What a scope gap actually is
A scope gap is any piece of work the spec implies you'll do but the takeoff or bid form doesn't explicitly capture. The owner's engineer drew it. The CSI spec referenced it. The bid form line item didn't. So when your bid gets accepted, you're contractually on the hook to perform the work — but you didn't price it. The cost eats your margin or worse.
Three concrete examples from real bids: a vibration-isolator requirement buried in §23 05 13 that the MEP sub didn't see and didn't price (+$11K post-award rework); a Suffolk County DOH witness fee mandated by a tank spec but not in the bid form (+$8.4K); a curtainwall mock-up requirement in 08 44 13 that nobody priced (+$72K). None are exotic. All are the kind of thing that hides between a 280-page project manual and a 142-sheet drawing set.
Where scope gaps hide
There are four high-probability hiding places. Knowing them is most of the battle.
1. Spec-vs-drawings conflict
The drawings show one fixture; the spec calls for another. The bid form picks one. The other one becomes a change order. Whichever costs more is what you owe.
2. Addendum reversals
Addendum #2 quietly changes a spec section. The base bid form wasn't updated. The estimator priced the original spec. Per the standard contract clause, “the addendum date controls” — and you're building to the addendum.
3. Equipment schedule vs. floorplan
The mechanical schedule lists 18 rooftop units. The floorplan shows 16 mounting locations. Two units have no location. Either two are missing from the plan, or two were added late to the schedule and never reconciled. Either way, you priced 16 and you're installing 18.
4. Demolition vs. new-work boundary
The demo plan stops at a wall. The new-work plan starts six inches into a different room. The strip between is implied scope and never explicitly assigned. Whoever bids low ends up owning it.
The manual estimator method
The standard manual method is a four-pass review: read the project manual cover-to-cover, then drawings sheet-by-sheet against a takeoff template, then every addendum against the base, then the bid form to make sure every spec-required item has a line. A good chief estimator on a 600-page bid does this in roughly 14–20 hours. Most are doing 6–10 hours and praying.
The places this manual method fails: parsing 600+ pages without losing concentration; carrying spec-vs-drawing conflicts across sittings; remembering every addendum change without a tracking sheet; treating equipment schedules as legible data rather than impressionistic art.
What BidLedger does instead
BidLedger runs a six-stage pipeline against the entire bid set in under 30 seconds on most bids. The verifier stage is the one that catches scope gaps: every LLM-proposed finding is checked word-for-word against the cited page. If the LLM hallucinated a clause, it gets rejected. If the LLM found a real spec requirement the bid form doesn't reflect, that finding ships in the “owner RFI” section of the brief along with the exact spec citation and the contradiction.
On a typical 600-page bid the pipeline surfaces 80–150 distinct findings, ranked by cost impact. The estimator's job becomes confirmation and judgment, not exhaustive reading. Free first bid: forward an email at getbidledger.com/send, get the brief back within one business day.