- Home
- AI agents
- Enterprise AI agent examples
- Supply chain and field service
Plan · Enterprise AI agent examples
AI agents in supply chain, logistics and field service: what companies report
A supply chain or field service agent turns messages into actions in the system of record: it reads quote requests, orders and shipping documents, quotes and books within rate rules, sets appointments, and prepares technicians with the likely fault, the parts and answers from the manuals. Pricing exceptions, customs judgement and safety decisions stay with people.
In this guide: 83 pages
Key takeaways
- C.H. Robinson has published detailed figures. Its agents produced more than 80,000 ocean and air quotes for more than 3,600 customers, with quote time down from up to three days to two or three minutes and customs time down 23%, as the company reported on 26 October 2026.
- In field service the measured gain so far comes from answers. Textron Aviation reports troubleshooting lookups down from up to 20 minutes to one or two minutes, with an agent that searches about 60,000 pages of maintenance documentation.
- Order intake is the acting pattern in manufacturing supply chains. Danfoss says agents automate up to 80% of transactional order decisions, with people handling pricing exceptions.
- Several logistics announcements are targets: DHL Supply Chain’s message volumes and FedEx’s aim to put agents in more than half of its operational workflows by 2028. Separate them from results before you build a business case.
What do AI agents do in supply chain, logistics and field service?
Logistics agents quote, book, set appointments and process customs paperwork. In manufacturing supply chains, agents turn order emails into ERP actions. Field service agents answer technicians from the manuals, brief them before a visit and diagnose line equipment. The table shows the level each runs at.
Supply chain and field service tasks by what the agent is allowed to do. "Answers" means it replies from content; "assists" means it drafts or diagnoses and a person acts; "acts" means it changes something itself.
| Task | Level | What it touches, and the control we add | Seen in |
|---|---|---|---|
| Ocean and air quotes, bookings, appointments and customs paperwork | Acts | Rates, the transport management system and customs files; quotes stay inside rate and margin rules, and classification questions go to a specialist | C.H. Robinson |
| Order and quote emails to ERP actions | Acts | ERP order entry; pricing exceptions go to people | Danfoss |
| Maintenance manual answers | Answers | Manuals with revision numbers; the technician decides | Textron Aviation, Talgo |
| Pre-visit brief: likely fault and spare parts | Assists | Service history, product model and the parts catalogue | Fisher & Paykel |
| Robot and line diagnostics | Assists | Diagnostics, service history and repair guidance | Toyota Motor North America |
| Service order analysis for a vehicle fleet | Acts | Service orders across the fleet | Renting Colombia (one-line source) |
The acting agents all sit on a message queue that used to be worked by hand: a quote inbox, an order inbox, a customs file. That is where the published time savings come from, because the agent removes the wait in the queue as well as the handling time.
What have companies published about supply chain and field service agents?
Five companies publish a measured figure across six entries: C.H. Robinson twice in its own releases, and four companies through a vendor. Read each as the publisher’s claim; C.H. Robinson also describes in detail what its agents do.
Publicly reported supply chain and field service agents, as published by each source and opened on 2 November 2026. None of these companies is a QueryNow client.
| Company | What the agent does | Published figure | Source and date | How to read it |
|---|---|---|---|---|
| C.H. Robinson (forwarding and customs) | Agents quote ocean and air shipments, handle booking requests and process customs documents; more than a dozen can work on one shipping task | More than 80,000 quotes for over 3,600 customers; quotes from a day (air) and two to three days (ocean) to two to three minutes; customs time down 23% | Company press release, 26 October 2026 | Company claim; rollouts to the rest of Asia and globally are planned for Q4 2026 and Q1 2027 |
| C.H. Robinson (truckload) | Agents quote, process emailed orders, post available trucks from carrier emails, set appointments and check loads in transit | More than 3 million shipping tasks; over 1 million quotes and 1 million orders; more than 5,200 customers getting loads accepted in under 90 seconds, against up to four hours in an email queue | Company press release, 16 April 2025 | Company claim; counts of tasks done |
| Danfoss | Agents read order emails in any format, including tables, PDFs and images, and trigger the matching ERP actions; people handle pricing exceptions | Up to 80% of transactional decisions automated; processing time halved; response "potentially" from 42 hours to near real time | Google Cloud customer story, undated, read 2 November 2026 | Vendor story on Go Autonomous; the 42-hour figure is hedged |
| Textron Aviation | TAMI answers technicians’ questions, in several languages, across about 60,000 pages for more than 50 aircraft models | Troubleshooting lookups from up to 20 minutes to one or two minutes | Microsoft customer story, 20 March 2025 | Company-reported through the vendor |
| Toyota Motor North America | GearPal helps technicians diagnose robots and line assets from diagnostics, service history and repair guidance | Diagnosis from 5 to 6 hours to 2 to 3 minutes; more than 50 agents in production across ToyotaGPT | LangChain blog, 24 August 2026 | Staff quotes in a vendor blog; savings per facility are a projection |
| Renting Colombia | An assistant identifies service and maintenance needs across a fleet of more than 35,000 vehicles | Service order analysis time down 92%, from up to four hours to seconds | Microsoft Cloud Blog roundup, 24 July 2025 | One roundup line; scope unclear |
Plans and figure-free stories sit beside these. DHL Supply Chain’s HappyRobot agents book dock appointments and call drivers, and the published volumes are targets. FedEx describes manager, worker and audit agents and a 2028 target. Fisher & Paykel’s pre-visit briefs on Agentforce list benefits without numbers.
What controls does a supply chain or field service agent need?
These agents write to systems that move goods and send people to sites, so the controls follow money and safety: price limits, a narrow write scope, people on customs and safety decisions, and a count of the rework the agent causes.
- Price limits. Quotes come from your rate engine and margin rules, and the agent cannot go below a floor. Anything outside the rules goes to a pricing desk with the reason.
- Write scope. The agent writes named objects in the transport management system, the ERP or the field service app: a quote, a booking, an appointment, an order line. Master data such as carrier rates and customer terms stays read-only.
- Customs and compliance. The agent sorts documents, extracts data and chases what is missing; classification questions and declarations stay with your customs specialists or broker.
- Safety. In field service the agent briefs and answers and the technician confirms every repair step. Each answer cites the manual page and revision, so a superseded procedure is visible.
- Measurement. Track quote time and quote-to-booking conversion, the share of orders a person touches, first-time fix and repeat visits, and the rework the agent caused: wrong bookings, wrong parts, missed documents.
Where should you start?
Start where messages pile up in a shared inbox and each one ends in the same system action, such as quote requests, order emails or appointment requests. In field service, start with answers from the manuals before any agent changes a work order.
- Sample a month of the inbox and classify each message by type and by the action it ended in. The two or three largest types are the candidates.
- Run the agent read-only first: it extracts the fields and proposes the action, and a coordinator accepts or corrects it.
- Switch on automatic action for the message types with high acceptance, inside the price and write limits.
- For field service, index the manuals with revision numbers and test the answers against questions your senior technicians write.
- Report quote time, touch rate and first-time fix every month against the baseline.
Planning and forecasting agents are a separate case. None of the published examples with figures plans demand or inventory on its own; the measured results are all in message handling, quoting, order entry and technician support.
How do Microsoft, AWS and Google implement it?
Where the published examples ran, and what each cloud offers for the next one, with statuses as the vendor pages showed on 2 November 2026.
Capability, Microsoft, AWS, Google
| Capability | Microsoft | AWS | |
|---|---|---|---|
| Where the examples ran | Textron Aviation on Azure OpenAI; Renting Colombia on Azure AI services | No supply chain or field service example with figures in our research set | Danfoss with Go Autonomous on Google Cloud; Domina and Talgo in Google’s use-case list |
| Build the agent | Copilot Studio (low code) or Microsoft Foundry Agent Service (pro code) | Amazon Bedrock AgentCore Runtime and Gateway (GA 13 October 2025) | Gemini Enterprise Agent Platform with ADK and Agent Runtime |
| Connect the ERP or TMS | Power Platform connectors, or an MCP server in front of the system | AgentCore Gateway turns APIs and Lambda functions into MCP tools | Gemini Enterprise connectors; Agent Registry holds MCP servers and tools |
| Limits on actions | Foundry tool call guardrails (preview); approvals in Copilot Studio agent flows | AgentCore Policy, Cedar rules on every tool call (GA 3 March 2026) | Agent Gateway as the policy point (GA 18 June 2026) |
C.H. Robinson built its agents in-house. Danfoss runs Go Autonomous’ product on Google Cloud, Toyota built on LangChain’s Deep Agents, LangGraph and LangSmith, and Fisher & Paykel uses Agentforce for Field Service.
Frequently asked questions
Can an AI agent quote freight on its own?
C.H. Robinson’s agents do, for ocean and air freight: more than 80,000 quotes, with quote time down to two or three minutes, as the company reports. The quotes come from rate data and routing rules, and a person handles whatever the rules do not cover.
How do AI agents help field service technicians?
Mostly by answering and briefing. Textron Aviation’s technicians query about 60,000 pages of manuals; Fisher & Paykel’s agent writes a pre-visit brief with the likely fault and spare parts. The technician still decides and does the repair, and the agent saves the search.
How do you tell a plan from a result in a logistics AI announcement?
A result names a measured change on a task over a period, such as customs time down 23%. A plan uses words like target, expected or will, such as DHL Supply Chain’s volume targets or FedEx’s 2028 aim. Build a business case on the first kind only.
Does a supply chain agent need a data platform project first?
A first message-handling agent does not. It needs read access to the inbox, write access to a few objects in the transport or ERP system and the rate rules. Forecasting and planning agents are different, because they depend on clean demand and inventory history.
Which systems does a logistics agent need access to?
Usually the shared inbox, the transport management system or ERP, the rate engine and the document store, each through a connector whose write rights cover only the objects the agent creates. Carrier master data and customer terms stay read-only.
More in Enterprise AI agent examples
Sources
Each source was read on 2 November 2026. The date shown is the date on the page, or the page’s ms.date for Microsoft Learn; an undated page shows the date it was read.
- C.H. Robinson debuts Lean AI for ocean shipping, air freight, and customs
- At C.H. Robinson, artificial intelligence has now performed over 3 million shipping tasks
- Danfoss and Go Autonomous: reimagining order management at scale with Google Cloud
- Textron Aviation enhances maintenance efficiency with Azure AI
- Toyota scales enterprise AI with Deep Agents and LangSmith
- AI-powered success with more than 1,000 stories of customer transformation and innovation
- 1,302 real-world gen AI use cases from the world’s leading organizations
- Fisher & Paykel helps technicians arrive repair-ready with Agentforce
- DHL boosts operational efficiency and customer communications with HappyRobot’s AI workers
- FedEx plans agent workforce in over 50% of workflows by 2028
- Policy for Amazon Bedrock AgentCore is now generally available
- Gemini Enterprise Agent Platform release notes
- What is Microsoft Foundry Agent Service?
- Guardrails and controls in Microsoft Foundry
- Multistage and AI approvals in Copilot Studio agent flows
- Use connectors in Copilot Studio
- Introduction to connectors and data stores
- Amazon Bedrock AgentCore Gateway
- Amazon Bedrock AgentCore is now generally available
Planning an AI program?
Talk to the team that builds it and runs it after launch.
- Built to SOC 2, HIPAA and GDPR standards
- EU AI Act-aligned delivery
- Microsoft Solutions Partner