The Problem How It Works Under the Hood Features Team Request a Pilot
Weld torch

Win more jobs,
lose less money.

Scale Fab reads your engineering drawings and tells you exactly how long a weld job will take

See the Problem Request a Pilot
Problem

Weld shops are flying
blind on labor costs.

Estimators have been using the same flawed assumptions for decades. The result is predictable — and expensive.

40%
of weld shops miss ship dates because labor estimates were wrong from the start — not because the work was done poorly.
157,000
welders retiring this decade — taking decades of estimating knowledge with them permanently.
Solution

Scale Fab tells you exactly how long
a weld job will take — before you start.

Upload a drawing. Scale Fab reads the geometry, identifies every joint, and returns a precise labor estimate before the contract is signed.

01

Upload Drawing

Drop in your welding drawing from any CAD platform.

02

AI Reads the Drawing

Scale Fab's AI reads weld geometry, joint types, material specs, and process callouts automatically — with no manual data entry required. Every joint classified before you touch it.

03

Real Shop Data Fills the Gaps

Live measurements from your welding machines replace the assumptions that kill your margins — capturing actual arc-on time, purge, fit-up, and QA instead of guessing.

04

Receive Your Estimate

Get a precise, itemized labor estimate broken down by weld ID, process, and sequence — ready to drop straight into your bid and defend every line item.

No gut feel. No averaged assumptions. No margin surprises.
Under the Hood

From Drawing to
Estimate in
Seconds

Scale Fab extracts structured data directly from your spool drawings — reading every weld ID, material callout, and joint type automatically.

IC-001-SPOOL.pdf — Spool Drawing Rev A
IC-001-SPOOL isometric drawing
Analyzing... 6 welds found
What Scale Fab Read From This Drawing
Drawing IC-001-SPOOL.pdf
Welds detected 6  (A · B · C · D · E · F)
Butt welds — 100mm 5
Butt welds — 80mm 1
Pipe material A106-B  SCH 40
Fitting material A234-WPB
Actual arc-on time 21%  (measured) vs 38% assumed
14.2
Estimated Weld Labor Hours Includes fit-up, purge & QA time
Capabilities

Built for the shop floor
and the bid room.

Every feature was designed with estimators, project engineers, and shop managers in mind.

Process Modeling

Process-Aware Labor Rates

Rates calculated by process — SMAW, GMAW, GTAW, SAW, FCAW — factoring deposition rates, electrode efficiency, and interpass time. Not one blended average for everything.

Bid Integration

Export Anywhere

Push itemized estimates to Excel, your ERP, or quoting software via API with configurable rates and markup — no re-keying.

Continuous Learning

Accuracy That Compounds Over Time

Every job Scale Fab processes makes the next estimate tighter. The system learns your shop — your welders, your processes, your actual throughput — and gets more accurate with every spool you run.

Full Audit Trail

Every Assumption Explained

Every detected joint, standard, and calculation is visible and traceable to a drawing reference. Estimators can review, override, and defend every number — nothing is a black box.

Team

Built by someone who's
already done this.

Domain expertise and technical foundation built together from day one. This is not a software team that stumbled into welding.

Tony Flores, Founder
Tony
Flores
Founder
  • Two internships at SpaceX, working directly in high-stakes weld production environments
  • Built a model at SpaceX that was roughly 90% more accurate than junior engineers — and matched or exceeded the most experienced senior engineers on the team — using only basic weld dimensions as input. Scale Fab extends that foundation with richer data and a continuous feedback loop.
  • The key finding: the limiting factor in estimating accuracy is not the skill of the estimator — it's the absence of a systematic method for learning from actual job outcomes. Scale Fab is that method.
  • Prior experience at Halliburton in high-stakes industrial environments — understands how estimating decisions get made and where they go wrong
SpaceX Halliburton
IBM HCSS
The Math

An 8% improvement in estimate accuracy
on a $20M backlog recovers
$1.6M in margin per year.

Without adding a single welder. Without winning a single new job. Without changing your production process. Just by knowing what jobs actually cost before you sign them.

Request a Pilot
Get Started

Stop leaving margin
on the table.

We're looking for a handful of shops to join the Scale Fab pilot. We embed on-site, set up AI tracking on your floor, and show you what accurate estimates actually look like — compared side by side with what you're submitting today.

See the Problem First