Transportation AI agent · Quoting
Freight-Quoting Assistant AI Agent
Assemble a rate quote from named sources, bound it with a validity window and sanity limits, and hold it for the person authorised to commit the brokerage — a sent quote can bind.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Broker commits
- Pre-commitment
- After launch
- Agent Care
What this agent does
Assembles the quote, not the authority to commit it
In
- 01
Rate inputs, lane history and accessorial rules from supported TMS, rating or market-data sources.
- 02
Lane, equipment and accessorial fields normalised, each value carried forward with its source and vintage.
Reason
- 03
A priced quote for the lane, with the validity window and the expiry stamped on the artefact itself.
- 04
Sanity bounds you set applied to the number, so a rate outside them stops instead of going out.
- 05
Named sources per line — your own booked history, published indices — and no pooled non-public competitor rates.
Decide
- 06
Quote language checked against 49 CFR 371.7(b), so nothing reads as an offer to haul the freight.
- 07
Every quote routed to the person authorised to commit the brokerage, whatever the confidence.
Out
- 08
The inputs, their vintage, the model version, the quote and the person who committed it, kept on the file.
- 09
Write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent assembles the quote; a named person commits it, and your firm stays the registered broker of record.
Example workflow
One quote, request to commitment
- 1Quote request receivedTMS request, email RFQ, shipper portal or rating API
- 2Rate inputs gatheredLane, equipment, accessorials, your booked history and published indices, each with its source and vintage
- 3Quote assembledRate, expiry, sanity result and confidence
- 4Controls appliedSource and vintage checks, sanity bounds, binding-language checks and confidence threshold
No human action required
Stages 1 to 4 run unaided and commit nothing — an assembled quote is not a sent quote, and the committing lane opens at the confidence gate.
Goes to the authorised person to commit.
Adds a pricing-desk read first.
Broker commitment
The quote is held with its rate sources, their vintage and the confidence.
Approve · Amend · Send to pricing desk
Approved — released to send
- 6Quoting systems updatedOnly where write access and approval policy allow it
- 7Outcome evaluatedQuotes held, sanity stops, desk amendments and what the lane actually cost to cover
Amendments
Every desk amendment is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Sending a quote out to a shipper or a carrier.
- Committing the brokerage to a rate or a capacity.
- Issuing, signing or returning a rate confirmation.
- Waiving, amending or extending the quote terms.
Automation boundary
Agent acts unaided
- ✓Assemble a priced quote from the sources you configured.
- ✓Name each rate source on the quote, and how old it is.
- ✓Stamp the validity window and the expiry on the artefact for the named owner.
- ✓Stop a quote outside your sanity bounds and hold it for a person.
Any write happens inside the boundaries agreed at implementation, never ahead of approval.
- Accepting a counter-offer from a carrier or shipper.
- Invoking the manifest-error clause on a live quote.
- Deciding who at the brokerage may commit a rate.
- Changes to the sanity bounds or quoting thresholds.
Example output
One quote, annotated
Everything the agent assembles is attached to the sources it was priced from.
Quoting output · single lane
Illustrative example- Lane
- Regional dry van
- Quoted line
- Line-haul and fuel, accessorials itemised separately
- All-in rate
- $2,940.00
- Rate source
- Your own booked lane history
- Confidence
- 89%
- Quote terms
- Broker status shown, expiry stamped
As received
Taken from your booked history and the indices you configured — nothing on this side is invented by the agent.
Rate sources used
- Your booked lane history
- Published market index
- Your accessorial schedule
Why this can bind
It is an offer with an expiry — accepted inside the window, it binds the broker.
What the score decides
Below the configured threshold the quote picks up a pricing-desk read before.
Action
ApproveAmendSend to pricing 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 quote request
From the TMS or the inbox
Price from named sources
Draw on your own booked history and the published indices you configured for the lane and the mode.
Quote it, a person commits it
Routine lane quotes and accessorial lines arrive already assembled.
Send review to what can bind
Sanity stops, stale sources and low-confidence rates are marked, so the desk's read starts where risk concentrates.
Build an evidence trail
The quote, the rate sources behind it and the person who released it stay on the file.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
TMS and brokerage
- McLeod
- MercuryGate
- Revenova
- Turvo
Rating and market data
- DAT
- Truckstop
- Greenscreens
- SMC³
CRM and RFQ intake
- Salesforce
- HubSpot
- shipper portals
Agent
Freight-quoting assistance
- Reads the sources
- Assembles the quote
- Holds for commitment
Documents and confirmations
- Rate confirmations
- BOL
- E-signature systems
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 quote
The layers nest. What each one misses is named in the map below.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a quote may stand
- L3
- Puts the commitment in a person's hands
- L4 – L5
- Show what the quote was built from
- L6
- Reverts to rate lookup when signals degrade
How Nestack evaluates it
Evaluate the whole quoting workflow — not only the number at the end.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the quote the customer receives
- E1Final-output evaluationDid the quote hold against what the lane actually cost to cover?
- E2Step-level evaluationDid the agent use the right sources, the right vintage and your lane rules?
- E3Tool evaluationDid it read the correct lane and write the correct quote record?
- E4Confidence calibrationDo low-confidence quotes actually attract more desk amendments?
- E5Slice evaluationHow does performance change across specific lane and mode groups?
- E6Business outcomeHow many quotes were amended, and how many were honoured at a loss?
Floor — the margin the brokerage actually keeps
Failure modes
Where each failure originates in the agent
Seven ways a quote goes wrong, set at the stage it starts.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- AH-03
Stale rate source
The lane is priced from an index older than the file records.
Stage gathers
Lane, equipment, accessorials, history and indices
02 · Reasoning
2 modes
- AH-04
Plausible-but-wrong rate
A rate looks right, sits under market, and binds anyway.
- AH-06
Carrier-like wording
Quote language reads as an offer to haul the freight.
Stage proposes
Rate, validity window and confidence
03 · Tool / write
2 modes
- AH-02
Quote sent without a person
A write releases the quote before anyone commits it.
- AH-05
Expired quote re-sent
A lapsed quote goes out again with a live window.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- AH-01
Rate without its sources
A number is returned with no source or vintage behind it.
Stage returns
The quote the customer reads and may accept
05 · Change / Version
1 mode
- AH-07
Silent bound regression
A model or rule change widens what the sanity bounds let through.
Stage tracks
Model, prompt, sanity bounds and source config
- Sev-1 · a quote sent outside the boundary
- Sev-2 · a wrong rate reaches the customer
- Sev-3 · a source degrades, quote routes to review
Affected slices
Overall quote quality can hide one bad lane
An aggregate amendment rate can look acceptable while a small number of lane groups absorb most of the rework and most of the margin given away. Nestack reports the desk-amendment rate by slice, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Spot quotes on thin lanes | 5.1% | 3.4× | Review | |
| Accessorial-heavy quotes | 4.2% | 2.8× | Review | |
| Multi-stop and partial loads | 2.8% | 1.9× | Watch | |
| Contract lanes with history | 1.0% | 0.7× | Normal |
Bar: desk-amendment-rate lift vs. contract-lane baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
No cycle closes on an explanation
A cycle is done when the miss has become a test the next release has to survive. That suite is what the next quote out the door is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Desk-amendment rate rises in a lane group.
- 02
Diagnose
Not the number but the sources are read first, until one cause holds.
- 03
Improve
Version-stamp the change and attach the quotes that exposed it.
- 04
Verify
The affected cases run again, and a fail stops the release.
- 05
Learn
It becomes a standing test, and the quoting rules change with it.
Learn → Detect
The return edge. Every cycle hands the next one a longer suite to clear.
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, quoting workflow, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Quoting workflow discovery and automation-boundary definition.Week 1
- 02Rate-source and rating-engine assessment.Week 1, Week 2
- 03Sanity bounds, expiry and quote-terms rule mapping.Week 1, Week 2
- 04Rate-input ingestion and normalisation.Week 2
- 05Quote assembly and source binding.Week 2, Week 3
- 06Confidence scoring and sanity routing.Week 3
- 07Broker commitment workflow.Week 3, Week 4
- 08TMS and rating-engine integration.Week 5
- 09Binding-language cases.Week 4, Week 5
- 10Guardrails and approval controls.Week 4
- 11Quote-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 lane set, one desk | ProductionProduction quoting systems | AdvancedMultiple entities / modes |
|---|---|---|---|
| Introduced at Pilot | |||
| Quoting to your sources and bounds | ✓ | ✓ | ✓ |
| Broker commitment | ✓ | ✓ | ✓ |
| Quote-accuracy baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by lane and mode | — | ✓ | ✓ |
| Commitment workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Rating-engine integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-mode pricing rules | — | — | ✓ |
| Multi-stage pricing approvals | — | — | ✓ |
| High quote volume | — | — | ✓ |
| Multi-entity quoting 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 lanes, modes and accessorial rules
→Rate-input ingestion and source mapping
Week 102Representative historical quotes
→Quoting baseline, source binding and the expiry stamp
Week 203Your rate sources and your sanity bounds
→Sanity bounds, expiry and quote-terms rule mapping
Week 104Access to relevant APIs, feeds or exports
→Rate-source and rating-engine assessment, then integration setup
Week 205Quotes you would not want honoured
→Formation cases and the evaluation suite
Week 406What a quote may never commit on its own
→Confidence scoring, sanity routing, guardrails and approval controls
Week 307Named people authorised to commit a rate
→Broker commitment 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
Phases are drawn over the weeks they actually occupy, which is why week 5 doubles up rather than padding.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Quoting workflow discovery, bound mapping and the automation boundary
W2Rate-source and rating integration and the quoting baseline
W3Quoting workflow, confidence logic and commitment controls
W4Evaluation suite, binding-language cases and failure-mode testing
W5Rating-engine integration, pilot lanes and targeted corrections
W6One quoting cycle assembled under the desk, 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 last checks clear on live quotes and monitoring moves to Agent Care.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Transportation AI agent
Build a quoting agent that assembles, and leaves the commitment with a person.
Show us your rate sources, your lanes and your quote terms. If you can name who may commit a rate and the bounds they work inside, we'll set the automation boundary around them.
- Typical delivery
- 4–6 weeks
- Deployment
- Your quoting stack
- Broker commits
- Pre-commitment
- After launch
- Agent Care available
Nestack Agents · Freight quoting
AGT-TR-02 · Agent Care available after launch