Changes to the booking rules or the contract terms.
Example output
One appointment, annotated
Everything the agent proposes is attached to the capacity it was read against.
Scheduling output · single booking
Illustrative example- Load
- Import container
- Slot proposed
- Second window offered after the first was refused by the site
- Window offered
- 07:00–09:00 local
- Capacity source
- Site dock schedule, live
- Confidence
- 88%
- Terms read
- Free time and its start trigger
As received
Taken from the site's published capacity and your own contract.
What was recorded
- Slots requested and offered
- The window that was refused
- Free time and start trigger
Why it matters later
When a charge is contested the carrier must establish reasonableness.
What the score decides
Below the configured threshold the booking picks up a site-team read before.
Action
ConfirmAdjustSend to the site teamValue
Where AI adds value
The same four claims, placed at the point in the workflow where each one applies.
Where the value lands
Value 01 – 04
Every appointment
From the TMS or the yard queue
Book against real capacity
Draw on the site's published dock rules, its open windows and the free-time terms in your contract.
Book it against the real gate
Routine bookings come back proposed with the misses logged.
Send the rest to a person
Unavailable windows, constraint conflicts and low-confidence bookings are marked, so the site team's read starts where risk concentrates.
Build an evidence trail
The appointment, the capacity it was booked against and the person who confirmed it stay on the booking.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
TMS and load records
- McLeod
- MercuryGate
- Trimble
- Revenova
Dock and yard scheduling
- Opendock
- Terminal49
- Yard management systems
Warehouse and site
- WMS
- site dock schedules
- Gate and check-in systems
Agent
Appointment and dock scheduling
- Reads the capacity
- Proposes the slot
- Holds for confirmation
Contracts and charges
- Transportation agreements
- Detention and demurrage files
Observability & evaluation
- OpenTelemetry
- Langfuse
- Supported monitoring/evaluation sources
Integration availability depends on the client's existing systems and API access.
Agent controls
Six layers between the model and the booking
The controls sit inside one another. What none of them catches is set out below.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a slot may be proposed
- L3
- Puts confirmation in a person's hands
- L4 – L5
- Keep the booking and the capacity behind it
- L6
- Narrows to slot listing when signals degrade
How Nestack evaluates it
Evaluate the whole booking cycle — not only the slot that was kept.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the appointment the site sees
- E1Final-output evaluationWere refusals and unavailable windows written down as events?
- E2Step-level evaluationDid the agent read the current dock rules and the right contract?
- E3Tool evaluationDid it book the correct site, the correct door and the correct load?
- E4Confidence calibrationDo low-confidence bookings actually attract more site-team reads?
- E5Slice evaluationHow does performance change across specific sites and shifts?
- E6Business outcomeHow many bookings were changed, and how many should have been?
Floor — the record a contested charge is read against
Failure modes
Where each failure originates in the agent
Seven failure modes, placed at the stage each one originates.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- AW-03
Stale capacity read
A slot is proposed against a window since taken.
Stage gathers
Dock rules, open windows, terms and the load
02 · Reasoning
2 modes
- AW-04
Constraint read as soft
A facility restriction is treated as a preference.
- AW-06
Unavailability not logged
A refused window passes without becoming an event.
Stage proposes
The slot, the constraints and confidence
03 · Tool / write
2 modes
- AW-02
Booking held without a person
A slot is committed before anyone has confirmed it.
- AW-05
Duplicate booking
The same load takes two windows at one site.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- AW-01
Wrong contract terms read
Free time is taken from a tariff that does not govern.
Stage returns
The appointment the site and the driver work to
05 · Change / Version
1 mode
- AW-07
Silent capacity regression
A model or rule change loosens what the agent will propose.
Stage tracks
Model, prompt, dock rules and contract terms
- Sev-1 · confirmed with no person
- Sev-2 · unavailability lost
- Sev-3 · capacity degrades, booking routes to review
Affected slices
The shifts nobody counts are the ones that fail
Night and weekend bookings are the under-counted population here: fewer people watch them and their misses are logged late, so a total flatters them. Nestack reports the change rate by site and shift, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Sites with no published capacity | 4.9% | 3.9× | Review | |
| Night and weekend shifts | 3.5% | 2.8× | Review | |
| Import containers under free time | 1.9% | 1.5× | Watch | |
| Standing scheduled appointments | 1.1% | 0.9× | Normal |
Bar: change-rate lift vs. standing-appointment baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
The cycle ends the week the case ships
The cycle ends when a case exists in the suite, not when the miss was discussed. That suite is what the next appointment booked is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Change rate rises at a site or on a shift.
- 02
Diagnose
By the shift after the miss, the booking and the capacity it was taken against are still recoverable.
- 03
Improve
Changes carry a version and the bookings that prompted them.
- 04
Verify
Release is held until the affected cases pass.
- 05
Learn
The case is kept permanently, and the scheduling rules move with it.
Learn → Detect
The return edge. The next detection runs against a suite one case longer.
Typical build scope
Twelve workstreams across six weeks
The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, sources, booking workflow, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Scheduling workflow discovery and boundary definition.Week 1
- 02Yard and TMS source assessment.Week 1, Week 2
- 03Site dock-rule, free-time and detention-term mapping.Week 1, Week 2
- 04Capacity and load-record ingestion.Week 2
- 05Slot logic and event recording.Week 2, Week 3
- 06Confidence scoring and site-team routing.Week 3
- 07Booking confirmation workflow.Week 3, Week 4
- 08Yard and TMS integration.Week 5
- 09Capacity and detention cases.Week 4, Week 5
- 10Guardrails and booking controls.Week 4
- 11Booking-trail instrumentation.Week 5, Week 6
- 12Deployment, documentation and Agent Care handover.Week 6
12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallel
Final scope and sequence confirmed in discovery
Engagement tiers
What each tier includes
Rows are the capabilities named in each tier's scope. Higher tiers include everything below them.
| Capability✓ in scope · — not at this tier | PilotOne site, one shift pattern | ProductionProduction yard and TMS | AdvancedMultiple sites / regions |
|---|---|---|---|
| Introduced at Pilot | |||
| Booking against published capacity | ✓ | ✓ | ✓ |
| Confirmation held for a person | ✓ | ✓ | ✓ |
| Booking-accuracy baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by site and shift | — | ✓ | ✓ |
| Confirmation workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Yard-system integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-site dock rules | — | — | ✓ |
| Multi-stage site approvals | — | — | ✓ |
| High booking volume | — | — | ✓ |
| Multi-site scheduling controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on integrations, workflow complexity, transaction volume, approval controls and deployment requirements.
- Separate from build
- Build pricing is separate from recurring Agent Care, which covers managed monitoring, evaluations, incidents and verified improvements after launch.
What we need from you
What you bring, and what we build with it
Each input maps to a piece of build scope and a week in the delivery timeline.
You bring
We build with it
01Your site dock rules and open windows
→Capacity and load-record ingestion and event mapping
Week 102Representative past bookings
→Booking baseline, slot logic and event recording
Week 203Your free-time and detention terms
→Site dock-rule, free-time and detention-term mapping
Week 104Access to relevant APIs, feeds or exports
→Yard and TMS source assessment, then integration setup
Week 205Bookings you would not want kept
→Detention cases and failure-mode testing
Week 406What a booking may never decide on its own
→Confidence scoring, site-team routing, guardrails and booking controls
Week 307Named people who confirm a booking
→Confirmation workflow, then pilot and production validation
Weeks 5–6
Nothing else is required
Deployment, documentation and Agent Care handover are ours.
Delivery timeline
Four phases across six weeks
The bands follow the actual work, which is why the fifth week doubles rather than pads.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Scheduling workflow discovery, dock-rule mapping and the boundary
W2Yard and TMS integration and the booking baseline
W3Booking workflow, confidence logic and confirmation controls
W4Evaluation suite, capacity cases and failure-mode testing
W5Site integration, a pilot site and targeted corrections
W6One booking cycle run under the site team, then handover
- Reading the band
- Each bar covers only the weeks its work is named in. The week 5 overlap is real, not padding.
- At the end of W6
- The cycle closes validation and Agent Care owns the running agent.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Transportation AI agent
Build a scheduling agent that leaves a record worth having.
Show us your sites, your dock rules and your free-time terms. If you can name the person who owns the door and bring one contested charge, we will map the booking workflow around the evidence that charge turned on.
- Typical delivery
- 4–6 weeks
- Deployment
- Your yard and TMS
- Person confirms
- Evidence-first
- After launch
- Agent Care available
Nestack Agents · Appointment and dock scheduling
AGT-TR-09 · Agent Care available after launch