By Luis Lopez, AI transportation consultant, CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast
Hours-of-service compliance is one of those areas where small problems pile up quietly. A few unassigned driving events here, a pattern of edits there, a driver who keeps finishing right at the limit. None of it looks serious on a single day. Over months, it becomes the kind of record that draws attention in an audit or a roadside inspection.
ELDs produce a huge amount of data, and most fleets only look at it when something goes wrong. AI-assisted hours of service review changes that by scanning logs continuously and surfacing patterns that deserve a human look. This article covers the rules that matter, what to look for in ELD data, and how to use the findings without damaging trust with drivers.
The Rules in Brief
For property-carrying drivers, the federal hours-of-service rules in 49 CFR Part 395 include:
- An 11-hour driving limit.
- A 14-hour on-duty window within which that driving must happen.
- A 30-minute break required after 8 cumulative hours of driving.
- 60/70-hour limits over 7 or 8 consecutive days.
ELDs are required for most drivers who must keep records of duty status. There are exceptions and additional provisions, so fleets should confirm how the rules apply to their operation. For a shipper-oriented overview, see our article on hours of service explained for shippers.
What AI Looks for in ELD Data
An AI review tool works best as a pattern finder. It does not decide whether a violation happened. It points reviewers to the records most likely to need attention.
Unassigned Driving
When a truck moves without a driver logged in, the ELD records unassigned driving time. Some of it is legitimate, such as a mechanic moving a truck in the shop. Unresolved unassigned time is a common audit finding, though, and it can hide driving that should be on someone’s log. AI can match unassigned segments to likely drivers based on dispatch records, fuel transactions or the driver who logged in just before and after, then queue them for a person to resolve.
Log Edits
Edits are allowed with proper annotation, and many are routine corrections. Patterns are what matter: frequent edits that shift driving time to on-duty or off-duty status, edits clustered near limits, or edits with vague annotations. AI can flag edit patterns by driver, by time of day and by how close the original log was to a limit.
Personal Conveyance and Yard Move Use
Personal conveyance and yard moves are special driving categories with specific conditions on their use. They are easy to misuse, sometimes without the driver realizing it. AI can flag personal conveyance segments that are unusually long, that start right after a long on-duty period, or that end at a customer or terminal location. It can flag yard move segments that cover distances or speeds that do not fit a yard. Each flag is a question for review, not a conclusion.
Near-Limit Patterns
A driver who routinely ends shifts within minutes of the 11-hour or 14-hour limit is not necessarily breaking rules. But that pattern often signals loads that are planned too tight, and it raises the chance that a small delay turns into a violation. AI can track how often each driver and each lane runs close to limits.
Planning Loads That Fit
The best place to prevent hours-of-service problems is before the truck moves. When dispatch plans a load, the plan should account for the driver’s available hours, the 30-minute break, realistic transit time and likely detention at pickup and delivery.
AI tools can check proposed assignments against a driver’s remaining hours and flag loads that only work if nothing goes wrong. If one lane or one customer keeps producing near-limit days, that is a planning problem, often tied to long dwell at a facility. Our article on questions to ask about AI and ELD telematics data covers how that data can feed planning.
Coaching vs Discipline
How a fleet uses findings matters as much as the findings themselves. If every flag leads to punishment, drivers learn to distrust the system, and some will look for ways around it. If flags lead to conversations, the fleet learns why patterns happen.
As an AI transportation consultant, the first thing I look at in a log review program is what happens after a flag. A healthy process usually looks like this:
- The AI flags a pattern.
- A trained reviewer checks the records and context.
- If there is a real issue, a supervisor talks with the driver to understand it.
- The fleet fixes root causes, such as dispatch planning, customer dwell or unclear policy.
- Discipline is reserved for deliberate or repeated misconduct after coaching.
Our piece on AI driver safety scoring and coaching covers similar principles for safety data.
Privacy and Transparency
ELD and telematics data reveal where drivers are and what they are doing throughout the day. Fleets should be clear with drivers about what data is reviewed, why, and who can see it. Limiting access to people with a real need, and keeping review focused on compliance and safety rather than unrelated monitoring, helps maintain trust. Fleets should also make sure their practices fit any applicable privacy laws and their own written policies.
Keep Humans in the Review
AI flags are a starting point. A person should always review the underlying records before any conclusion is drawn, because context is often missing from the data. A long personal conveyance segment may have a legitimate explanation. An edit may correct a genuine device error. A near-limit day may trace back to a road closure.
Limits to Keep in Mind
- Data quality. Device malfunctions, connectivity gaps and incorrect driver logins all affect what the AI sees.
- Rule complexity. Exceptions and special provisions mean a tool can flag something that is actually permitted. Reviewers need to know the rules.
- False confidence. A clean AI report does not guarantee compliance. It only means the tool found nothing it was built to find.
Used well, AI-assisted review helps fleets find and fix log risk before an auditor or inspector does, and it points to the planning problems behind many violations.
For more conversations on freight and AI, subscribe to the Freight Guru Podcast.
About the author: Luis Lopez is a Miami-based AI transportation consultant and logistics entrepreneur, the CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast.
