AI buyers license the records a company produces while it works: email, team chat, shared documents, support and project tickets, CRM history, accounting records and source code. What makes them valuable is history and real workflow: years of records showing how a real team handled real customers, projects and decisions. Buyers generally want US companies with 20 to 30 or more employees and several years of records in business systems.
Records buyers ask for
| System | Examples | Buyers that name it |
|---|---|---|
| Email and calendar | Google Workspace, Microsoft 365 | Mercor, Troveo; micro1 says “internal communications” |
| Team chat | Slack, Microsoft Teams, Google Chat | Mercor, Troveo; micro1 says “internal communications” |
| Documents and drives | Google Drive, SharePoint, OneDrive, Notion, Confluence | micro1, Mercor, Troveo |
| Tickets and projects | Zendesk, Jira, Asana, Linear | micro1, Nyne, Troveo |
| CRM and sales | Salesforce, HubSpot | micro1, Mercor, Nyne, Troveo |
| Accounting and ERP | QuickBooks, Xero, NetSuite | micro1, Nyne, Troveo |
| Code | GitHub, GitLab | Mercor, Troveo; micro1 names engineering workflows and code documentation |
micro1 describes the target as “operational data”: SOPs, project histories, internal knowledge bases, support workflows and reconciliations. Mercor lists industry examples such as contracts and case files in legal, clinical notes and claims in healthcare, work orders in manufacturing and dispatch logs in logistics. Nyne focuses on the systems that run day-to-day work in service businesses: invoicing, CRM, support tickets and scheduling.
What makes records more valuable
- Years, not months. micro1 says several years of records across several systems make a stronger fit; Nyne asks for three years in business.
- Several connected systems. A support ticket that links to a CRM record, an email thread and a document shows a whole piece of work. Mercor says connecting more tools generally means a higher payout.
- Real work, not finished files. Drafts, discussions, handoffs and corrections matter more than polished outputs.
- A clear right to license. The company must own the records or have the right to license them. Records you hold for clients often fall outside that right.
What is usually excluded
Some records cannot be licensed even after de-identification, or need extra steps first:
- Client files you hold for someone else. A law firm’s privileged matter files, an accounting firm’s client books or an agency’s client assets usually belong to, or are protected for, the client.
- Health records covered by HIPAA, such as patient charts.
- Payment card data and full bank account numbers.
- Data you promised customers not to share in a privacy policy or contract.
- Records under a legal hold, for example in pending litigation.
Our guide to legal considerations covers each of these. Buyers also let you exclude whole categories: Nyne, for example, says you can leave out categories entirely.
Company size and status
| Buyer | Published size rule | Operating or closed |
|---|---|---|
| micro1 | 30+ full-time employees | Both, through separate programs |
| Nyne | 20+ full-time US W-2 employees, 3+ years in business | Operating only |
| Mercor | Not stated | Not stated |
| Troveo | Not stated | Not stated |
A company that has closed can still qualify, as long as its records still exist. That is the catch: cloud accounts are deleted soon after they are cancelled. See what to do with company email and Slack when you close.
Questions people ask
Do we need a minimum amount of data?
micro1 says it has no fixed minimum number of files or terabytes. What buyers look for is history: several years of records across several business systems.
Can a small company qualify?
The lowest published bar among the buyers we cover is 20 full-time US W-2 employees and three years in business (Nyne). micro1 asks for 30 or more full-time employees.
Does the data need to be in English?
micro1 says data should be primarily in English. Other buyers do not state a language rule, but all of them target US companies.
Sources
More on how it works
Privacy and de-identification: who removes personal data
How AI data buyers remove personal information from company records before licensing, what you control, and what to put in the contract.
Legal considerations before you license company data
Employee and customer data, privileged and health records, privacy policies and bankruptcy rules: what to check with a lawyer before licensing records.
Timeline and payment: how long licensing takes
How long it takes to license company data, when buyers pay, what payouts they publish, and what slows a deal down.