Spec Book to Submittal Register (CSI MasterFormat → Procore CSV)

Upload a construction spec book / project manual PDF and get, in about a second per 500 pages:

1. CSI MasterFormat section index: every section with number, title, division and page range.
2. Submittal register: every submittal required in PART 1 of each technical section: product data, shop drawings, samples, O&M data, warranties, certificates, test reports, qualification data, LEED/CHPS documentation, mix designs, and more. Each one comes with its paragraph reference (e.g. 1.3.A) and page number, so you can check it against the spec.
3. Procore-ready import CSV, using the exact column headers of Procore's Submittals import template.

The extraction is rules-based: no AI, no LLM, no third-party APIs. Your spec book is not sent anywhere, and the same input always gives the same output.

Who it is for

General contractors, subcontractors, project engineers and submittal coordinators who build submittal logs by hand at the start of every project. A 500-page project manual usually takes a day or more to log by hand. This Actor gives you a first-pass register to review in minutes.

Example input

The Console form is prefilled with a real public spec book: the 380-page project manual for the Story County (Iowa) Justice Center Remodel. Click Start without changing anything:

{
  "pdfUrls": [
    "https://www.storycountyiowa.gov/DocumentCenter/View/16034/2024-09-11-Project-Manual-SC-Justice-Center-Remodel?bidId="
  ]
}

Example output

That run takes about 6 seconds, finds 82 CSI sections (79 validated) and returns 171 submittal rows. The demo book is free. One register row (run of 1 Nov 2026, trimmed):

{
  "recordType": "submittal",
  "sectionNumber": "03 30 00",
  "sectionTitle": "Cast-In-Place Concrete",
  "division": "03",
  "divisionName": "Concrete",
  "article": "1.04 SUBMITTALS",
  "paragraphRef": "1.04.B",
  "page": 66,
  "submittalNumber": "033000-01",
  "title": "Product Data",
  "submittalType": "product data",
  "category": "Action",
  "procoreSubmittalType": "Product Information",
  "description": "Submit manufacturers' data on manufactured products showing compliance with specified requirements and installation instructions."
}

The OUTPUT record summarises each book and links the SUBMITTAL_REGISTER.csv, PROCORE_IMPORT.csv and SECTION_INDEX.csv files:

{
  "fileName": "2024-09-11 Project Manual SC Justice Center Remodel.pdf",
  "status": "ok",
  "pageCount": 380,
  "sectionsFound": 82,
  "sectionsValidated": 79,
  "submittals": 171,
  "pagesCharged": 0,
  "demo": true
}

Input

Field	What it does
pdfUrls	Public links to spec book PDFs.
pdfStore + pdfKeys	Or pick one of your Apify key-value stores holding the PDFs (read-only access, granted only to that store).
includeDivision01	Also mine Divisions 00–01 (off by default: those sections describe procedures, not product submittals). They are always listed in the section index.
includeWarranty	Add special, manufacturer and installer warranties as closeout submittals (on by default).
sectionNumberFormat	08 71 00, 087100, 08.71.00 or as printed. Match your Procore Specifications tool.
procoreSubmittalManager	Email written into the Procore CSV (Procore requires a Submittal Manager).
procoreSubmittalStatus	Optional Procore status. Leave it blank to get Procore's default (Draft).
maxPages, maxFileSizeMb	Safety limits.

Output

Dataset: one row per submittal (submittalNumber, sectionNumber, sectionTitle, title, submittalType, procoreSubmittalType, category Action/Informational/Closeout, paragraphRef, page, article, description), plus one recordType: "error" row for each book that failed.

Key-value store (one set per book, prefixed book1-<name>-):

• SECTION_INDEX.json / .csv: Section Number, Section Title, Division, Division Name, Start Page, End Page, Page Count, Validated, Submittal Count.
• SUBMITTAL_REGISTER.json / .csv: the full register.
• PROCORE_IMPORT.csv: Package Title, Package Spec Section Number, Package Spec Section Description, Package Number, Submittal Title, Submittal Spec Section Number, Submittal Spec Section Description, Submittal Number, Description, Submittal Manager, Submittal Status, Submittal Type, Location, Received Date, Issue Date, Submit By Date, Responsible Contractor Name. There is one package per spec section. Types are mapped to Procore's default types (Product Information, Shop Drawing, Sample, Product Manual, Plans, Document, Other), and the description ends with the paragraph and page reference.
• OUTPUT: a run summary with links to every file, warnings and the pages charged.

Example register row:

{"submittalNumber": "102113-01", "sectionNumber": "10 21 13", "sectionTitle": "Toilet Compartments",
 "title": "Product Data", "submittalType": "product data", "procoreSubmittalType": "Product Information",
 "category": "Action", "paragraphRef": "1.03.A", "page": 43, "article": "1.03 ACTION SUBMITTALS",
 "description": "For each type of product indicated. Include construction details, material descriptions, ..."}

Accuracy (measured)

We tested on four public project manuals: Story County (IA) Justice Center, 380 pages; UNLV, 514 pages; Erie County (NY) District Attorney lobby, 530 pages; and a 47-page school package.

• Section detection: 203 of 203 table-of-contents sections were found, with no false sections.
• Submittal items, against 39 hand-checked sections:
  • Development set (25 sections, used to write the rules): precision 100%, recall 99.2%, type accuracy 100%.
  • Held-out set (14 sections, labelled after the rules were frozen): precision 98.6%, recall 98.6%, type accuracy 100%.
• The rules prefer precision. A paragraph that states a requirement but names no deliverable (for example "Provide message list" with no label) is left out rather than guessed. Always review the register before you issue it.

Limits

• Text PDFs only. Scanned pages are not OCRed (they are reported in the warnings).
• Built for CSI 3-part sections (SECTION header … PART 1 GENERAL … END OF SECTION), in MasterFormat 2004+ (6-digit) or 1995 (5-digit) numbering. Sections without a clean header/end pair are listed in the index but not mined.
• If a spec has no list letters in its text layer, the letters are inferred from indentation; those rows are flagged paragraphRefInferred: true.

Pricing

Pay per event: $0.05 per spec-page event, one for each page of the validated CSI sections returned. You are charged only after the register has been saved, and only if it has at least one submittal. Failed downloads, unreadable or scanned PDFs, books with no recognisable sections or no submittals, and the prefilled demo book are free. If your max run cost is reached, the output stops at the last whole section that fits, and you pay only for those pages.

Errors

INVALID_URL, BLOCKED_URL, DNS_ERROR, DOWNLOAD_FAILED, DOWNLOAD_TIMEOUT, NOT_FOUND, HTTP_ERROR, ACCESS_DENIED, FILE_TOO_LARGE, EMPTY_FILE, UNSUPPORTED_FORMAT, CORRUPT_FILE, ENCRYPTED_PDF, NO_SECTIONS_FOUND, BUDGET_EXCEEDED. Warnings: PAGES_TRUNCATED, LOW_TEXT_YIELD, SECTIONS_NOT_VALIDATED, NO_SUBMITTALS_FOUND, BUDGET_PARTIAL_OUTPUT.

Privacy and security

The Actor runs with limited permissions. It can read only the key-value store you select, and it downloads only public http(s) URLs (private/internal addresses are blocked). Nothing is sent to third parties. "Procore" and "MasterFormat" are trademarks of their owners. This Actor is not affiliated with Procore Technologies or CSI.

Other Actors by Haniel

• Floor Plan Takeoff: PDF Room Areas, Wall LF & Door Counts: Construction takeoff from vector floor-plan PDFs: room square footage, wall linear feet and door counts, with a scale self-check. $2.50 per sheet.
• ACORD 25 COI Checker: Certificate of Insurance Compliance: Read ACORD 25 certificate of insurance PDFs into JSON and get PASS / FAIL / REVIEW against your limits, expiry and additional insured rules. $0.08 per certificate.
• SAM.gov Scraper: Federal Contract Opportunities Monitor: New SAM.gov contract opportunities, RFPs and sources sought by NAICS, set-aside and agency. No API key. $2 per 1,000 notices.
• PDF to Markdown Converter & PDF Text Extractor (OCR, DOCX): PDF, scanned PDF, DOCX, HTML and images to clean Markdown or text for RAG and AI agents. $3 per 1,000 documents.