A 90-day pilot on one work unit: cut coordination drag, return general-foreman and foreman hours to the field, and put cited safety and spec answers in a foreman's hand — without a call, a binder, or a hunt through Teams.
Forge doesn't do smoke-and-mirror cost savings. "Hours saved × hourly rate" looks big and changes nothing on the P&L — returned salaried hours are capacity, not cash. Nobody comes off payroll. So we lead with the line closest to real money.
| Scenario | Idled hrs/mo | Crew-hrs | Value / mo |
|---|---|---|---|
| 15% of lookup time idles a 3-person crew used — conservative | 3.15 | 9.45 | $1,134 |
| 25% of lookup time idles a 3-person crew base case | 5.25 | 15.75 | $1,890 |
| 40% of lookup time idles a 3-person crew | 8.40 | 25.20 | $3,024 |
| Activity | Hrs/mo | Rate | Value/mo | Basis |
|---|---|---|---|---|
| Route sheet & scheduling | 16 | $130 GF | $2,080 | ~45 min/day: collect, compile, correct, distribute — every working day |
| Spec & safety lookups | 21 | $120 FM | $2,520 | 8 crews × ~2/day × ~8 min hunting Teams, binders, or the GF |
| 811 locate status | 7 | $125 mix | $875 | ~10/wk × 10 min round trips: foreman → GF → portal → foreman |
| 811 locate drafting | 6 | $130 GF | $780 | ~4/wk × 20 min, single-property & short-run, drafted for GF approval |
| Gross hypothesized | 50 | $6,255 | Per month, per work unit |
Returned admin hours from salaried staff don't hit the P&L. A rigorous case applies a realization haircut — the share of returned time that converts into useful output rather than being absorbed as slack. We model three and use the middle:
This line alone doesn't justify the investment, and we won't pretend it does. Stacked with crew standby it clears comfortably. Presented alone, it should be discounted — which is why it isn't the headline. The 50 hours is one general foreman's estimate of his own unit's drag; we're not asking anyone to take it on faith, we're proposing a cheap test with the breakeven set at 40% of it.
This pilot came out of a safety initiative discussion. The mechanism is narrow and testable: when the right clearance, approach distance, or procedure is one text away — cited to the page in AES's own manual, from the phone a foreman already carries — it's more likely to actually get checked than when the answer costs a call, a binder, or a hunt through Teams. Blue does not make the safety decision. It removes the friction between a qualified worker and the standard.
| Unit headcount (approx.) | 25 |
| Expected recordables/yr @ 2.3 per 100 FTE | ~0.57 |
| 5% reduction | $1,360/yr |
| 10% reduction used | $2,721/yr |
| 20% reduction | $5,441/yr |
Expected value is the honest way to carry this line: $227/month. One unit of 25 people doesn't generate many claims.
To be explicit, because these two boxes must not be conflated: the expected value is $227/month. The figures at left are tail risk, not a forecast. Forge is not claiming HWC will avoid a claim, and this proposal does not rest on it. The point is narrow: the downside of running the pilot is capped at one month's fee. The downside of a preventable incident is not capped at all.
Workers' comp for power line and utility construction (NCCI class code 7538) is priced off the Experience Modification Rate, recalculated annually on three years of claims history. An EMR above 1.0 raises premium on every payroll dollar — and in this industry it can disqualify a contractor from bidding work at all. Safety adherence isn't only a cost line here; it's a revenue-eligibility line. Claims avoided this year compound into premium and bid eligibility for the next three.
| Line | Conservative | Base case |
|---|---|---|
| Crew standby avoided — closest to cash | $1,134 (15% idle) | $1,890 (25% idle) |
| Hours returned — capacity | $3,128 (50% real.) | $4,378 (70% real.) |
| Safety expected value | $227 (10% red.) | $340 (15% red.) |
| Monthly benefit | $4,488 | $6,609 |
| Less: monthly fee | ($2,500) | ($2,500) |
| Net monthly benefit | $1,988 | $4,109 |
| Return multiple | 1.8× | 2.6× |
| 90-day net (after 2-wk commissioning) | +$2,029 | +$6,911 |
| Payback | 1.8 months | 1.3 months |
| Year-1 net benefit | $23,056 | $48,508 |
| Input | Value | Source & note |
|---|---|---|
| Foreman base rate | $60.00/hr | Provided by HWC (G. Crabtree) |
| General foreman base rate | $65.00/hr | Provided by HWC (G. Crabtree) |
| Foreman fully burdened | $120.00/hr | Provided by HWC — implies a 2.00× burden multiplier |
| GF fully burdened derived | $130.00/hr | $65 × 2.00 burden. Overwrite with the actual figure if it differs. |
| Unit composition | 1 GF · 8 foremen | Indianapolis unit, per field walkthrough |
| Crew size behind a foreman assumed | 3 | Conservative. Used only for standby modeling. |
| Recordable injury rate | 2.3 / 100 FTE | BLS, private industry, 2024 — lowest since 2003 |
| Avg lost-time claim, all causes | $47,316 | NCCI Workers Comp Statistical Plan, accidents 2022–23 |
| Avg burn claim (arc-flash proxy) | $64,973 | NCCI, by cause of injury, 2022–23 |
| Avg fall/slip claim | $54,499 | NCCI, by cause of injury, 2022–23 |
Documents loaded, unit configured, foremen onboarded. We measure today's actual route-sheet build time, lookup frequency and duration, and locate round-trips — before Blue changes anything. This becomes the denominator. Blue produces no claimed value during commissioning, and the 90-day math above reflects that.
Every interaction logged: question asked, answer given, source cited, time to answer. Route sheets timestamped. Locate checks counted. No self-reported numbers.
Baseline vs. actual, in hours and dollars at HWC's own rates. Answer accuracy and citation rate. Foreman adoption. Delivered to leadership as-is.
| Metric | Target | Why it matters |
|---|---|---|
| Hours returned / month | ≥ 20 (breakeven) | The floor. Below this the pilot hasn't earned continuation — and we'll say so. |
| Route sheet build time | 45 min → < 10 min | Directly measurable, every working day |
| Answer accuracy | Spot-audited | HWC's safety lead has standing rights to audit any Blue answer against the source document. Every answer carries its citation, so verification takes seconds. |
| Citation rate | 100% | Every safety answer cites document, section, and page — or Blue declines and routes to the GF |
| Foreman adoption | ≥ 6 of 8 active | An unused tool returns nothing. Adoption is the leading indicator of everything else. |
| Locate round-trips eliminated | ≥ 60% | The most visible daily coordination friction |
Billed monthly in advance. Cancel any time in the first 90 days on 30 days' notice. Hardware is Forge-owned and retrieved; data destroyed and certified. No penalty, no clawback, no minimum during the pilot window.
Reviewed and approved by USCCO IS&T. No domain join, no AD, no VPN, no Azure, no OAuth, no M365 or Teams. Runs on an HWC-supplied hotspot. Nothing to provision, nothing to de-provision, no ports opened.
It surfaces the cited standard from AES's own manual and routes to the GF when it can't cite one. The qualified worker decides, every time. This is a faster path to the manual — not a substitute for it, and not an authority over it. Answer accuracy is spot-audited by HWC's safety lead throughout.
It comes from a field walkthrough, not a measurement — which is exactly why weeks 1–2 baseline it. If the real number is 25 hours, the pilot still clears breakeven. If it's below 20, the scorecard will say so plainly and HWC shouldn't continue.
Forge is a small company. That risk is real, and it's why the pilot is structured this way: one device, one unit, cancel on 30 days' notice with hardware retrieved. HWC is never deep in, and nothing about the unit's operation becomes dependent on Forge remaining heroic.
Direct email/calendar comms, 811 ticket submission, and project-management support are deliberately out of scope and not priced here. Each requires its own scoping and IT review — as VP IS&T directed. We're proving the narrow version first, on purpose.
One work unit, 90 days, $3,300 to start. If the numbers aren't there at day 90, cancel and we pull the hardware.