Change log: US Freight & Logistics 2026

Companies with $50 million to $10 billion in revenue

Every correction that changes the leader list or a profile in the Roiva AI in Practice Index, US Freight & Logistics 2026, dated and with the reason.

This log starts when the edition is published. From then on it records, each dated and with the reason:

  • every correction that changes a published result, meaning the leader list or a profile, naming the company;
  • corrections that change only a score we don't publish, counted without naming the company;
  • leaders who asked not to be listed, counted without naming them.

A Roiva staff member approves the wording of each entry, never whether it appears: every such change is posted. Changes made while the edition was being drafted, before publication, are revisions of a draft rather than corrections, and aren't listed here. The method doesn't change for a published edition; changes for later editions are dated on the method page.

What has changed since publication?

The other companies read a second time (November 17, 2026)

A second blind sample was read again, and then a second reader, working without the first reading, read again every use and governance practice at the companies that aren't leaders. Writers' words, figures credited to AI in general, plans and quotes that don't name AI were corrected or dropped. The leader list doesn't change. No profile changed. Scores we don't publish changed for four companies.

Leaders' uses checked against a sharper test (November 17, 2026)

A reader checked every use behind the leaders against a sharper test of what counts as a deployment. That test applies from the next edition. For this edition we corrected only what its own rules already rule out: plans and hopes, AI as an industry trend, and AI in a product a company sells. The leader list doesn't change. What changed on a profile:

  • Schneider National: AI to triage driver calls, which it is "looking at", no longer counts. The use stands on other sources, so only its count of passages quoted changed.

A blind sample read again (November 17, 2026)

A separate reader read again a random sample of 199 items, about a tenth of every item the four editions count, without seeing the earlier readings, and we corrected what it found. The leader list doesn't change. No freight result changed.

Leader items read a second time (November 17, 2026)

A second reader, working without the first reading, read again each item behind the leaders' results that could be opened. Plans and hopes were dropped; programs and training recorded as uses of AI became general statements; and quotes, statuses and dates were brought to the sources' own words. The leader list doesn't change. What changed on a profile:

  • ArcBest: an invoice-automation use is now "status not stated": its source marks it as including AI but doesn't say it is live.
  • Landstar System: its Microsoft Copilot rollout is now "status not stated", for the same reason.
  • Schneider National: a web page couldn't be reliably dated in the study's window (its own data gives 2022), and its use is off the profile.

One passage, several signals (November 17, 2026)

A passage can give more than one governance signal when each rests on its own words in it. Before, we had counted one signal per paragraph. We re-checked the signals left out under that shorthand. What changed:

  • ArcBest: its Responsible AI policy is back on the profile: its review model balances automation "with the appropriate level of human oversight".

The leader list doesn't change.

General statements and governance practices read on their sources (November 17, 2026)

We read on their sources the governance practices and general statements about AI that this edition counts and that hadn't yet been read under the rules it was published with. A general statement now counts only when it says what the company is doing with AI now: plans, hopes and AI as an industry trend don't count, and neither do risk sentences that name no use, trade-press writers' words or quotes that don't name AI.

What changed on a profile:

  • Landstar System: Motive's AI dashcams, offered for its owner-operators to buy, aren't Landstar's own use of AI, and its AI task force's plan to work with startups is a plan; both are off the profile.

The leader list doesn't change. Scores we don't publish changed for three other companies.

Leaders' figures re-checked (November 17, 2026)

We re-checked the figures behind the leaders' results, under the rules this edition was published with. A figure a company credits to AI in general, such as a company-wide ratio or a cost program with AI as one of several levers, is a general statement about AI, not a figure for an AI initiative. What changed:

  • ArcBest: "AI-driven automations" that cut emails names no specific use of AI, so that use and its figure are off the profile. The carrier phone-agent quote now stops at the sentence that names AI.
  • RXO: its costs taken out since the spin-off, through initiatives "including AI investment", and a productivity figure credited to AI in general no longer count, and the two uses they described are off the profile. Its agentic AI for carrier calls is now listed as live rather than at scale, as its sources describe it.
  • Schneider National: a productivity figure credited to "very effective AI", and two statements of AI investment beside cost and margin targets, name no specific use of AI. The figure and the two uses are off the profile.

All three are still leaders. No company joined or left the list.

Published (November 16, 2026)

US Freight & Logistics 2026 is published, with results as of November 14, 2026, under the method set out on the method page.

Leaders who asked not to be listed: 0.