Employees × years × data types → indicative range · 9 deals reported in 2026
AI labs now license real operating records (emails, tickets, ledgers) to train agents. Set your headcount, years of records and data types to see an indicative range, next to the deals reported this year.
A rough range based on deals data-licensing firms have published in 2026. It is not an offer: what a buyer pays depends on demand for your sector, how well the records can be anonymised, and whether you hold the rights to license them.
The same records are also what an AI rollout runs on. The audit shows what they can automate inside your own business.
Employees × years of records × $680 × a weight for the records you keep (email & chat 1.0, projects & tickets 0.7, accounting & ledgers 0.8; 0.5 if none are selected). The range runs from 0.6× to 1.28× that figure.
The $680 rate is the anchor used by public estimators from data-licensing brokers, and it lines up with the deals they publish below. It is an assumption, not a market price. For a lower bound: SimpleClosure, which winds down startups, reported about 100 such deals in a year at $10,000 to $100,000 each.
As published by data-licensing firms and the financial press; the businesses are anonymous. Elements was not party to these transactions.
Communications, product knowledge, order history, supplier records, and operational documents.
Rate quotes, shipment documents, dispatch records, delivery exceptions, and customer claims.
Dispatch messages, SOPs, logistics workflows, and customer service operating records.
CAD drawings, calculations, technical reports, revisions, and project specifications.
Operational work orders, job notes, and customer project records, anonymized before delivery.
Five years of workpapers, reconciliations, financial models, and tax-preparation files.
Operational bids, subcontractor agreements, change orders, and site logs.
Project installations, engineering checklists, permits, and customer service logs.
Thirteen years of Slack, Jira, email, and Google Drive records.
Use employees now or at your peak, whichever is higher. More people means more emails, tickets and decisions on record, and that volume is what buyers pay for.
Count only the years you still have archived and could export. Five years of reconciliations or dispatch logs is worth more than one, because it shows how the work changed over time.
Email and team chat weigh the most: they hold the reasoning behind decisions. Projects and tickets, then accounting and ledgers, follow. Untick anything you couldn't hand over in a clean export.
The figure uses the same model data-licensing brokers publish, and its rate is an assumption. A real offer depends on buyer demand for your sector, how well the records anonymise, and whether you hold the rights to license them. The deals below the estimator show what has actually been reported.
A free tool gives you a hypothesis. The 30-minute diagnostic is where we pressure-test it against your actual workflows — and decide whether the project is worth building, buying, or skipping.