1. ELD / Hours of Service
  2. Work with your agent
Reference

Put an AI agent to work in ELD / Hours of Service

Ask for help preparing loads, checking review queues and tracing operation evidence.

Set it up

  1. Teach your agent ERP.AI

    Load the ERP.AI skill and paste the output into your agent. It explains how to read and update records in your workspace.

    curl -fsSL https://www.erp.ai/skill.md
  2. Give it a role

    Assign the agent its own user and one of the 5 roles. It gets exactly the access that role has, and no more.

  3. Check what it can reach

    Confirm which records, actions and connections the agent can use in your workspace before you point it at real data.

  4. Review its work and usage

    Check the drafts and changes it proposes, and watch its usage receipts as you give it more to do. Agent work uses prepaid credits.

What the agent can do

  • Dispatcher: Find my assigned loads, compare their promised times with pickup observations and prepare a summary of duty intervals waiting for review.
  • Transport Supervisor: Gather the submitted interval, source evidence and matching load revision so I can make the independent decision.
  • App Administrator: Identify roster entries whose eligibility dates need attention and list vehicle references that the team should check before the pilot.
  • Auditor: Trace the original operation receipt, reviewer reason and compensating evidence for a selected load within my access scope.

What stays with a person

  • The agent needs an authorized person to make an independent transport approval.
  • It cannot establish a physical pickup or repair from an unexplained label.
  • The responsible team supplies observed evidence, confirms the source identity and decides how a disputed external result should be handled.
  • Keep requests tied to a recognizable load and operating scope.

The agent needs an authorized person to make an independent transport approval. It cannot establish a physical pickup or repair from an unexplained label. The responsible team supplies observed evidence, confirms the source identity and decides how a disputed external result should be handled.

Keep requests tied to a recognizable load and operating scope. Ask the agent to show the existing receipt before requesting another action when a response is uncertain.

Agent activity and connected operations use the platform's metered services. Estimate expected usage from the pilot's import and review activity; see [Pricing](/eld-hours-of-service/pricing).

Set up ELD / Hours of Service your way.

Start free, then use Proto to change fields, steps and approvals before your team goes live.

How to get started