Operating records: what they are and how to license them

Operating records are the everyday digital logs your company generates, and you can license them to AI labs under a data-license agreement. The process starts with a free fit check and moves forward only with your approval.

Defining operating records

In plain terms, operating records are the digital footprints of work that happens inside your organization. Anything that captures a request, the actions taken, and the final outcome belongs here. Typical examples include

• Support tickets that show the problem, the steps tried, and the resolution.
• Chat threads where a decision is reached, such as sales negotiations or internal approvals.
• CRM entries that track leads, outreach, and closed deals.
• SOPs and internal documentation that describe how tasks are performed.
• Knowledge-base articles, QA records, and project-history logs.
• Accounting exports that detail invoices, purchase orders, and expense reports.

These records are sometimes called "business records" or "operating records" in legal language, but they refer to the same everyday data. An email archive alone is usually less valuable because it often lacks the clear chain from request to outcome that AI labs look for.

Why AI labs want this data

AI labs train models that need to understand how real work gets done. When a model sees a full workflow, say, a support ticket that moves from opening to resolution, it learns the reasoning steps, the language used, and the decisions made. That context lets the model generate more accurate suggestions, automate routine tasks, and build reinforcement-learning environments that mimic real business processes.

Not sure your company is a fit? Take the free check

Can you license your operating records?

Yes. Companies with at least 30 full-time employees are eligible, and there is no upper limit on size. Larger organizations usually generate more data, which can raise the price of the license but also makes the data more attractive to buyers. The license you grant covers only the data; you retain ownership and continue to use the records internally.

Typical financial terms

Completed deals generally fall between $100,000 and $2 million USD, with many landing around $200,000 to $300,000 USD. The exact amount depends on several factors, such as the volume of data, the richness of the records, the length of continuous history on the same tools, visible reasoning in the workflow, rarity of the industry niche, and the cleanliness of rights. Structured formats, useful metadata, multi-language content, and compliance with industry standards can also increase value.

How the licensing process works

You can read the full step-by-step guide on our site (see how a deal runs). In short, the process is

1. Fit check. You answer a few basic questions on our free check page (who qualifies); no data is shared at this stage.
2. Quote gathering. We turn your answers into a one-page summary and share it with several AI labs.
3. Offer review. You receive the best quote and its conditions, then decide whether to proceed.

If you move forward, the data is exported securely, de-identified, and any personally identifiable information or confidential markings are removed. Names become stable pseudonyms like "employee 632." The de-identification method and file handling are spelled out in the contract, and you review a sample before any data leaves your environment. Legal counsel confirms that you own the data and that your customer contracts allow de-identified use.

The buyer licenses the data, not a trained model, and use rights are usually perpetual because a model cannot reproduce the original records. Exclusivity is limited, typically to 24 months; a shorter term can lower the price.

Sample operating record

Below is a simplified illustration of a support ticket that would be attractive to an AI lab:

Field	Example
Ticket ID	45231
Date opened	2023-07-15
Issue description	Customer reports error code E-502 when exporting reports.
Steps taken	(1) Reproduced error on test system. (2) Checked server logs, found timeout. (3) Applied patch v3.2.1.
Resolution	Patch resolved error; customer confirmed successful export.
Time to close	4 hours
Agent pseudonym	employee 632

The chain from problem description to resolution gives a clear decision path that AI labs can use to train models on troubleshooting, communication style, and problem-solving logic.

Getting started with licensing your data

If you think your company's operating records could be valuable, begin with the free fit check. The button below will guide you through a brief questionnaire and let us see if you qualify. After you submit, we will handle outreach to AI labs, bring back the best offer, and walk you through the contract details.