Transportation AI agent · Registration & fuel tax
Carrier Registration & Fuel Tax AI Agent
Derive each biennial update month from the USDOT number that carries it, hold every jurisdiction distance and fuel line against the trip behind it, and leave the return for the person who signs.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Signed by name
- Number-derived
- After launch
- Agent Care
What this agent does
Assembles the distance, never the assessment
In
- 01
A quarter closes, and IFTA R960.100 sets return and payment at the last day of the following month.
- 02
A registration month arrives, and § 390.19T(b) reads it off the last digit of the USDOT number.
Reason
- 03
An odd or even year is owed, and § 390.19T(b) takes that from the next-to-last digit of the same number.
- 04
A trip ends, and P720.250 wants total distance in all jurisdictions, trip-permit operations included.
- 05
A fuel receipt lands, and P560 wants a monthly summary of distance and fuel put into each vehicle.
Decide
- 06
A UCR year opens, and 49 CFR 367.50 brackets the fee; registration must complete before 1 January.
- 07
A registration year is set by the base jurisdiction, and no uniform national IRP renewal date was found.
Out
- 08
A return is drafted, and P720.750 wants the signature of the person filing, unless filing is electronic.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent assembles the distance and runs each clock. A named person signs the return and lodges the biennial update; the base jurisdiction assesses.
Example workflow
One quarter, distance to signature
- 1Distance and fuel records receivedTelematics traces, trip sheets, fuel-card transactions and retail receipts
- 2Return context assembledThe fleet, the jurisdictions crossed, the fuel each unit burns and the period it falls in
- 3Return draft preparedThe jurisdictions, their distance and volume, the gaps and completeness
- 4Controls appliedProvenance checks, clock checks, jurisdiction-coverage checks and completeness confidence
No human action required
Stages 1 to 4 run unaided, and nothing is lodged at any of them — the agent is assembling, and the fuel-tax lane opens at the completeness gate.
Goes to the named person to sign and lodge.
Adds a fuel-tax desk read first.
Fuel-tax review
The draft is held with its gaps, its jurisdictions and the period it covers.
Sign · Append record · Send to the fuel-tax desk
Signed — by the named person, lodged
- 6Fleet and registration records updatedOnly where write access and records policy allow it
- 7Outcome evaluatedDistance provenance, jurisdiction coverage, desk corrections and what review found
Corrections
Each fuel-tax desk correction is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Signing the quarterly return under P720.750.
- Lodging the biennial update under § 390.19T.
- Assessing the interest a jurisdiction is owed.
- Registering a fleet with the base jurisdiction.
Automation boundary
Agent acts unaided
- ✓Bind every jurisdiction line to the trip and the receipt behind it.
- ✓Run the biennial clock that the USDOT number sets.
- ✓Chase a missing fuel receipt before it is due.
- ✓Mark the distance that reaches a draft with no trip underneath it.
Nothing is signed or lodged except by a named person, inside the agreed boundaries.
- Deciding what a jurisdiction may collect.
- Telling an auditor the distance record is sound.
- Setting the fleet a registration is held under.
- Changes to fleet, registration or mileage records.
Example output
One jurisdiction line, annotated
Our dispatch and routing agent plans against recorded hours and our maintenance planner books the shop; this record is what one jurisdiction line consumed.
Return draft · single jurisdiction
Illustrative example- Jurisdiction line
- Taxable distance, P720.250
- Recorded as
- Distance traced to trip and vehicle
- Fuel type
- Diesel
- Source of record
- Telematics trace, 3 August 2026
- Confidence
- Held unsigned
- Held for
- The person who signs, by name
As received
Taken from the telematics trace and the fuel-card record — it reaches as far as those sources do.
What the record holds
- Trip and vehicle
- Fuel-card receipt
- Jurisdiction, dated
Why no tax call here
What a jurisdiction is owed is its own assessment, not a model output.
What the score decides
Below the configured threshold the draft picks up a desk read before the signer sees it.
Action
SignAppend recordSend to the fuel-tax deskValue
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 distance line
From the trip that produced it
Where the distance is used
Our hours-of-service records agent corroborates the duty day; one trip also leaves a distance record, kept on a different clock by a different regime.
Your number sets the date
§ 390.19T(b) takes the month from the last digit of the USDOT number and the year from the digit before it.
What was checked, and not found
At 88 FR 80184 on 17 November 2023 § 390.19 was suspended indefinitely, leaving § 390.19T operative for US-domiciled carriers; and the duty to lodge an update when nothing has changed sits in agency guidance, not in the codified text of § 390.19T(b).
Build an evidence trail
The distance, the trip it came from and the person who signed the return stay on the record.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Distance and telematics
- GPS traces
- trip sheets
- Odometer and border crossings
Fuel purchases
- Fuel-card transactions
- Retail receipts and bulk draws
Fleet and vehicle records
- Asset register
- unit list
- Weights, fuel types and plates
Agent
Carrier registration and fuel tax
- Reads the distance
- Runs each clock
- Holds for the signer
Registration systems
- Base jurisdiction portal
- USDOT and UCR records
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 signer
Six layers, tightening as they go inward. Whatever reaches the centre is described in the map below.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a draft may stand
- L3
- Puts the signature in a person's hands
- L4 – L5
- Keep the distance and the trip behind it
- L6
- Holds the return undrafted when signals degrade
How Nestack evaluates it
Evaluate the whole assembly — not only the return that comes out.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the return an auditor reads
- E1Final-output evaluationDid the draft carry what each jurisdiction column actually asks for?
- E2Step-level evaluationDid the agent read the right fleet, the right period and the live trace?
- E3Tool evaluationDid it read and write the correct vehicle record and the correct period?
- E4Confidence calibrationDo low-confidence drafts actually attract more desk corrections?
- E5Slice evaluationHow does performance change across specific jurisdictions?
- E6Business outcomeHow many drafts needed a correction before the signer put a name to one?
Floor — the record the carrier answers for
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
- JI-03
Stale distance read
The trace read is not the one the trip actually ran.
Stage gathers
The fleets, the trips, the jurisdictions and the fuel
02 · Reasoning
2 modes
- JI-04
Distance asserted, not traced
A jurisdiction line stands with no trip behind it.
- JI-06
Proposed fee read as law
The UCR increase proposed for 2027 is worked as settled.
Stage proposes
The jurisdictions, their distance and coverage
03 · Tool / write
2 modes
- JI-02
Thin draft passed forward
A draft moves on without the fuel-tax read.
- JI-05
Distance bound to wrong unit
A trace is filed against the wrong vehicle.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- JI-01
Signed, provenance unrecorded
The record shows a signature but not what stood behind it.
Stage returns
The return a person signs and an auditor reads
05 · Change / Version
1 mode
- JI-07
Silent retention regression
A configuration change moves the clock, not the record.
Stage tracks
Model, prompt, return rules and distance fields
- Sev-1 · a return signed on untraced distance
- Sev-2 · wrong distance reaches the return
- Sev-3 · source degrades, return holds unsigned
Affected slices
Low distance jurisdictions absorb the corrections
A jurisdiction-level distance-provenance figure can read clean while low distance jurisdictions carry most of the corrections. Nestack reports the correction rate by jurisdiction, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Low distance jurisdictions | 7.6% | 3.6× | Review | |
| Trip-permit operations | 5.4% | 2.6× | Review | |
| Bulk-fuel jurisdictions | 3.4% | 1.6× | Watch | |
| High distance jurisdictions | 1.6% | 0.8× | Normal |
Bar: correction-rate lift vs. high distance jurisdiction baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
What a short retention clock costs
A cycle closes when the short retention window is a regression case. That suite is what the next return built is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Correction rate rises on low distance jurisdictions.
- 02
Diagnose
The month nobody could find on a calendar, because it was sitting in the USDOT number itself, is read back until one cause remains.
- 03
Improve
Changes go out numbered, and the jurisdictions that drove them ride along.
- 04
Verify
One jurisdiction case still failing keeps the release where it is.
- 05
Learn
It joins the suite for good, and the retention rules move in the same commit.
Learn → Detect
The return edge. The next return 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, return assembly, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Distance-clock discovery and boundary-setting work.Week 1
- 02Distance, fuel and registration sources.Week 1, Week 2
- 03Jurisdiction-column and retention-clock mapping.Week 1, Week 2
- 04Distance and fuel ingestion.Week 2
- 05Trip, vehicle and jurisdiction binding.Week 2, Week 3
- 06Completeness scoring and review routing.Week 3
- 07Named-signer filing workflow.Week 3, Week 4
- 08Fleet and registration-system integration.Week 5
- 09Distance-source and retention cases.Week 4, Week 5
- 10Guardrails and filing controls.Week 4
- 11Distance-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 jurisdiction, one quarter | ProductionProduction fuel-tax workflow | AdvancedMultiple fleets / base jurisdictions |
|---|---|---|---|
| Introduced at Pilot | |||
| Return assembly to your jurisdictions | ✓ | ✓ | ✓ |
| Named-signer release | ✓ | ✓ | ✓ |
| Distance-provenance baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by jurisdiction | — | ✓ | ✓ |
| Filing workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Telematics and fuel-card integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-fleet registrations | — | — | ✓ |
| Cross-regime record packs | — | — | ✓ |
| Large mixed fleets | — | — | ✓ |
| Multi-regime retention 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 fleet and the jurisdictions it runs
→Fleet mapping and distance capture
Week 102Representative traces, trip sheets and fuel records
→Source binding, clock logic and the distance baseline
Week 203Your USDOT number and UCR bracket
→Jurisdiction mapping, trip binding and the automation boundary
Week 104Access to relevant APIs, feeds or exports
→Telematics, fuel-card and registration-source assessment, then integration setup
Week 205Returns you would not want audited
→Distance cases and failure testing
Week 406What no distance record may support
→Completeness scoring, review routing, guardrails and release controls
Week 307A named person to sign the return
→Named-signer 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
Each band is the honest length of a phase in working weeks, and that is why one week carries two.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Registration discovery, jurisdiction mapping and the automation boundary
W2Source integration and the distance-provenance baseline
W3Return assembly, clock logic and release controls
W4Evaluation suite, distance cases and failure-mode testing
W5Record integration, pilot returns and targeted corrections
W6One registration year run under the fleet administrator, 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
- Once the distance record validates, Agent Care picks up the agent.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Transportation AI agent
Build a registration and fuel tax agent around the return your named signer has to put a name to.
Show us one quarter and the distance behind it. Which month does your biennial update fall in — the last digit of the USDOT number already answers that, under § 390.19T(b). A transactional tax on sales is a different agent.
- Typical delivery
- 4–6 weeks
- Deployment
- Your distance records
- Signed by name
- Number-derived
- After launch
- Agent Care available
Nestack Agents · Registration & fuel tax
AGT-TR-22 · Agent Care available after launch