Commercial general contractor project records and mail, 16 firms

Summary

Project management records and project mail from 16 US commercial general contractors, covering about 1,900 jobs from preconstruction to closeout. RFIs, submittals, change events, daily logs and meeting minutes stay linked to their job, so a model can follow one project across years of threads and decisions. People, firms, sites, phone numbers and bank details are replaced with consistent placeholders, and HR, payroll and legal mailboxes are never collected.

Collected to order. We collect it from US companies for your order. Scope, volume and timing are confirmed in writing before you sign.

Pricing on request

Request scope and pricing

At a glance

Planned scope. Volume, sources and timing are confirmed in writing before you sign.

Records: 19.5Memails, project records and documents
Size: 58.7 GB
Tokens, estimated: 6.6B
Years: 2013 to 2026
Firms: 16

Owner profile: 16 US commercial general contractors, 15 to 110 staff each, Southeast, Texas and the Mountain West
Industry: Construction
Data type: Company operations archive
Headcount: 15 to 110 staff per firm
Region: United States (Southeast, Texas, Mountain West)
Languages: English, Spanish
Formats: JSONL
Updated: 2026-10-14

Sources

System	Kind	Records
Microsoft 365 mail	Email	18.3M messages
Procore RFIs, submittals and change events	Projects and wiki	475K records
Procore daily logs	Field operations	512K daily logs
Shared drives (minutes, schedule narratives, closeout)	Shared files	212K documents

Good for

evaluating long-context reasoning over multi-year project threads
training agents to draft RFI responses and route submittals to the right reviewer
predicting which RFIs become change events from the question text
summarizing daily logs into weekly owner progress reports

Fields

Field	Type	Description
record_id	string	Unique record id within the pool
project_id	string	Placeholder for the job, stable across every record of that job
firm_code	string	Placeholder for the contributing contractor
record_type	enum	email, rfi, submittal, change_event, daily_log or document
created_at	timestamp	Creation time with the original UTC offset
subject	string	Subject line or record title
body	string	De-identified text, quoted replies stripped for email
spec_section	string	CSI MasterFormat section where the record cites one
status	enum	open, answered, closed or void
thread	array<object>	Responses in order, each with role, time and text

Example record

{
  "record_id": "GC16-SUB-0207714",
  "project_id": "[PROJECT_311]",
  "firm_code": "[COMPANY_7]",
  "record_type": "submittal",
  "created_at": "2023-02-02T10:18:00-06:00",
  "subject": "Submittal 08 11 13-02 Hollow metal doors and frames, Rev 1",
  "body": "Resubmitting the hollow metal shop drawings with the Rev 0 comments addressed. Stair and corridor frames are now 16 ga with labels for 90 and 20 minute ratings. Door 214B hand corrected to LHR. Fabricator needs approval by 2/16 to hold the 6 week lead time.",
  "spec_section": "08 11 13",
  "status": "closed",
  "thread": [
    {
      "role": "architect",
      "at": "2023-02-09T16:40:00-06:00",
      "text": "Make corrections noted. Door 118 frame needs a 4 in head to fit the closer arm, not 2 in. Hardware sets to match the revised hardware submittal 08 71 00-03."
    },
    {
      "role": "general_contractor",
      "at": "2023-02-10T08:12:00-06:00",
      "text": "Corrections sent to the fabricator. Released for fabrication 2/14, delivery to site week of 3/27."
    }
  ]
}

One record, de-identified. Placeholders such as [PERSON_12] and [COMPANY_4] stay the same across the whole dataset, so threads still read correctly.

Licenses available

Exclusive
One buyer. Nobody else can license it during the exclusive term, usually 2 years.
Non-exclusive
Licensed to more than one buyer.
First look
One buyer gets 2 to 4 weeks to make an offer before anyone else sees the dataset.

What ships with it

Datasheet: what it is, what it is good for, verified size, sources, years, format and limits.
Provenance file: the owner shown by code, staff notice and consent status, exclusions, mapped to the EU AI Act Article 53 training-data summary and California AB 2013.
Manifest with a hash for every file.
QC report and de-identification report.
Your own copy, with unique canary records.

How it is de-identified

Default exclusions removed, files flattened to text, identifiers found in three layers and replaced with consistent placeholders, and the result measured against a 99% bar for each identifier type in each source.

The full method