Skip to content

AI workforce analytics

Turn field crew data into answers.

Construction labor analytics turns recorded hours into answers about where labor went, where overtime is building and which jobs are drifting over budget. ZeroBadge AI lets an office admin ask those questions in plain English. It reads only the company's own records, never edits anything, links every number to the records behind it, and is switched on or off by your admins.

  • Every number links to its records
  • Read-only, your company's data only
  • Access set by your admins

Compare

You asked

How did Riverside compare with Northgate this week?

Riverside recorded +129.9 h on the prior week while Northgate held steady (−1.4 h). Riverside also carries 38.3 h of overtime to Northgate's 17.6 h.

Oct 5–9RiversideNorthgate
  • Recorded hours375.0 h328.9 h
  • Change vs prior week+129.9 h−1.4 h
  • Overtime38.3 h17.6 h
  • Crew on Fri 998
  • Labor cost (loaded)$28,070$24,008
  • Schedule exceptions12

Supporting records

  • 43 shift records · Riverside
  • 39 with overtime
  • Baseline Sep 28–Oct 2

● Grounded in 240 shift recordsRead-onlyYour company's data only

Ask it like you'd ask your office manager

Ask what you'd ask in the Monday meeting.

Type the question the way you'd say it in the Monday meeting. ZeroBadge AI works out the jobsites, crews and dates you mean, runs the calculation against your records and comes back with a short finding, the supporting numbers and the records you can open to check it.

Answers use the same timesheets, scores and forecasts as your dashboard, so the two never disagree. If a question needs something you haven't set up, like a job budget or crew schedule, it says what's missing instead of estimating.

Ask

Which jobs are on pace to finish over their labor budget?

FindingProject to date · through Fri Oct 9

Two of three jobs are projected to finish over budget. Riverside is furthest out: 5,578 h projected against a 5,200 h budget (+7.3%), with 42% of the work still ahead.

Supporting numbers
JobUsedDoneProjectedvs budget
Riverside62%58%5,578 h+7.3%
Harbor Yard71%73%7,234 h−2.2%
Northgate44%44%9,702 h+1.1%

View records · Inputs behind the Riverside projection

  • Labor budget · Riverside

    Entered by the project manager

    5,200 h
  • Completion update

    Entered Fri Oct 9

    58%
  • Recorded hours to date

    Timesheets through Oct 5–9

    3,235 h
  • Harbor Yard · status

    71% of budget used

    On track

Open all 5 records →

Decision it informs. Re-staff or price a change order on Riverside while 42% of the work is still ahead, not at closeout.

● Grounded in labor budgets, entered completion and 120 shift recordsRead-onlyYour company's data only

Answers are computed from records and link back to them. ZeroBadge AI never edits a record.

What it answers

Nine kinds of questions, and the decision each one informs.

Nine areas cover most of what the office asks about crews, each with the call it helps you make.

  • Which jobs used more labor hours this week?

    What comes back

    Recorded hours by jobsite, worker, day or week, with the change against the prior period.

    The decision it informs

    Where to look first on Monday, and whether a crew move last week shows up where you expected it to.

  • How did Riverside compare with Northgate this month?

    What comes back

    Side-by-side hours, overtime, crew size and cost, with the baseline dates stated.

    The decision it informs

    Which project gets the next available crew, and which one can give one up.

  • Where did overtime increase, and on which days?

    What comes back

    Overtime by site and crew against your daily and weekly thresholds, with the shifts behind it.

    The decision it informs

    Add a body or pay the premium: a staffing call made before the next pay period, not after it.

  • Which jobs are on pace to finish over their labor budget?

    What comes back

    Budget used against work complete, labor variance and projected final hours at the current pace.

    The decision it informs

    Re-sequence the work, re-staff, or price a change order while there's still work left to price.

  • Are late starts concentrated at one site?

    What comes back

    Late starts and early finishes against each crew's schedule, by site, weekday and crew.

    The decision it informs

    Fix an unrealistic scheduled start, or handle a one-off in the review queue. They need different conversations.

  • Which records need attention before payroll?

    What comes back

    Missing clock-outs, time outside approved sites, unclassified time and other open exceptions.

    The decision it informs

    What to resolve before the week is approved, so corrections don't land in next week's run.

  • How much paid time went to driving between sites?

    What comes back

    Transit by route and worker, supplier stops and crew transfers, as hours and as labor cost.

    The decision it informs

    Batch supply runs, stage material on site, or change which crew covers which jobs.

  • What moved Harbor Yard's workforce score?

    What comes back

    The weekly score with its drivers: punctuality against schedule, share of the shift on site, transit.

    The decision it informs

    Whether the issue is start times, time away from the site or the routing between jobs.

  • Who's on shift right now, and where?

    What comes back

    Live crew status from each badge's latest report, with week-to-date hours against plan.

    The decision it informs

    Confirm coverage before pulling someone to another job mid-day.

Where the numbers come from

Every number in an answer comes from a record you can open. ZeroBadge AI calculates; it doesn't guess.

The figures come from ZeroBadge's own workforce calculations (timesheets, overtime rules, weekly scores, labor forecasts); the AI explains them. Each answer states its period, baseline and scope, and links to the shifts behind it. Trends and anomalies are flagged as patterns worth a look, never as conclusions about why something happened or how hard anyone worked.

How an answer is built

Four steps between the question and the records.

  1. 01

    You ask

    In plain English, about any site, crew, worker or period your role can see.

    Office admin

  2. 02

    It scopes the question

    Your company and role are applied on the server before anything runs. Date ranges and row counts are capped.

    Automatic

  3. 03

    It computes from records

    Timesheets, schedules, budgets and scores are read, never written.

    Read-only

  4. 04

    You review, then act

    Open the records, check the finding, and make the call. Nothing changes until a person changes it.

    Human decision

Example questions

Ten ways an office admin might ask, in their own words.

Short, specific questions work best. Name a job, a crew or a week when you can, and say if you want a comparison.

  • Labor by job

    "How many hours did Riverside use last week, and how does that compare with the week before?"

  • Overtime

    "Which crews went over 40 hours this week, and on which jobs did the extra hours land?"

  • Budgets

    "Which jobs are on pace to use more labor than budgeted at the current completion?"

  • Late starts

    "Are late starts bunched at one site, or spread across the company?"

  • Transit

    "How many paid hours went to driving between sites last month?"

  • Exceptions

    "What is still open on this week's timesheets before I approve?"

  • Scores

    "What moved Harbor Yard's weekly score?"

  • Live status

    "Who is on shift right now, and at which sites?"

  • A single worker

    "What did this worker's week look like, day by day?"

  • Time off

    "Which of last week's missed days were covered by approved time off?"

What an answer looks like

A finding, the numbers behind it, the period it used and the records to open.

The wording and figures below are an example of the shape, not data from any company. The parts are the same every time.

  • The finding

    Example

    Riverside used 412 hours this week, up 44 hours (12.0%) on last week's 368.

    Why it is there

    A short answer first, so you do not have to read a table to learn the point.

  • Supporting numbers

    Example

    Overtime was 36 hours against 21 the week before. Crew on site averaged 9 a day against 8.

    Why it is there

    The figures that explain the headline, from the same timesheets as your dashboard.

  • Period and baseline

    Example

    Saturday to Friday this week, compared with the prior Saturday to Friday.

    Why it is there

    You can see exactly what was compared, so a number is never read against the wrong week.

  • Links to records

    Example

    Each figure opens the shifts and timesheets behind it.

    Why it is there

    You can check it yourself before you act on it.

  • What is missing

    Example

    "No labor budget is set for Riverside, so no budget comparison is shown."

    Why it is there

    It says what it could not calculate instead of estimating.

Describing a change is as far as an answer goes. It does not say why hours rose, and it does not judge how hard anyone worked.

Guardrails

Limits on what ZeroBadge AI can see and do.

These are product settings, built into how it works.

  • Your company's data only

    Answers draw only on your organization's records. Company data is isolated at the database level, and the AI can't be pointed at another tenant.

  • Read-only

    ZeroBadge AI never edits a timesheet, resolves an exception, approves a week or sends anything to payroll. It reads and explains.

  • Admin-controlled access

    Office admins decide whether the assistant is on for the company and who can use it. Workers and site supervisors don't see it.

  • Summaries, not location trails

    Answers use hours, sites, schedules and scores, not raw location history, which is deleted after 90 days.

  • People make the calls

    An answer can point to an open exception; a supervisor still resolves it, and the audit trail records the change.

Limits

What ZeroBadge AI will not do, and when it will say it cannot answer.

  • It will not answer from outside your records

    If the question needs something that was never recorded, such as installed quantities or materials cost, it says so. It does not fill the gap from general knowledge.

  • It will not explain why

    It can show that overtime rose at one site. It cannot tell you whether the cause was weather, rework or a late delivery, and it does not guess.

  • It will not give legal or payroll advice

    Overtime rules and pay rules vary by state and by agreement. It reports hours against the thresholds you set, and it does not interpret the law.

  • It will not change anything

    No timesheet edits, no approvals and no payroll sends. Those stay with a person, as in the timesheet approval workflow.

  • It will not reach another company

    Your company and role are applied on the server before anything runs, and limits on date range and rows apply to every question.

  • It is not for everyone on your team

    Admins decide who has access. Workers and site supervisors do not see it.

ZeroBadge AI

Ask ZeroBadge AI a question about your crews

Type a question the way you would ask a project manager. ZeroBadge AI answers from your company's own badge records, links every number to the records behind it, and cannot change anything. Admins decide who gets to use it.

  • You ask

    Which jobsite ran the most overtime last week?

    The Elm Street warehouse, with 46 overtime hours across 9 workers, against 31 hours at the Harbor Road clinic. The 46 is each worker's hours past your weekly threshold added up, and every worker's line opens their timesheet.

  • You ask

    Which workers had the lowest weekly scores, and why?

    Two workers scored under 60. Both started late on 3 of 5 scheduled days, and neither had a missed day or approved time off. One score carries a lower confidence label because that badge missed two reports.

  • You ask

    How many labor hours went to the Pine Ridge school job this month?

    412 hours so far, 268 regular and 144 overtime. That is 61% of the 680 hours budgeted for the job. Drive time to the job is counted separately as transit.

Answers come only from your company's own records. Every number links to the records behind it. Read-only; admins control access.

How ZeroBadge AI works

Questions

Questions about ZeroBadge AI.

What kinds of questions can ZeroBadge AI answer?
Questions about your workforce: labor hours by site, worker or period; week-over-week and site-to-site comparisons; overtime; labor budget used and projected final hours; late starts against schedule; open exceptions; transit and crew movement; weekly workforce scores and their drivers; and who is on shift right now.
Where do the answers come from?
From your company's own badge-recorded shifts, the timesheets built from them, your crew schedules and the job budgets you enter. The numbers are calculated by the same rules as your reports, and every answer links to the records behind it.
Can ZeroBadge AI change timesheets or approve payroll?
No. It's read-only. Corrections, approvals and payroll stay with the people responsible for them, and every correction keeps an audit trail. See payroll integration for how approved hours reach payroll.
Who in my company can use it?
Office admins control it. They switch it on for the company and decide who has access. It isn't available to workers or site supervisors, and it only ever sees your company's data.
Do I still need the dashboards and weekly reports?
Yes. They show the standing picture; the AI answers the follow-up you'd otherwise build a spreadsheet for. See workforce productivity for the scores it explains.

Last updated

Get started

Ask it about your own crews.

We'll set up a walkthrough with questions from your jobs: labor by site, overtime, budgets and crew movement.

What we will walk through

  • 01The questions your office asks every week
  • 02How each answer links to its records
  • 03Access controls and guardrails for admins