Home / Data types / Business operations
Business operations data for AI agents
Datadhara licenses the internal records of how real companies worked, such as email, chat, support tickets, CRM, docs, SOPs, project histories, financials and invoices, from companies that own them, including startups that have shut down. People and customers are pseudonymised, and the data is licensed for AI training under a signed agreement.
What we can source
| Record type | Format | Typical labels |
|---|---|---|
| Email and chat | Mailbox and Slack/Teams exports | Threads, participants (pseudonymised), timestamps |
| Support tickets | Zendesk, Intercom, Jira exports | Status history, resolution, linked conversations |
| CRM and sales | Salesforce, HubSpot exports | Accounts (masked), deal stages, activities |
| Docs, wikis and SOPs | Notion, Confluence, Google Docs | Version history, owners, links |
| Financials and invoices | Accounting exports, PDFs | Ledgers, invoices, budgets, reports |
We collect to your spec, so volumes, languages and labels are agreed per project. Tell us what you need.
Example record
{
"source": "support_ticket",
"opened": "2024-03",
"customer": "acct_7f2c",
"messages": 9,
"resolution": "refund_issued",
"linked": ["crm_deal_311", "slack_thread_88"],
"consent_ref": "src-0452 / v1"
}
Values are illustrative. Real deliveries follow the schema agreed in your spec.
How personal information is removed
Employee and customer names, emails, phone numbers and account details are replaced with stable pseudonyms, so threads and accounts can still be followed across systems. Secrets, payment card data and special-category personal data are removed, and contracts that restrict sharing customer data are checked before anything is licensed. See the full methodology.
What teams use it for
- Training and evaluating enterprise agents that work across email, chat, tickets and CRM
- Realistic tool-use and workflow environments
- Summarisation, search and question answering over company knowledge
- Evaluation sets for customer-support and sales assistants
How licensing works
Every source signs a written agreement that covers AI training. You start with a small pilot and a dataset card, and if the data works for you we agree volume, timeline and license terms (exclusive or non-exclusive) in writing. Read the full licensing walkthrough.
Samples
These records are private by nature, so samples are shared under NDA once an owner has agreed to license. Tell us your spec and we'll confirm what we can source.
Have data like this?
Did your company keep years of email, tickets and CRM records? See how licensing to AI labs works. You keep ownership, and you don't pay us.
Further reading
- Selling a shut-down startup's code and data to AI labs
- Private codebases with real git history
- License your company's records to AI labs
Questions
Why does data from several systems matter?
Real work spans tools: a ticket links to a Slack thread, a CRM deal and an invoice. Records that keep those links let agents learn how work actually moves through a company.
Can companies that have shut down license their records?
Yes, when the founders, a successor company or an administrator controls the records. Customer and employee data is pseudonymised or removed first.
How far back do records go?
It varies by company. Two or more years across several systems is the most useful range.
Tell us what your models are missing
A few lines is enough: the domain, the format, and roughly how much.
Request data →