Built for mid-market GCs · not AI demos
Defensible VE proposals from your plan set.
Every suggestion cited to the source sheet.
Upload your CD set — get a VE log of substitutions, issues, and design observations, every line cited to sheet, spec section, and schedule row.
See extraction accuracy before you talk to anyone.
- Substitution
VRF heat-recovery → staged split DX
22–28%M-20123 81 26SourceSheet M-201 · Schedule HVAC-14Spec23 81 26 — substitution permittedSupplier basisDaikin RXYQ · Mitsubishi City MultiDownstreamOr-equals · 30-day submittal window - Substitution
Rift white oak → engineered white oak
18–25%A-61109 64 29SourceSheet A-611 · Schedule FIN-03Spec09 64 29 — substitution permittedSupplier basisShaw Patcraft · Mohawk GroupDownstreamLEED MR-1 unchanged · warranty 25y - Design Insight
Type A luminaire overspec on back-of-house
12–17%E-40126 51 00SourceSheet E-401 · Schedule LF-A / LF-A1Spec26 51 00 — substitution permittedSupplier basisLithonia WF · Cooper 6LFDownstreamArchitect review · photometric calc - Drawing Issue
Fixture schedule row 7 not located on plan
—A-301 ↔ Q-203—SourceSheet A-301 ↔ Q-203 · Schedule FIX-07Spec—Supplier basis—DownstreamRFI · flag for review
3 days → 30 min
The three days you’d spend re-keying schedules and legends, compressed to half an hour. Every line still cited to sheet, spec, and schedule row.
The takeoff isn’t the part you want to automate.
You’re a hyper-specialist. You spent years learning to catch the gotchas in a technical book that nobody else reads carefully. Then you got the wrong tool: raster-based software with no intelligence for selecting lines or planes, specs that live in a separate PDF your software pretends doesn’t exist, and AI takeoff tools that are, in one estimator’s words, “hit and miss on the particular drawings we use, which are full of info.”
VEbase automates the grunt so you don’t hire around it. It doesn’t do your job. It gives you back the hours you were never supposed to spend re-keying schedules — so you can spend them interpreting design intent and proposing substitutions the architect won’t kill.
013–4 days per bid re-keying schedules and legends
“Used to spend three to four days per project doing manual takeoffs in Bluebeam and OST.”
02AI takeoff tools that need double-checking anyway
“Would still require double checking all the information it produces. Useful only for rough orders of magnitude.”
03VE suggestions the architect kills on sight
“The architect was sued for acquiescing to changes made by a construction firm, despite not recommending them.”
Upload the CD set. Review the VE log. Approve with sources.
- 01
Upload plan set
Drawings, specs, schedules, addenda. Up to 500+ pages. We parse the plan set as structured data, not pixels — schedule rows, spec sections, plan callouts, legend keys. The intelligence the raster-based tools never had.
- 02
Review the VE log
Every suggestion shows: source sheet, spec section, schedule row, and downstream impact — warranty, LEED, coordination, or-equals submittal timing. No “AI says so.” No black box. No hallucinations dressed up as confidence.
- 03
Approve with sources
Export to Excel or Bluebeam markup. Every line is defensible to the owner, the architect, and your insurer. Your PM sees the sourced reasoning. Your firm sees fewer rejected substitutions.
Not a takeoff tool. VEbase is a VE tool. We stay in our lane. Keep Bluebeam. Keep your cost database. Keep your subs.
Every line is defensible. Or we don’t output it.
Every VEbase suggestion ships with five things an architect or owner can verify in under a minute. If VEbase can’t cite all five, it abstains. Flag for review is a feature, not a failure.
- 01A-501 · rev 04
Source sheet
The exact drawing page and revision ID.
- 0208 11 13.2.1 — or-equals permitted
Spec section
The CSI-formatted spec clause that permits or restricts substitution.
- 03Door Schedule · row D-14
Schedule row
Which row of which schedule produced this line.
- 04LEED MR-1 unchanged · 30-day submittal
Downstream impact
Warranty, LEED credit, coordination risk, submittal timing.
- 05Daikin RXYQ-TATJU · March 2026 catalog
Supplier / catalog citation
The unit-price basis — not a number we made up.
Built for estimators who’ve already been burned.
| Capability | VEbase | Togal.AI / Kreo | Procore Estimating | PlanSwift / OST |
|---|---|---|---|---|
| Parses specs as structured data | visual geometry only | |||
| Every suggestion cited to source sheet + spec | ||||
| VE proposals, not just takeoff | primary | |||
| Excel + Bluebeam export, no lock-in | data lock-in widely reported | |||
| Cancel anytime, transparent pricing | varies | enterprise contract | licenses revoked 2025 | |
| Built for mid-market GC (15–30 bids/yr) | varies | enterprise |
Priced against the estimator you can’t hire.
A full-time estimator runs $70,000–$90,000 per year — before benefits, overhead, and the nine months to ramp. Precon is $8,388 a year — roughly 10% of one junior estimator, billed monthly, cancel anytime, your data exports to Excel or Bluebeam on your way out.
No 50%renewal hikes. No perpetual-license bait-and-switch. No “call sales for a quote.” Unlimited estimate revisions on every paid plan — schematic through GMP — at no extra charge.
“The most dangerous pattern is not that AI makes errors. It is that AI produces errors with an authoritative tone and speed.”
Fair questions.
Upload a plan set. See what you get back.
No demo call. No sales email thread. Upload a CD set, watch the extraction, review the VE log. If the first suggestion isn’t defensible, you’ll know in five minutes — and nothing stops you from closing the tab.
- Transparent pricing
- Cancel anytime
- Excel + Bluebeam export
- No perpetual-license tricks