Getting started
Licensing company data for AI training: an owner’s guide
By Automati · Updated
Company data licensing starts with a question: does your retained operating history contain useful work that your company may have the right to license? Start with a description of the records, then establish buyer fit, rights and terms before sharing anything.
What would a data license actually cover?
A proposed license should describe the selected records and the uses a buyer is allowed to make of them. AI training and evaluation are different uses worth asking about explicitly. A discussion about licensing is not the same as selling your business, transferring every file or granting ongoing access to your systems.
The exact rights come from the agreement. Ask which company will receive the records, whether onward licensing is allowed, what copies can be retained and how long the arrangement lasts. Do not infer these terms from a marketing page or an introductory conversation.
Look for the decisions around the work
A finished quote tells only part of a story. Earlier versions, the requirements behind them and the outcome may explain how a team handled a real task. These connected records are a useful starting point for an inventory; they are not a promise that a buyer will want them.
| Work | History to investigate | Question to resolve |
|---|---|---|
| An order | Request, estimate revisions, planning and delivery exception | Can the steps be connected without exposing restricted material? |
| A support issue | Problem description, investigation and resolution | Who controls the ticket contents and related correspondence? |
| An internal procedure | Versions, reasons for changes and training notes | Is the material company-owned or supplied by someone else? |
Buyer requirements are specific to the opportunity
Public buyer programs describe different needs. Treat their pages as evidence about that particular program, at that point in time. Industry, languages, retained history, record quality and available rights all belong in a conversation; no single employee count establishes universal eligibility.
Company age and accessible record history are separate facts. A company founded twenty years ago may have only two years of searchable records after a migration. Describe what can actually be accessed, and label estimates or unknowns.
A practical sequence for the first conversation
Prepare a high-level inventory without attaching files. Agree who can speak for the company and who needs to answer questions about the systems and restrictions. An initial description can be enough to decide whether further review is worthwhile.
- Identify one workflow and the systems that retain it.
- Estimate accessible years, languages and volume; keep unknowns visible.
- Flag customer-owned material, personal information and contractual restrictions for separate review.
- Ask for the buyer’s current requirements and proposed next step in writing.
- Agree a separate process for any sample, export or access request.
Where Automati fits
Automati is an independent referrer. We help describe the opportunity and prepare a possible buyer introduction. We do not buy the records, set a price or guarantee an offer, and we may receive a referral fee if a deal completes.
The initial review asks for contact details and descriptions only. You decide whether to permit an introduction. The buyer evaluates fit and proposes any license; your company reviews the offer and data-handling terms before deciding.
Sources and context
Public pages checked October 16, 2026. Programs can change; confirm current terms with the provider.
- micro1: data partnerships
A buyer’s public description of its own operational-data program; not an Automati partnership or universal eligibility rule.
- Supersoniq: operational data licensing
An intermediary’s explanation of its own process. Its service commitments and pricing claims do not describe Automati.
Automati is an independent referrer and may receive a referral fee if a deal completes. Buyers decide fit and any offer. This guide helps prepare questions; it does not determine rights, eligibility or a dataset’s price.