New playbook: when your own AI model pays. Read

C.H. Robinson: 2–3 minutes per ocean or air freight quote, down from one to three days

C.H. Robinson (freight logistics, ~11,250 employees (Q3 2026 average)) reported this on 28 Oct 2026 in earnings call (source). It belongs to the agents, traceability and evals layer.

What they built

Hundreds of in-house AI agents embedded across the shipment lifecycle (quoting, orders, appointments, missed-pickup checks) and now ocean and air quoting, ocean booking and customs brokerage, built and run under a lean waste-removal discipline by 450+ in-house engineers and data scientists.

The result

NAST adjusted operating margin 40.3% and Global Forwarding adjusted operating margin 36.6% in Q3 2026; annual token cost under $2M on a run-rate basis. Ocean and air price quotes that took a day (air) or two to three days (ocean) now take two to three minutes; missed-pickup agents automate 95% of checks and save 350+ hours of manual work a day.

The number covers segment margins are Q3 2026 earnings figures; the token cost is C.H. Robinson's annual run rate; quote time covers ocean and air price quoting (Oct 2026 release); missed-pickup figures are one NAST workflow (company blog, Sep 2026).

The layer: agents, traceability and evals

Each agent has one narrow task, a named owner and an action log.

  • Agents with one narrow job and an identity
  • A trace of every tool call and decision
  • Eval sets that gate every change

Read the guide to agents, traceability and evals.

Playbooks that use this case

More cases on agents, traceability and evals

  • 12,500

    banker hours saved in the first month of operation, July 2026

    Five specialist agents on Amazon Bedrock AgentCore that classify and read payslips and bank statements, run serviceability calculations and verify them against lending policy, then hand the banker a result to review for mortgage and credit-card applications (source).

  • $10M

    EBITDA uplift from two agent systems, Partners Group says

    Two agentic systems built on a cleaned proprietary policy-level dataset: one runs new-client policy processing, the other automates human-intensive policy checking; delivered by a seven-person team with Version 1 (source).

  • 80%

    less manual submission triage time

    An agentic intake platform that reads broker submissions, digitises them and triages risks into a decision-ready workflow for underwriters across more than 20 countries (source).

We map how your company runs, pick the first processes with your leadership team, and put them on agents.