"Duct weight drift used to kill us. You'd take off a job at 14,200 pounds of sheet metal and the shop would fabricate it at 15,800. That's a real number — that's our margin on a $3M tenant improvement. Pilrs catches the gauge transitions the old tools missed. First quarter we ran it, zero drift complaints from the shop."
Pacific Mechanical started in a rented shop in Milwaukie in 2004 when founder Dale Reilly cashed out of a larger Portland mechanical and went on his own with four ironworkers and a sheet metal brake. Twenty-one years later the shop runs 60 people — 14 in fabrication, 38 in the field, 8 in the office — across commercial HVAC retrofit, light industrial process work, and the ground-up tenant improvement projects that have defined the post-2020 Portland market. They self-perform ductwork fabrication and equipment setting; piping and controls subout to two long-standing partners.
The estimating department, led by Tom Reilly (Dale's son), ran on Trimble Autobid Mechanical for equipment and piping, FastDuct for ductwork takeoff, and a tonnage-and-gauge Excel workbook Dale built in the mid-2000s that everyone still used to cross-check shop weights. The workflow worked on small jobs. On anything over $2M — which increasingly was every job after 2023 — the hand-off between FastDuct and the tonnage sheet was the bottleneck. Tom was routinely pulling 60-hour weeks to keep four bids moving simultaneously.
The signature failure mode at Pacific Mechanical was duct weight drift. FastDuct produced fitting counts and linear feet, the tonnage sheet applied gauge-by-gauge weight factors, and the shop fabricated from the approved cut list. Problem: the cut list came from the engineer's drawings, which specified gauges by static pressure class — and the class boundaries shifted partway through some runs, where a 2" w.c. duct transitioned into a 4" w.c. plenum. Manual takeoff frequently applied the lower class weight to the entire run. On a tenant improvement with 14,200 lb of takeoff and 1,600 lb of actual plenum weight, that was a $9,800 sheet metal miss before labor.
The second wound was equipment lead-time blindness. A 40-ton rooftop unit with an 18-week lead time in 2025 needed to be flagged at bid time so the GC could plan around procurement. Autobid didn't track live supplier lead times — Tom kept a separate spreadsheet that was usually 3 weeks stale. Three times in 2024 Pacific Mechanical won a bid, then blew the schedule when the RTU showed up two weeks after hot weather set in, and on design-assist contracts the liquidated damages absorbed the entire project margin.
Throughput was the third problem. Nine bids a month on a three-estimator team meant Pacific Mechanical was walking away from about 40% of the RFPs the GCs were sending them. Their bid-hit rate was a respectable 22% — but applied to nine bids, that's two wins a month, which was not enough backlog to keep the shop cutting at capacity. Dale made the call in late 2024: find a way to double the bid volume or start laying off sheet metal apprentices. Neither option was acceptable.
Tom piloted Pilrs in March 2025 on a single 88,000 SF light industrial retrofit in Hillsboro. The AI takeoff pulled duct, fittings, and gauge transitions directly from the M-sheets with explicit class boundary flags — exactly the thing the manual workflow had been missing for 20 years. On the pilot job, Pilrs's sheet metal tonnage landed within 1.8% of the shop's actual fabrication weight, versus the 8–11% drift Pacific Mechanical had been living with.
Rollout in production started with ductwork only. Piping and equipment stayed in Autobid through Q2. By June the team had integrated Pilrs output into the legacy Excel tonnage sheet for cross-check, and by August they retired the tonnage sheet entirely. Autobid remained for equipment schedules and piping — Pilrs exports piping LF and fitting counts into Autobid via CSV, and Autobid handles the pricing assembly.
The one thing Pacific Mechanical insisted on was shop-drawing lag integration. Pilrs now flags any equipment item with a lead time exceeding the project's procurement window and kicks that to the GC as a clarification line in the bid package. That feature alone has changed how Pacific Mechanical negotiates escalation clauses — they now win 4 out of 5 RTU escalation protection requests, versus 1 in 10 before.
Pilrs ingests every mechanical sheet and extracts verified quantities in minutes, eliminating symbol-fatigue miscounts.
Every takeoff is reviewed against construction specs before it reaches the estimator — no hallucinated quantities, no phantom line items.
CSV export into Trimble Autobid Mechanical preserves the existing pricing and assembly library — no rip-and-replace.
Monthly bid volume climbed from 9 to 14 in the first full quarter on Pilrs. Tom's 60-hour weeks dropped to 45. The bid-hit rate moved from 22% to 31% — partly because Pacific Mechanical was bidding on a broader slice of the available work, and partly because the bids themselves were tighter: the duct weight drift fix alone let them price sheet metal 2% lower on 70% of jobs without touching margin.
The most telling metric is one Tom didn't expect to track: complaints from the fab shop. Pre-Pilrs, the shop foreman called Tom roughly twice a month with a cut list that didn't match the field measurement or the gauge schedule. Since August 2025, that number has been zero. The shop is cutting to spec, the field is hanging to spec, and the estimating office isn't getting the 7 AM phone call anymore.
Upload a plan set and see the same takeoff velocity these contractors unlocked. 14-day trial, no credit card required.