Logistics and transportation
How freight brokers, carriers and shippers use Microsoft technology: email and order agents on Microsoft Foundry, transportation management in Dynamics 365 and connected fleet data, with NIS2 duties for transport operators.
The problem
Much of freight still runs on email. C.H. Robinson receives tens of thousands of customer emails a day, and many routine requests, such as price quotes, could wait hours before an employee reached them.
Requests come in many forms: truckload, less-than-truckload, intermodal and air freight each need different details, and planners must also choose carriers, routes and rates for inbound and outbound loads.
In the EU, transport is one of the sectors covered by NIS2, so operators must manage cyber risk and report significant incidents.12
The approach
Microsoft groups transport under Microsoft for Mobility, whose connected fleets scenario combines Azure messaging services, Azure Maps, Microsoft Fabric, Dynamics 365 and Power Platform. For planning and execution, transportation management in Dynamics 365 Supply Chain Management helps identify routes and rates for inbound and outbound loads, for example the fastest route or the least expensive rate.
C.H. Robinson built its own generative AI tools on Azure AI Foundry (now Microsoft Foundry), Azure OpenAI in Foundry Models, Azure SQL Database and Azure Cosmos DB. They classify incoming emails, assemble the details, identify missing information and carry out the steps a person would take, with a human feedback loop. A key account manager says quote time fell from hours to an average of 32 seconds. In April 2025 the company said its agents had performed over 3 million shipping tasks, from quotes and orders to capacity, appointments and tracking; that release does not mention Microsoft.
A smaller forwarder with modest email volume may get further with shared mailboxes, templates and its TMS's own automation. Agents pay off when request volume is high, request types repeat and the TMS has APIs the agent can call.1345
Conceptual architecture
Diagram as a list
Copilot & business apps
- Customer emails and portal requestsWhere quote, order and appointment requests arriveConnects to Classification and extraction (Azure OpenAI in Foundry Models)
- Human feedback loop for exceptionsStaff correct and handle unusual requests
Agents & automation
- Task agents for quotes, orders and appointments (Microsoft Foundry)Complete routine steps a planner would take
AI models & platform
- Classification and extraction (Azure OpenAI in Foundry Models)Identifies request type and missing detailsConnects to Task agents for quotes, orders and appointments (Microsoft Foundry)
Data & analytics
- Operational data (Azure SQL Database, Azure Cosmos DB)Rates, loads and conversation stateConnects to Task agents for quotes, orders and appointments (Microsoft Foundry)
- Transportation management (Dynamics 365 Supply Chain Management or an existing TMS)Routes, rates, loads and carrier records
- Connected fleet telemetry (Azure messaging services, Azure Maps, Microsoft Fabric)Location and status data for trackingConnects to Transportation management (Dynamics 365 Supply Chain Management or an existing TMS)
Technologies and their roles
Microsoft Foundry1
Agent platform
C.H. Robinson built its generative AI tools on Azure AI Foundry, now Microsoft Foundry.
Microsoft Foundry Models1
Language models
Azure OpenAI in Foundry Models classifies and processes C.H. Robinson's customer emails.
Azure SQL1
Relational data
Azure SQL Database is part of C.H. Robinson's stack.
Azure Cosmos DB1
Operational data
Azure Cosmos DB is part of C.H. Robinson's stack.
Microsoft Dynamics 3654
Transportation management
Transportation management in Dynamics 365 Supply Chain Management plans routes and rates for inbound and outbound loads.
What you need first1
- Rate, capacity and order data reachable through APIs
- A catalog of request types with the fields each one needs
- Exception routing to named staff when the agent is unsure
- Quality monitoring of agent output, sampled every week
Risks and how to reduce them
- NIS2 duties for transport operators2
- Transport is a NIS2 sector; include agent platforms and TMS integrations in cyber risk management and significant-incident reporting.
- Wrong quotes or bookings sent automatically1
- Keep a human feedback loop and route unclear requests to people, as C.H. Robinson does.
- Attributing customer-wide results to one vendor5
- C.H. Robinson's 3 million task figure comes from a release that does not mention Microsoft; do not read it as a product benchmark.
Documented examples
C.H. Robinson
In productionC.H. Robinson: AI agents for freight quotes, orders and tracking
from hours to an average of 32 seconds per quote (measured)
Logistics and transportation, United States (global operations)
Related
Sources
- Customer story: C.H. Robinson (Azure AI Foundry) (opens in a new tab)
- NIS2 Directive: securing network and information systems (opens in a new tab)
- How does Microsoft support mobility? (opens in a new tab)
- Transportation management overview - Dynamics 365 Supply Chain Management (opens in a new tab)
- At C.H. Robinson, Artificial Intelligence Has Now Performed Over 3 Million Shipping Tasks (opens in a new tab)
Thank you. Your correction was sent.
The editors check it against the sources. If you left an email address, they may reply about it.