Gather the entry data set from your shipping documents, carry each field with the source it came from, and hand the prepared package to a licensed customs broker, who decides and files.
Integration availability depends on the client's existing systems and API access.
Agent controls
Six layers between the model and the entry
Controls stack inward. What the stack does not catch is set out in the map below.
L6 · Outermost — last line of defenceInward → L1 · closest to the model
L6Rollback / safe modeNarrow the agent to data assembly when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt, rule and classification-scope changes.Track
L4TraceabilityRecord the source documents, the assembled fields, the flags and the broker's decision.Record
L3Broker reviewHold the package for a named licensed broker; the hold governs release, not whether the filing is right.Gate
L2Scope guardrailsTest outputs against the six-digit cap and the held-act list; a failure returns the package. A notice on screen is not tested; the cap is.Restrict
L1Confidence thresholdsRoute low-confidence packages to a compliance read before the broker sees them.Require review
Model corePackage prepared — entry data set, open gaps, flagged fields and confidence
L1 – L2Test whether a package may stand
L3Puts the filing in a broker's hands
L4 – L5Keep the entry data and the basis behind it
L6Narrows to data assembly when signals degrade
How Nestack evaluates it
Evaluate the preparation workflow — not only the fields handed over.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the data the broker files from
Depth of coverage ▼
E1Final-output evaluationDid every assembled field match the document it was read from?
E2Step-level evaluationDid the agent use the right documents, product library and prior decisions?
E3Tool evaluationDid it read and write the correct entry and the correct field?
E4Confidence calibrationDo low-confidence packages actually attract more broker corrections?
E5Slice evaluationHow does performance change across specific shipment types?
E6Business outcomeHow many packages needed a broker correction or a post-entry amendment?
Floor — the entry the importer answers for
Failure modes
Where each failure originates in the agent
Seven ways a package goes wrong, placed by stage.
Agent lifecycleDirection of processing →
01 · Retrieval1 mode
AJ-03
Superseded document read
A field is taken from an invoice that was later replaced.
Stage gathersInvoice, packing list, transport document and product library
02 · Reasoning2 modes
AJ-04
Candidate past the cap
A candidate is extended past six digits for goods being imported.
AJ-06
Assertion carried as fact
A supplier's stated origin is passed on as though it were determined.
Stage proposesEntry data set, six-digit candidates and confidence
03 · Tool / write2 modes
AJ-02
Package released early
A package moves on with held fields still unresolved.
AJ-05
Duplicate entry data set
The same shipment is assembled and handed over twice.
Stage writesOnly where write access and approval policy allow it
04 · Output1 mode
AJ-01
Missing statutory element
A required invoice element is absent and not flagged.
Stage returnsThe entry data set the broker files from
05 · Change / Version1 mode
AJ-07
Silent scope regression
A model or rule change widens what the agent will assert.
Stage tracksModel, prompt, scope rules and product-library config
Sev-1 · acts outside the boundarySev-2 · a wrong field reaches the brokerSev-3 · source degrades, package routes to review
Break the correction rate down by cohort before you trust it — a headline that reads acceptably is usually a few shipment types absorbing most of the rework. Nestack reports it by slice, not only in total.
Slice performance — reported separately, not only in aggregateIllustrative example
Slice
Failure rate
Lift
Lift vs. threshold
Status
Section 232 derivative lists
9.2%
3.9×
Review
Mixed-material assemblies
6.1%
2.6×
Review
New supplier, first shipment
4.2%
1.8×
Watch
Repeat parts, product library
1.4%
0.6×
Normal
Bar: broker-correction-rate lift vs. repeat-part baseline · scale 0–4.0× · tick marks the 2.0× review threshold2 of 4 slices over threshold
Evidence-linked improvement
A cycle ends in a case, not a note
An apology does not close a cycle; a case the next release must pass does. That suite is what the next entry prepared for filing is measured against.
Improvement cycle · five stagesSwitchback — the path turns at Improve and returns at Learn
01Detect
Broker-correction rate rises in a shipment slice.
02Diagnose
The prepared packages and the documents behind them are read side by side until the cause narrows to one.
03Improve
The correction is versioned, with the entries that motivated it attached.
04Verify
A failing case holds the release until it passes.
05Learn
The case is added permanently, and the classification notes move with it.
Learn → DetectThe return edge. The suite the next package meets is 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, documents, preparation workflow, evaluation, integration, then production validation and handover.
WorkstreamWeek 1Week 2Week 3Week 4Week 5Week 6
01Entry workflow discovery and boundary definition with your broker.
02Document and system assessment.
03Product library, prior decisions and held-act mapping.
04Document ingestion and normalisation.
05Assembly logic and source binding.
06Confidence scoring and gap routing.
07Licensed broker review workflow.
08Customs-platform and record integration.
09Classification and origin cases.
10Guardrails and filing controls.
11Entry-trail instrumentation.
12Deployment, documentation and Agent Care handover.
12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallelFinal 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 tierPilotOne importer, one brokerProductionProduction customs systemsAdvancedMultiple entities / countries
Introduced at Pilot
Preparation to your documents and rules✓✓✓
Licensed broker review✓✓✓
Preparation-accuracy baseline✓✓✓
Introduced at Production
Reporting by tariff chapter—✓✓
Broker review workflow in your systems—✓✓
Approved write-back—✓✓
Customs-platform integration—✓✓
Introduced at Advanced
Multi-programme duty rules——✓
Multi-stage compliance approvals——✓
High entry volume——✓
Multi-country entry controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build priceConfirmed after discovery based on integrations, workflow complexity, entry volume, approval controls and deployment requirements.
Separate from buildBuild 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 bringWe build with it
01Your document set and the fields your broker needs→Document ingestion and field mappingWeek 1
02Representative filed entries→Preparation baseline, product-library extraction and source bindingWeek 2
03Your product library and prior broker decisions→Product library, prior decisions and held-act mappingWeek 1
04Access to relevant APIs, feeds or exports→Document and system assessment, then integration setupWeek 2
05Entries you would not want filed→Classification cases and the evaluation suiteWeek 4
06What only a licensed broker may do→Confidence scoring, gap routing, guardrails and filing controlsWeek 3
07Named licensed brokers to review packages→Broker review workflow, then pilot and production validationWeeks 5–6
Nothing else is requiredDeployment, documentation and Agent Care handover are ours.
Delivery timeline
Four phases across six weeks
Each band covers the weeks the work really takes, so week 5 carries evaluation and launch together.
PhaseW1W2W3W4W5W6
DiscoveryW1
BuildW2 – W3
EvaluateW4 – W5
Pilot & LaunchW5 – W6
Week focusW1Entry workflow discovery, held-act mapping and the automation boundaryW2Document integration and the preparation baselineW3Preparation workflow, confidence logic and broker review controlsW4Evaluation suite, scope guardrails and failure-mode testingW5Customs-platform integration, pilot entries and targeted correctionsW6One entry cycle prepared under the licensed broker, then handover
Reading the bandA bar spans only the weeks its work is named in. Week 5 carries two because that is how the work runs.
At the end of W6Validation finishes on live entries and Agent Care assumes monitoring.
DurationSix-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Transportation AI agent
Build a customs-documentation agent around your broker's filing chain.
Show us your documents, your product library and your filing chain. A licensed customs broker signs the entry, and nothing here moves that. We'll map the preparation workflow and set the boundary.