Skip to content
be-human.ai
← Back to the journal

AI by industry

AI for Excavation Companies: Locate Requests and Dig Tickets

How an excavation company can use an AI agent to track locate requests, dig-ticket expirations and crew notes, with human approval on every safety-critical step.

Be Human typographic cover with the title: AI for Excavation Companies: Locate Requests and Dig Tickets
Conceptual illustration created with AI · Be Human

An AI agent for an excavation company can keep locate requests, dig-ticket expiration dates and crew field notes in one shared record, and alert the office before a ticket lapses or a locate response is missing. It should never confirm that a site is clear, release a dig or change a safety instruction, because those decisions belong to qualified people. The practical starting point is one narrow role with written approval boundaries and a success check the office can observe.

Key takeaways

  • Locate requests, dig tickets and crew notes are linked records. Keeping them in one place lowers the chance that a crew digs on a stale or incomplete ticket.
  • An agent is well suited to watching deadlines, matching tickets to job sites and turning field notes into searchable records. It should not decide whether ground is cleared.
  • Every safety-critical step needs a named human owner, a recorded approval and a permission the agent cannot exceed.
  • Start with one role, such as the ticket and locate tracker for the office, before adding anything crew-facing.
  • Measure the rollout with a plain weekly review of every ticket that expired, changed status or was escalated, and a written reason for each.

Why locate and dig-ticket tracking breaks down

Excavation work depends on information that changes from day to day. A locate request is filed, utility responses arrive on their own schedules, tickets carry validity windows set by local rules, and weather or crew changes move the work. When these facts live in separate inboxes, spreadsheets and a foreman's phone, the failure is rarely dramatic at first. A ticket expires while the job is waiting on a reschedule, a locate response arrives in an email nobody forwarded, or a note about a moved marker stays on a clipboard in the truck.

The office usually knows the dates. The field knows what was actually seen. An operations leader needs both in one record, with a clear owner for every status change, so that a date on a screen and a condition in the ground are never confused.

What the agent should track

A useful tracker starts with a short list of records tied to decisions people already make, not with a general assistant that reads everything.

  • Locate request: site address, requesting party, filing date, the notification reference and the response status returned by each utility operator.
  • Ticket: validity dates as printed on the ticket, the local rule that governs renewal, the assigned crew and the planned dig area.
  • Marks and responses: which utilities responded, which marked their lines, which reported no facilities and which responses are still open.
  • Crew notes: photo references, observations about marks, changes to the planned dig path and anything that conflicts with the ticket.
  • Approvals: who reviewed the site, when they did so, and exactly what they were allowed to decide.

The agent should show the source of every field. If a status came from a utility email, the record should link to that email. An unsourced status is a question for a person, not a fact the agent may repeat. The same discipline of field-level sources appears in post orders and incident logs for security firms, where a log entry is only useful if someone can trace where it came from.

Where the safety-critical line sits

The useful question is not whether the agent is accurate most of the time. It is which decisions must stay with a qualified person regardless of accuracy. Use this guide to draw the line before any permissions are granted.

  • The agent may list tickets nearing expiry, draft renewal requests for a human to send, flag a locate response that has not arrived and summarize crew notes into the job record.
  • The agent may draft but not send: messages to utility operators, schedule updates to customers and the daily ticket summary for the dispatcher.
  • A human must decide: whether a site is cleared to dig, whether marks are acceptable, how to work around a marked conflict and whether any work may continue past a ticket's validity.
  • The agent must never change a ticket to active or cleared, mark a locate complete without the utility response on file, or alter an instruction given by a crew lead.
  • Any note that mentions damage, an unmarked line or an unexpected material is escalated at once to the named safety contact, through a path that does not depend on the agent working correctly.

Write these rules into the agent as permissions, not only as instructions in a prompt. Instructions can be misread or overridden by unusual input, while a tool the agent cannot call cannot be misused. The principle is the same least-privilege access used across business software, and the [OWASP authorization guidance](https://cheatsheetseries.owasp.org/cheatsheets/Authorization_Cheat_Sheet.html) describes it in detail. Operations teams that already run quality and nonconformance tracking will recognize the pattern of a logged finding that must be closed by a named person.

A step-by-step plan for the first rollout

  • Step one: list every place a locate or ticket fact is recorded today, including texts, email, spreadsheets and dispatch software.
  • Step two: pick one crew or one region, and define the three fields that matter most for expiry and clearance.
  • Step three: connect read access first. The agent reads the sources and builds a shared view, with no write access to ticket or utility systems.
  • Step four: add draft-only actions such as renewal requests and expiry reminders, each reviewed by the dispatcher before anything is sent.
  • Step five: run a weekly review of every ticket that expired, changed status or was escalated, and record why each one happened.
  • Step six: widen permissions only after the review shows that the agent's flags were consistently correct and that no required human step was skipped.

Expand slowly. A narrow scope with a clear review is easier to trust and easier to correct than a broad one that nobody has time to check.

Illustrative scenario: a crew at a site with a lapsed ticket

This scenario is illustrative and does not describe a specific company. A crew lead messages the office that a site has been rescheduled and asks whether the existing ticket still covers the work. The agent checks the ticket record, sees that its validity window ended before the new date and posts a flag to the dispatcher with the ticket reference and the renewal rule noted in the record. It drafts a renewal request but does not send it.

The dispatcher submits the renewal through the proper channel, and the agent updates the record only when the new response is on file. The crew does not dig until the site supervisor records that the ground has been cleared. Nothing in that sequence required the agent to make a safety judgment. It made a gap visible early enough for a person to act on it.

How Be Human runs this as a managed service

Be Human designs and operates this kind of agent as a managed service. A first engagement defines the one role, the records and tools it needs, the approval boundaries described above and a success criterion the office can observe, such as the weekly review of expired and escalated tickets. Ongoing operations can include monitoring of the integrations, periodic evaluation of the agent's flags against the decisions people actually made, permission reviews when crews or utilities change how they work, and controlled updates when a local rule changes. Memory, tools and permissions stay under your company's control, so the approval record remains yours.

If you are not sure which of your locate or ticket tasks is the right first one to hand to an agent, the free Be Human AI diagnostic is a practical way to find it.

Frequently asked questions

Can an AI agent approve a dig or clear a site? No, and it should not be set up to try. The agent can track locate responses, flag missing marks and show a ticket's validity, but the decision to dig belongs to a qualified person who has reviewed the site. Permissions should block any change to clearance status, even if an input asks for one.

Does this replace the locate notification process? No. The agent works alongside the notification and locate processes your company already uses and the rules set by your jurisdiction. It reads responses and records them, but it does not file or replace official submissions.

What data does the agent need to start? It needs the current ticket list, the locate responses from the recent period and the scheduling or dispatch record that shows where crews are working. Begin with read access, and add write access for drafts only after the office confirms that the fields are accurate.

How do you know it is working? Use a weekly review of expired tickets, escalations and status changes. The question is whether the agent surfaced each gap in time and whether any required human step was skipped. Adjust permissions when the review shows a repeated pattern, not after a single good week.

Your next step

See where an AI agent fits in your company first.

Take the free Be Human AI diagnostic. A few questions about your work show which task to hand to an agent first, what it needs to know and where a person approves.

Be Human editorial guidance. Examples are illustrative unless explicitly attributed. Sources provide background; implementation choices depend on your workflow.

Less busywork. More room for you.

Bring us the work you wish you had time for.

Book a call to try a personalized AI agent free for 7 days ↗Or take the free AI diagnostic →