All case studies

Case studies: shipping & logistics ops

Your couriers have APIs. Your ERP doesn't. We connect both.

We automate the shipping desk: orders come out of your ERP (even without an API), shipments get registered with each courier, labels print, and pickups get booked, with no re-typing. Then we add invoice checks against your contracted rates.
Book a 30-min workflow reviewGet a free 90-day AI plan

Order to pickup, without re-typing

  1. ERP order

    No API: browser automation

  2. Courier registration

    Courier API

  3. Label printed

  4. Pickup booked

  5. Tracking back to ERP

  • 10 wks

    Concept to live invoice-checking tool

    BCG case

  • Up to 30%

    Of procurement manual work streamlined

    BCG

  • −25%

    Fulfillment cost in AI-run warehouses

    JD.com

  • 6-12 wks

    Typical first deployment

    ATI Lab delivery

Case studies

From routing to invoice checks: what AI already does in logistics

The large operators show the ceiling. The invoice and process-redesign cases show what a mid-size team can copy.

Third-party results from other companies, linked to source. They are not ATI Lab clients. How we would build it for you is below. See ATI Lab's own client builds.

Cases a mid-size team can copy

Leading German Energy Provider - Procurement AI

Energy / Procurement - GermanyThird-party result

The bottleneck

Supplier invoices came in above contract rates, and at that volume nobody could check every line.

What fixed it

Custom GenAI tool built in 10 weeks that scans incoming invoices, compares charged amounts against contract pricing, flags discrepancies, and auto-drafts reconciliation messages for claims managers.

Published result

  • Tens of $M

    in potential value from identified overpayments

  • 10 weeks

    from concept to live tool

  • Auto-drafted

    supplier reconciliation messages

What an invoice check looks like

LineChargedContractStatus
Freight, lane 1€1,240€1,240Matches
Fuel surcharge (flagged)Reconciliation message drafted for the claims manager€310€248Above contract
Handling€95€95Matches
Storage€420€420Matches
IllustrativeExample data, not from the published case.

Source: Boston Consulting Group (opens source in a new tab) - "How Four Companies Capitalize on AI to Deliver Cost Transformations" (Sep 2025)

Global Biopharma Company - GenAI Process Redesign

Pharma / Supply chain - GlobalThird-party result

The bottleneck

Wanted to go beyond simply speeding up existing processes. Needed to rethink workflows with GenAI capabilities in mind.

What fixed it

Instead of bolting AI onto old processes, teams evaluated each workflow, determined where GenAI improved it, and ran "mirror" processes - old and new in parallel - to validate before switching over.

Published result

  • Mirror testing

    old and new process run in parallel before switch-over

  • Redesigned

    processes around AI capabilities, not vice versa

Source: Boston Consulting Group (opens source in a new tab) - "How Four Companies Capitalize on AI to Deliver Cost Transformations" (Sep 2025)

Enterprise references

General Mills - AI Supply Chain Optimization

Enterprise reference - CPG, USAThird-party result

The bottleneck

Over 5,000 daily shipments from manufacturing plants to warehouses. Suboptimal routing and load planning driving up transportation costs.

What fixed it

AI models deployed to assess and optimize transportation across full shipment volume - route selection, load consolidation, and warehouse allocation.

Published result

  • $20M+

    saved since FY2024

  • 5,000+

    daily shipments optimized by AI

  • Service levels up

    alongside cost cuts

Sources: FactR / DataPeak (opens source in a new tab), Master of Code (opens source in a new tab) - industry comparison and supporting analysis

UPS - ORION AI Route Optimization

Enterprise reference - logistics, USA / GlobalThird-party result

The bottleneck

Fixed delivery routes could not adapt to real-time conditions. Millions of packages, thousands of drivers, static route planning wasting fuel, time, and money.

What fixed it

AI agent called ORION (On-Road Integrated Optimization and Navigation) plans delivery routes in real time using live data - traffic, package priority, driver constraints - instead of relying on fixed paths.

Published result

  • $300M/yr

    annual savings

  • Real-time

    route optimization using live conditions

  • Per driver

    path calculated for each delivery run

Source: Botpress (opens source in a new tab) - "10 Real World Case Studies of AI Agents" (May 2025)

JD.com (Chinese E-Commerce) - AI Warehouse & Fulfillment

Enterprise reference - e-commerce / logistics, ChinaThird-party result

The bottleneck

Massive e-commerce scale requiring constant optimization of procurement, warehousing, and order fulfillment to maintain margins.

What fixed it

AI-driven supply chain optimization across procurement efficiency, warehouse robotics and digital twins for logistics planning.

Published result

  • −25%(decrease)

    order fulfillment costs in advanced warehouses

  • +1.2%(increase)

    revenue rise from AI-optimized procurement

  • +92%(increase)

    profit increase in Q2 2024 (company-wide; AI one of several drivers)

  • $10B

    projected annual savings by 2030

$10B is a projection. The profit figure is company-wide and not attributable to AI alone.

Order fulfillment cost in advanced warehouses, indexed (before = 100)

  • Fulfillment cost25% lower
    BeforeBefore: 100
    AfterAfter: 75
Third-party published result.

Source: Master of Code (opens source in a new tab) - "How Does AI Reduce Costs? Save 5-20% Across Operations" (Aug 2025), citing analyst reports

Intake and quoting

How are logistics companies using AI to automate shipment intake and quoting?

Logistics companies use AI agents to read shipment requests from email and attached documents, create the order or booking, and draft a quote in seconds for a person to approve. The largest published case is C.H. Robinson, which receives more than 11,000 pricing emails a day. Its agents create emailed truckload orders in about 90 seconds and have produced over 1 million quotes at 32 seconds each. The platforms below each cover part of this. A mid-size shipper or forwarder usually needs the intake step built around its own inbox and ERP.

What the main freight platforms offer, October 2026

PlatformShipment intakeQuotingPublished figure
C.H. RobinsonOrders agent reads emailed tenders and attachments and creates the orderQuoting agent returns customer-specific quotes5,500 truckload orders a day, about 90 seconds each; 1M+ quotes at 32 seconds each (June 2025)
FlexportEmail booking details with a document; an agent returns a quote for review or books the shipmentAgent converts Flexport rates into the client's own rate-sheet template"80% of emails eliminated" per shipment, platform-wide
FreightosForwarders paste emails, packing lists or PDFs; AI extracts commodities, ports and dimensionsAI drafts air, ocean and trucking quotes, each approved by a personQuoting cut from 7 steps to 3
DescartesAI reads commercial invoices, bills of lading and packing lists for customs and shipment creationSpot quoting tool (not described as AI)"90% of customer quotes in under 30 seconds" (Sept 2026)
project44No published shipper-intake agentProcurement agent benchmarks contract rates (shipper side)About 1M automated carrier messages (April 2026)
Uber FreightDocument agent for carrier paperwork; no published shipper-order intakeInstant-quote API (not described as AI)30+ AI agents in its network (May 2025)

Sources: each company's own site or newsroom, checked 18 October 2026. The figures are self-reported and come from carriers and brokers at a scale most shippers never reach. C.H. Robinson (June 2025) · C.H. Robinson, 11,000 pricing emails a day (May 2024) · Flexport AI · Freightos AI for forwarding · Descartes (Sept 2026) · project44 (April 2026) · Uber Freight (May 2025)

What we built: shipment intake from email

On one client's shipping desk, shipment requests arrived as free-text emails with attachments, and staff retyped the details into several systems. The system we built reads each inbound email, checks the details against the client's booking rules, then creates the labels and carrier bookings. The result was 95% less manual work and zero monthly entry errors. The operations team still works from email, so nobody had to learn a new tool.

We have not published a quoting build. Where rates already sit in a rate sheet or the ERP, the same intake step can fill in the quote request and leave the price approval to a person, as Freightos does.

Read the full case

How we'd build it for you

What we build for shipping & ops teams

  • The pain: Re-typing orders into each courier portal

    What we build: ERP → courier registration → label → pickup booking workflow (API + browser automation)

    Published proof: Typical first deployment: 6-12 weeks

  • The pain: Supplier and carrier invoices overcharge unnoticed

    What we build: Invoice-vs-contract checker that drafts the dispute message

    Published proof: Tens of $M flagged, live in 10 weeks (BCG case)

  • The pain: Fear of breaking shipping during the change

    What we build: Mirror run: new flow in parallel until outputs match

    Published proof: Switch-over without disruption (biopharma case)

Couriers have APIs, your ERP doesn't

The ERP has no API, so the ATI Lab automation layer reads it through browser automation and connects to each courier through its API.. Your ERP (No API: read via browser automation) feed into ATI Lab automation layer (Registration, labels, pickups, tracking), which updates Courier A (API), Courier B (API), Courier C (API or portal).

How we switch over without breaking shipping

  1. Map the order-to-pickup flow

  2. Build the automated flow

  3. Run old and new in parallel

    Mirror run

  4. Switch over once outputs match

Industry benchmarks

Where AI pays back in supply chain

Cost savings from supply-chain AI

of companies using AI in supply chain report cost savings

Source: AI Index Report 2025. Third-party published result.

Procurement cost reduction possible with AI

  • Procurement costMcKinsey / BCG15–45%
Third-party published result.

Procurement manual work streamlined

  • Manual work streamlinedUp to 30%
Source: BCG. Third-party published result.

  • 43%

    of companies using AI in supply chain report cost savings

    AI Index Report 2025

  • 15-45%

    overall procurement cost reduction possible with AI

    McKinsey / BCG

  • >5%

    revenue upside from GenAI in supply chain (highest function)

    McKinsey Global Survey

  • Up to 30%

    manual work in procurement streamlined by AI

    BCG

Next step

Send us one week of shipments. We'll show you what can run itself.

We map your order-to-pickup flow (ERP export, courier registration, label, booking, tracking update), automate it on your existing systems, and run it in parallel with your team before switching over. Typical first deployment: 6-12 weeks.

Map my shipping workflowLogistics & freight tasks we automate