The machine that runs
freight operations
A fully autonomous multi-agent system that moves a freight shipment from supplier routing request to carrier award to DC notification to invoice reconciliation — with a human director making exactly two decisions along the way.
Built on tools the client already operates
The intelligence layer is supplied by West Atlas. The infrastructure is Google Workspace — email, storage, and data the client already owns. No new software, no proprietary databases, no vendor lock-in. Every component is auditable and replaceable.
One dedicated Google Workspace inbox (tms-agent@) and a service account with scoped API access to Gmail, Sheets, and Drive. West Atlas handles everything else — build, deployment, and ongoing operation.
Freight coordination is a full-time job hiding inside someone else's job
For a CPG brand shipping weekly to retail distribution centers, the freight quoting process is a relentless coordination tax on the logistics team. Every shipment is a new email chain — chasing suppliers for routing info, soliciting quotes one-by-one, selecting a carrier, notifying the DC, filing the documents, and months later trying to reconcile an invoice against a quote no one can find. None of it is value-adding. All of it is necessary.
The real cost isn't just time. It's what doesn't get done — strategic vendor relationships, cost analysis, DC performance tracking — because the team is occupied with coordination that should be invisible.
Seven agents. The director makes two decisions.
Every agent has one job. Google Sheets is the shared state layer — each agent reads the current shipment status and writes to it when its job is done. The director approves once to solicit quotes, and once to award a carrier. Everything else runs without human intervention.
Two decisions. Complete visibility.
The director touches the system twice per shipment — once to approve quoting, once to select a carrier. Every other action is handled by the agents. What the director sees is designed to make those two decisions fast, informed, and auditable.
Gate 2 — quote digest with one-click award
| Forwarder | Quoted Cost | Transit | Est. Delivery | vs. L90 avg | |
|---|---|---|---|---|---|
| Forwarder A | $X,XXX | 4 days | Apr 23 | +2.5% | Award A |
| Forwarder B | $X,XXX | 5 days | Apr 24 | -9.9% | Award B |
| Forwarder C | $X,XXX | 3 days | Apr 22 | +14.2% | Award C |
| Window | Avg Cost | Avg / Pallet | Shipments |
|---|---|---|---|
| L30 | $X,XXX | $XXX | 3 |
| L90 | $X,XXX | $XXX | 9 |
| L180 | $X,XXX | $XXX | 22 |
Award confirmation — full audit record in browser
Invoice variance alert — agent reads the PDF so the director doesn't have to
| Line Item | Quoted | Invoiced | Variance | Note |
|---|---|---|---|---|
| Base freight | $X,XXX | $X,XXX | — | Matches quote |
| Fuel surcharge | $XXX | $XXX | +$10 | Rate change |
| Liftgate fee | $0 | $65 | +$65 | Not in quote |
| Address correction | $0 | $75 | +$75 | Not in quote — disputable |
A complete operations database. Built automatically.
Every action the agent takes is written to a single Google Sheets master record — one row per shipment, columns spanning intake through award through invoice reconciliation. The reporting capability isn't designed in. It's the natural result of a system that logs everything it does. The client never enters data manually. The database builds itself.
| Shipment ID | PO # | Supplier | Pallets | Weight | Origin | DC Code | Mode | Ready Date | Drive Folder | Forwarder | Award Cost | Award Date | Cost/Pallet | L90 Avg | vs L90 | Invoice # | Base Freight | Fuel Surch. | Liftgate | Addr. Corr. | Detention | Other | Total Invoice | Variance $ | Variance % | Inv. Flag | Invoice PDF | Status | DC Notice |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TMS-0038 | PO-XXXX01 | Supplier Co A | 3 | 1,100 lbs | Origin City | DC-01 | LTL | Apr 8 | 🔗 Open | Forwarder B | $X,XXX | Apr 8 | $XXX | $X,XXX | -2.4% | INV-8821 | $XXX | $XXX | $0 | $0 | $0 | $20 | $X,XXX | +$15 | +1.3% | GREEN | COMPLETE | ✓ Sent | |
| TMS-0039 | PO-XXXX02 | Supplier Co B | 5 | 2,100 lbs | Origin City | DC-02 | LTL | Apr 11 | 🔗 Open | Forwarder A | $X,XXX | Apr 11 | $XXX | $X,XXX | +3.1% | INV-8834 | $X,XXX | $XXX | $65 | $75 | $0 | $0 | $X,XXX | +$150 | +12.1% | RED | COMPLETE | ✓ Sent | |
| TMS-0040 | PO-XXXX03 | Supplier Co C | 2 | 840 lbs | Origin City | DC-01 | LTL | Apr 14 | 🔗 Open | Forwarder C | $X,XXX | Apr 14 | $XXX | $X,XXX | +8.6% | INV-8847 | $X,XXX | $XXX | $0 | $0 | $0 | $15 | $X,XXX | +$15 | +1.1% | GREEN | COMPLETE | ✓ Sent | |
| TMS-0041 | PO-XXXX04 | Supplier Co A | 3 | 1,200 lbs | Origin City | DC-01 | LTL | Apr 18 | 🔗 Open | Forwarder B | $X,XXX | Apr 18 | $XXX | $X,XXX | -9.9% | — | — | — | — | — | — | — | — | — | — | — | — | COMPLETE | ✓ Sent |
| TMS-0042 | PO-XXXX05 | Supplier Co D | 6 | 2,800 lbs | Origin City | DC-03 | LTL | Apr 20 | 🔗 Open | Forwarder A | $X,XXX | Apr 19 | $XXX | $X,XXX | +1.8% | — | — | — | — | — | — | — | — | — | — | — | — | AWARDED | ✓ Sent |
| TMS-0043 | PO-XXXX06 | Supplier Co B | 2 | 840 lbs | Origin City | DC-02 | LTL | Apr 22 | 🔗 Open | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | QUOTING | — |
| TMS-0044 | PO-XXXX07 | Supplier Co C | 4 | 1,600 lbs | Origin City | DC-01 | LTL | Apr 25 | 🔗 Open | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | PENDING_APPROVAL | — |
← Intake columns | Award columns | Invoice columns | Status — color banding marks each lifecycle section · All rows written by agents · Zero manual entry
The machine gets smarter over time
The v1 pipeline automates the complete freight lifecycle. As the Shipment Master accumulates history, the benchmark engine gets more accurate, the invoice variance patterns become actionable, and the enhancement roadmap unlocks carrier-level intelligence and multi-client deployment.