Transportation AI agent · Vessel emissions
Vessel Emissions Compliance AI Agent
Split one consumption dataset across the four instruments it answers to — EU MRV, EU ETS, FuelEU Maritime and IMO DCS — each on its own deadline, set by a different body.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Verifier first
- Four deadlines
- After launch
- Agent Care
What this agent does
Assembles the record, never the assessment
In
- 01
A bunker is delivered, and the note behind it has to reach the voyage leg that actually burned the fuel.
- 02
A voyage leg closes, and the scope split decides whether it counts in full or at half.
Reason
- 03
A reporting year ends, and the MRV report is due 31 March 2026, or earlier if the authority says so.
- 04
A window opens 31 January 2026, and the ship-specific FuelEU report goes to the verifier.
- 05
A surplus is banked or a pool is formed, and 30 April 2026 shuts the window for recording it.
Decide
- 06
A consumption total is aggregated, and the Administration issues the Statement of Compliance by 31 May 2026.
- 07
A surrender falls due 30 September 2026, and the phase-in share, not the whole year, is what is covered.
Out
- 08
A monitored gas becomes a surrendered one: methane and nitrous oxide enter the obligation in 2026.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent assembles and splits the record. An accredited verifier assesses it; the designated company person submits the report and surrenders the allowances.
Example workflow
One voyage leg, record to verifier
- 1Consumption record receivedBunker delivery notes, flow meter readings, noon reports and port call records
- 2Voyage context assembledThe vessel, the leg it sailed, the instrument the data answers to and the scope that leg carries
- 3Report state draftedThe legs, their scope split and provenance
- 4Controls appliedProvenance checks, scope-split checks, deadline checks and completeness confidence
No human action required
Stages 1 to 4 run unaided, and nothing is verified at any of them — the agent is assembling, and the compliance lane opens at the completeness gate.
Goes to the designated company person to submit.
Adds a marine compliance read first.
Compliance review
The report is held with its gaps, its legs and the instrument it is owed to.
Submit · Append record · Send to compliance
Submitted — by the designated company person
- 6Vessel and voyage records updatedOnly where write access and records policy allow it
- 7Outcome evaluatedProvenance coverage, scope accuracy, compliance corrections and what review found
Corrections
Each compliance correction is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Assessing whether a report is free from misstatement.
- Submitting the verified report to the authority.
- Surrendering allowances against the phase-in share.
- Recording a pooling arrangement in the FuelEU database.
Automation boundary
Agent acts unaided
- ✓Bind every bunker delivery note to the leg that consumed the fuel.
- ✓Split every leg into full-count and half-count scope.
- ✓Pursue any unprovenanced consumption figure early.
- ✓Mark the gases that entered the surrender obligation this year.
Nothing leaves this agent as a filing; a named person submits, inside agreed boundaries.
- Judging whether a voyage leg was scoped correctly.
- Telling a flag State that the fleet is compliant.
- Setting the intensity limit a vessel is held to.
- Changes to voyage, bunker or emissions records.
Example output
One voyage leg, annotated
Our track and trace agent shows silence as silence rather than a last-known point; this record is what one voyage leg contributed.
Report state · single leg
Illustrative example- Consumption
- Bunker delivery note, one leg
- Recorded as
- Delivered quantity matched to the leg
- Scope
- Half-count leg
- Source of record
- Bunker delivery note, 3 August 2026
- Confidence
- Held unsubmitted
- Held for
- The designated company person, by name
As received
Taken from the bunker delivery note and the noon report — it reaches as far as those sources do.
What the record holds
- Bunker note date
- Flow meter reading
- Voyage leg, scoped
Why no verification call here
Whether a report is free from material misstatement is the verifier's call.
What the score decides
Below the configured threshold the report picks up a compliance read before submission.
Action
SubmitAppend recordSend to complianceValue
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 consumption record
From the source that raised it
Where the record is used
Our emissions data certification agent answers a domestic regulator where a designated representative certifies under penalty of law; no criminal certification exists on this side.
One dataset, four instruments
The MRV report is owed by 31 March 2026, and an administering authority may pull that date forward, though never before 28 February 2026.
What was checked, and not found
The IMO Net-Zero Framework was approved in April 2025 but never adopted: the extraordinary session of October 2025 adjourned without adopting it, spring 2026 reached no compromise, and as of 23 August 2026 it is not in force, with the resumed session set for 4 December 2026.
Build an evidence trail
The consumption, the voyage leg it sits on and the verifier who assessed stay on the report.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Bunker and consumption sources
- Bunker delivery notes
- Flow meters
- noon reports
Voyage and port data
- Port call records
- Voyage leg records
Verification and reporting
- Verifier query log
- Verified report filings
Agent
Vessel emissions compliance
- Reads the consumption
- Splits by leg and scope
- Holds for the company person
Compliance instruments
- EU ETS
- FuelEU database
- IMO DCS submissions
Observability & evaluation
- OpenTelemetry
- Langfuse
- Supported monitoring/evaluation sources
Integration availability depends on the client's existing systems and API access.
Agent controls
Six filters between the model and the company person
Six filters set in series, each finer than the last. Whatever survives is set out in the map below.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a report may stand
- L3
- Puts the submission in a person's hands
- L4 – L5
- Keep the consumption and the leg behind it
- L6
- Holds the report unsubmitted when signals degrade
How Nestack evaluates it
Evaluate the whole assembly — not only the report that comes out.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the report a verifier reads
- E1Final-output evaluationDid the report carry what each instrument actually requires?
- E2Step-level evaluationDid the agent read the right vessel, the right leg and the current record?
- E3Tool evaluationDid it read and write the correct voyage record and the correct scope?
- E4Confidence calibrationDo low-confidence reports actually attract more compliance corrections?
- E5Slice evaluationHow does performance change across specific vessel groups?
- E6Business outcomeHow many reports needed a correction before the company person submitted?
Floor — the report the shipping company owns
Failure modes
Where each failure originates in the agent
Seven failure modes, each fixed at the stage where it first surfaces.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- JH-03
Stale consumption read
The figure read is not the one the verifier later saw.
Stage gathers
The vessels, the legs, the records and the dates
02 · Reasoning
2 modes
- JH-04
Report asserted, not shown
A report is called complete without its records.
- JH-06
Unadopted framework read as live
The IMO Net-Zero Framework is worked as law.
Stage proposes
The legs, their scope split and the provenance
03 · Tool / write
2 modes
- JH-02
Thin report passed forward
A report moves on without the compliance read.
- JH-05
Consumption bound to wrong leg
A record is filed against the wrong voyage.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- JH-01
Submitted, provenance unrecorded
The report shows a submission but not what backed it.
Stage returns
The report a verifier assesses and an authority reads
05 · Change / Version
1 mode
- JH-07
Silent scope regression
A configuration change moves the split, not the report.
Stage tracks
Model, prompt, scope rules and consumption fields
- Sev-1 · a report submitted on no provenance
- Sev-2 · wrong consumption reaches the report
- Sev-3 · source degrades, report holds unsent
Affected slices
Ships on half-count voyages absorb the corrections
A vessel-level consumption-provenance figure can read clean while ships on half-count voyages carry most of the corrections. Nestack reports the correction rate by vessel group, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Ships on half-count voyages | 8.4% | 3.6× | Review | |
| Bareboat-chartered ships | 6.0% | 2.6× | Review | |
| Ships entering a pool | 3.7% | 1.6× | Watch | |
| Ships on intra-EU liner runs | 2.1% | 0.9× | Normal |
Bar: correction-rate lift vs. intra-EU liner baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
What a missed pooling window costs
A cycle ends when the missed pooling window is a standing case. That suite is what the next report assembled is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Correction rate rises on ships running half-count voyages.
- 02
Diagnose
The voyage that counted twice at one end and once at the other is read back until one cause remains.
- 03
Improve
Changes leave numbered, and the voyages behind them travel attached.
- 04
Verify
A single failing voyage case is enough to stop the release.
- 05
Learn
The case is kept, and the scope rules are rewritten alongside it.
Learn → Detect
The return edge. The next report is measured 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, report assembly, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Consumption-record and automation-boundary scoping.Week 1
- 02Bunker, voyage and verifier sources.Week 1, Week 2
- 03Record-to-instrument and scope-split coverage mapping.Week 1, Week 2
- 04Consumption record ingestion.Week 2
- 05Vessel, leg and source binding.Week 2, Week 3
- 06Provenance scoring and review routing.Week 3
- 07Company-person submission workflow.Week 3, Week 4
- 08Vessel and bunker-system integration.Week 5
- 09Scope-split and pooling cases.Week 4, Week 5
- 10Guardrails and verifier controls.Week 4
- 11Voyage-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 vessel group, one year | ProductionProduction emissions workflow | AdvancedMultiple vessels / instruments |
|---|---|---|---|
| Introduced at Pilot | |||
| Report assembly to your instruments | ✓ | ✓ | ✓ |
| Designated company person release | ✓ | ✓ | ✓ |
| Consumption-provenance baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by vessel | — | ✓ | ✓ |
| Submission workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Bunker-record integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-flag fleets | — | — | ✓ |
| Cross-instrument report packs | — | — | ✓ |
| Large fleets | — | — | ✓ |
| Multi-instrument deadline controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on integrations, workflow complexity, fleet size, 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 vessels and the instruments they owe
→Fleet mapping and consumption capture
Week 102Representative bunker, voyage and verifier records
→Source binding, scope logic and the provenance baseline
Week 203Your monitoring plans as the verifier holds them
→Record mapping, vessel binding and the automation boundary
Week 104Access to relevant APIs, feeds or exports
→Bunker, voyage and verifier-source assessment, then integration setup
Week 205Reports you would not want verified
→Scope cases and the evaluation suite
Week 406What no emissions report may settle
→Provenance scoring, review routing, guardrails and release controls
Week 307A designated company person to submit
→Submission 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 bars are drawn to the weeks the work genuinely takes, which is why the fifth carries a pair.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Emissions workflow discovery, record mapping and the automation boundary
W2Source integration and the consumption-provenance baseline
W3Report assembly, scope logic and release controls
W4Evaluation suite, scope cases and failure-mode testing
W5Record integration, pilot reports and targeted corrections
W6One reporting year run under the designated company person, then Agent Care handover
- Reading the band
- Each bar covers only the weeks its own work is named for. The fifth week carries two because the work does.
- At the end of W6
- When the voyage record validates, Agent Care picks the agent up.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Transportation AI agent
Build a vessel emissions agent on the single dataset that feeds all four instruments.
Show us one voyage leg and the bunker note behind it. Not a binding global carbon price: the IMO Net-Zero Framework adjourned in October 2025, stayed unadopted through spring 2026, and resumes 4 December 2026. Cargo paperwork at the port is a different agent.
- Typical delivery
- 4–6 weeks
- Deployment
- Your fuel dataset
- Verifier first
- Four deadlines
- After launch
- Agent Care available
Nestack Agents · Vessel emissions
AGT-TR-21 · Agent Care available after launch