By system
Include one of these or several at once. You decide for every system.
Microsoft 365 or Google Workspace inboxes holding years of customer, vendor and team threads.
Files and shared drives
The proposals, how-to procedures, templates and reports stored on SharePoint, Dropbox or Google Drive.
Team chat
Slack or Teams threads and channels where daily problems get sorted out.
CRM
Salesforce, HubSpot or similar tools, with call notes, pipeline stages and account history.
Service tickets
The requests, answers and fixes logged in your service inbox or help desk.
Project management
Boards, tasks and status updates in tools such as Jira, Asana or Monday.
Finance systems and ERP
Purchasing, order, inventory and invoicing workflows in your accounting system or ERP.
Code
Repositories, pull requests and review comments, if your team builds software.
What makes ordinary records valuable
To train and test models on real work, AI labs license business records. If a model has read only the public web, it knows very little about how a company runs inside.
What your records hold is a chain of events. A request arrives, someone asks something, a file is revised, a call is made and the job is closed. That chain is what has value. It can be messy and unorganized.
Having years of records helps. A sizable team helps too, with many people writing, replying and deciding in the same tools.
Examples of a useful thread
- A customer asks for a quote. A team member checks stock, consults a coworker, writes up a price and follows up a week on.
- A job runs late. The team trades messages, updates the schedule and tells the customer what changed.
- A ticket is opened. The support team asks questions, tries a fix, passes it up and closes it out with a note.
- A process gets written down. The draft is reviewed, fixed and saved as the way things are now done.
The minimum to qualify
- 20 or more employees. Enough people working in the same systems.
- 3 or more years in business. Enough time for the records to show work changing.
- Records in business software. For example Slack, Google Workspace, Microsoft 365, an ERP, a CRM or project tools.
What is not required
You do not need clean data, a data team or a single system of record. Nor do you need to decide now which systems to include. Those choices come later, during scoping, and every one of them is yours.
Records you would rather keep private can simply stay out. Sensitive data that remains in scope is excluded or de-identified before anything is shared.
See what your records might be worth.
Show me my estimate