How Lineage delivers high-touch service across 2 million shipments
Customer
Lineage Logistics
Headquarters
Novi, MI
Operation Type
Cold chain network
Services
Warehousing, Cold Chain, Trucking, Freight Forwarding
Summary
Lineage partnered with Pallet to automate shipment processing across three business units: inbound receiving, freight forwarding, and warehouse order management. Pallet's agents now handle these workflows roughly two million times a year at operator-level accuracy, enabling touchless processing and high-touch service across the massive network. With cost no longer climbing in lockstep with volume, Lineage is expanding Pallet across its US and international sites in 2026.
About the company
Lineage is the world's largest temperature-controlled warehouse and logistics operator, moving tens of billions of pounds of food annually across 480+ facilities for many of the world's largest food and beverage producers, retailers, and distributors.
The company has committed significant spending to LinOS, its warehouse execution platform. However, front-end order processing was a workflow that hadn't seen enough rigor. It still ran on unstructured inputs, tribal knowledge, and manual effort across three distinct business units.
Lineage partnered with Pallet to make that processing touchless, without sacrificing the high-touch service its customers expect.
At our scale, Pallet had to exceed our existing accuracy levels. The challenge was that the knowledge required to do so lived in our operators’ heads. Pallet captured that expertise, reached production in just 49 days, and helped us elevate our service.
Sudarsan Thattai
CIO and Chief Transformation Officer
Elliott Wolf, Chief Data Scientist, founded and leads Lineage’s data science team, applying advanced mathematical techniques to optimize the cold chain.
Chloe Inestroza, VP of Transportation Data Science, focuses on predictive analytics, machine learning, and AI to optimize Lineage's cold chain logistics network.
Thomas Kuijpers, System & Implementation Manager, leads freight forwarding and customs process automation and digital transformation.
Sudarsan Thattai, CIO and Chief Transformation Officer, oversees information technology and transportation across the company’s 480+ facility global network.
The problem
Across three business units, Lineage faced the same core issue: skilled operators spending hours on manual processes that were slow, error-prone, and nearly impossible to scale.
- Inbound warehouse receiving generates millions of shipment packets, including BOLs, inventory lists, POs, and arrival notes, that take up to 10 minutes each to key into structured data by hand.
- Freight forwarding offices field thousands of shipping-request emails a day. Each order takes an operator 15 minutes to extract 30+ structured fields and enter into CargoWise by hand.
- Warehouse order management is the hardest of the three. Many customer has different requirements across multiple WMSs, and coordinators spend can 10 minutes per order matching emails, attachments, and product codes to WMS records.
Multiply those minutes across millions of documents, and the cost adds up fast. Adding people could never keep pace with the volume, so the work itself had to change.
Solution
In a matter of weeks, Pallet deployed three AI agents, each purpose-built for a distinct workflow that spans Lineage's internal systems.
Process
|
Before
|
After
|
|---|---|---|
Inbound Warehouse Receiving | Millions of shipment packets (BOLs, POs, arrival notices) each converted into structured data for WMS ingestion by hand. | A Receiving Agent extracts structured data from shipment packets at scale, at 99.8% accuracy. |
Freight Forwarding Order Automation | Thousands of new shipment requests keyed into CargoWise by hand, including implicit details that exist only as tribal knowledge. | An Order Intake Agent, trained on Lineage's processes and tribal knowledge, enters orders into CargoWise at 99.8% accuracy. It reached full production in just 49 days. |
Warehouse Order Automation | Customer emails and attachments parsed manually into WMS pick-and-pack records, despite inconsistent product codes and ambiguous customer IDs. | An Order Management Agent automates the entire workflow, ambiguous orders included, at accuracy that matches or beats experienced operators. |
1,000 shipment packets a day, converted to WMS-ready data at 99.8% accuracy
What makes this hard is the variety and scale. Packets arrive from different carriers and countries as handwritten notes, scanned photos, and multi-page tables, and traditional OCR could never read across that range.
The Receiving Agent reads BOLs, POs, arrival notices, and packing lists the way an operator would, extracting around 40 fields per packet across shipper, carrier, and cargo details. By capturing and embedding tribal knowledge into the Pallet platform, it holds that accuracy at scale: across 53 facilities, it processes up to 1,000 documents a day at 99.8% accuracy, even on peak days.
We’ve tried a lot of different tools, and they all struggle with the complexity of our documents. Across thousands of handwritten notes, rough photos, and complex packets, Pallet handles them with ease and a high degree of accuracy.
Elliott Wolf
Vice President, Chief Data Scientist
Order entry from 15 to 2 minutes, live in 49 days
The hard part is that much of what an order requires was never written down. Forwarding teams serve hundreds of customers, each with their own conventions, and the details live only in coordinators’ heads.
Pallet captured and operationalized that judgment, from customer-specific formats to recognizing forwards and follow-ups to pulling track-and-trace data from ocean and air cargo systems. The Order Intake Agent monitors forwarding inboxes, extracts around 30 fields per order, checks for duplicates, and creates the order directly in CargoWise. It went from first deployment to full production in just 49 days.
Judgment calls at higher accuracy than experienced operators
This is the hardest of the three because item resolution takes judgment. Customers describe a product one way, the WMS stores it another, and a single customer name can map to multiple entity IDs across subsidiaries.
The Order Management Agent runs the full pipeline from customer email to validated, WMS-ready order. It pulls orders from Salesforce, opens attachments in any format, and resolves every line item, including item codes, lot numbers, and delivery instructions, against the facility's WMS catalog. Experienced operators always knew which record to use. Pallet captured that tribal knowledge, connected data across systems, and ran thousands of simulations before go-live. Now the agent matches a seasoned operator: near 100% on unambiguous orders and 95% on ambiguous resolution, in a fraction of the time.
Outcomes
Across three Lineage business units, Pallet now runs shipment processing that used to be done by hand, roughly two million times a year, at the accuracy the work demands.
The bigger shift is structural. Costs and effort no longer climb in lockstep with shipment volume, so Lineage can grow the network and improve its operating efficiency at the same time. Coordinators spend their time on customers and exceptions instead of data entry, and service levels hold as shipments scale.
In 2026, Pallet is expected to process over 3 million shipments for Lineage, with expansion continuing across US and international sites. Lineage customers get better service, and global operations can scale as fast as the network grows.
