The scope is in the documents.
A commercial bid set runs to hundreds of sheets plus a spec book. Each trade’s scope is spread across plans, details, schedules, and spec sections, and today an estimator reads all of it by hand before bid day.
RAW Logic is building a preconstruction platform that reads commercial drawing sets and specs, extracts scope by trade, and cites the sheet and page behind every line, so estimators catch scope gaps before bid day.
In development. Early access for estimating teams at GCs and trade contractors.
A commercial bid set runs to hundreds of sheets plus a spec book. Each trade’s scope is spread across plans, details, schedules, and spec sections, and today an estimator reads all of it by hand before bid day.
Each scope item carries the sheet, detail, or spec section it came from, down to the page. An estimator can open the drawing behind any line and check it, and every edit stays on the record.
When the drawings conflict or leave something out, the platform says so. The item is marked uncertain, the sheets involved are listed side by side, and an RFI is drafted for the estimator to send or discard.
Founded by people who have worked inside a contractor’s office, next to the estimators.
Roy Rosenberg built the data backbone of a billion-dollar general contractor from the ground up, pulling Procore and Vista into one source of truth that the company’s monthly Project Executive meeting runs on. Working alongside its preconstruction team, he saw where bid time goes: estimators reading full drawing sets by hand to build scope for each trade, and gaps that only surface after award.
RAW Logic turns that problem into a product. Jeff Webster, co-founder, is the design partner on every screen.
Request early access
I’ve been writing code since I was 13, including a few things I shouldn’t have. That became a career in cybersecurity, then years shipping software, long before I set foot on a jobsite.
When I did, I kept seeing skilled people buried in documents. Superintendents rebuilding the same report every week. Estimators reading a few hundred sheets by hand to build scope for one trade, then learning after award what the set said on a sheet nobody had time to open.
So that’s what I’m building. The person who writes the code is the person who picks up the phone.
Roy
A scope sheet for each trade, built from the full drawing set and spec book. Every line shows where it came from and whether it passed the trade checks, and every open question comes with a drafted RFI. The example below was made for this site; it is not output from a client project.
| Pass | Description | Result |
|---|---|---|
| 1 | Extract scope from plans, details, schedules, and Division 09 specs | 38 items |
| 2 | Check each item against drywall trade rules and the rest of the set | 4 flagged |
| 3 | Draft RFIs for items the documents do not resolve | 2 drafts |
| Item | Source | Check |
|---|---|---|
| 5/8″ Type X board, 1-hr corridor walls | A-501 / Det. 4 | OK |
| 3-5/8″ 20 ga studs at shaft walls | A-201, S-001 | OK |
| Abuse-resistant board to 4′-0″ AFF | 09 29 00, p. 4 | OK |
| Level 5 finish, lobby feature wall | 09 29 00, p. 7 | Flag |
| Rated head-of-wall joints | A-502 / Det. 11 | Flag |
Illustrative example built for this site. Not output from a client project, and not a measure of accuracy.
A set of agents works through the documents in three stages, each with its own job. Open a stage to see what it does on a real bid set.
Every drawing sheet and spec section is read at full size, as an image and as text, and scope is pulled out item by item, sorted by trade.
A few hundred sheets split into scope for drywall, doors and hardware, concrete, and every other trade in the set.
Plans, sections, details, schedules, and general notes are read sheet by sheet. Spec sections are matched to the trades that own them, and addenda are applied on top.
Output: a scope list for each trade, with the sheet and page on every line.
Separate agents check each line against the rules for its trade and against the rest of the set: specs against plans, schedules against details, notes against assemblies.
A rated wall on the plan with no rated assembly in the details gets caught here.
Each trade has its own rule set. When two documents disagree, or a required item has no source, the line is marked uncertain and both sources are listed. Nothing is filled in by assumption.
Output: every line marked as checked or flagged, with the reason on the flag.
Scope sheets for each trade and draft RFIs for every open question, ready for the estimator to review, edit, and price.
Scope sheets to send to subs for pricing, and RFIs to send to the design team.
Nothing leaves without estimator review. Edits are kept, so the record shows what the platform found and what a person changed.
Output: scope sheets and RFI drafts, each line traceable to the documents.
One project means hundreds of full drawing sheets per run. The platform reads every one of them.
The estimator still prices the job. The platform makes sure nothing is missing from what gets priced.
The drywall sub should not be the one who finds the missing rated wall.
Scope pulled from plans, details, schedules, and specs, and sorted into the trades you buy out, so each scope sheet starts from the full set.
Ask about scope by tradeAny line, back to the sheet it came from.
Every scope item names its source: sheet number, detail, spec section, and page. When a sub or owner questions a line, the answer is already on it.
Ask about citationsUnclear documents get a flag, not a guess.
Each line is checked against the rules for its trade and against the rest of the set. Conflicts and missing information are marked uncertain, with the sheets involved listed side by side.
Ask about trade checksThe questions you need answered before bid day, already written.
Scope sheets ready to send to subs for pricing, and a draft RFI for every open conflict. The estimator reviews, edits, and sends.
Ask about scope sheets and RFIsThe platform proposes, the estimator decides. Nothing goes to a sub, an owner, or a design team until a person approves it, and every edit is logged.
Where the documents are unclear, the output says so. Uncertain items are marked and sourced, never filled in to look complete.
Built by someone who came up in cybersecurity. Project documents are encrypted, kept separate per company, and auditable end to end. Your drawings are never pooled and never used to train models.
Commercial drawing sets and spec books as PDFs: plans, sections, details, schedules, general notes, and addenda. One project is typically hundreds of full-size sheets per run, and every sheet is read.
Estimators and preconstruction teams at general contractors, and estimators at trade contractors who bid from the full set.
No. It builds scope your team can price, with the source on every line. Pricing stays with the estimator.
The item is marked uncertain, the conflicting or missing sources are listed side by side, and an RFI is drafted. The platform does not fill gaps with assumptions.
No. It takes on the first full read of the documents, so estimators spend their time checking scope, pricing it, and working with subs.
Not yet. RAW Logic is in development. A small number of estimating teams are being invited to run past bid sets through it and compare the output with what they actually bid.
They stay yours. Project documents are encrypted, kept separate per company, and never used to train models.
For a few contractors, yes. Write to roy@rawlogic.io.
RAW Logic is in development. We are inviting a small number of estimators and preconstruction leads at general contractors and trade contractors to run a past bid set through the platform and compare the scope sheets with what they actually bid. You see the output on your own documents; we learn where it misses.
Your documents are used only for your project. Tell us your company, your role, the trades you bid, and a typical project size.
Tell us the trade and project type where scope gaps hurt you most, and we will show you how the platform handles it today.
Email usor write directly: roy@rawlogic.io
“Sounds verytabular… fine, it’s better now.”
Jeff Webster
Co-founder, and impossible to impress