Company Data: The Overlooked Asset in a Business Wind-Down

Before a company sells or closes, its records may qualify for an AI data license. Advisers who make the introduction can earn a referral fee, on clear conditions.

Every wind-down has a spreadsheet. Receivables are chased, leases surrendered, equipment listed for sale, subscriptions cancelled one by one. The CFO knows the date the last payroll runs. The accountant knows which filings remain.

What rarely makes that spreadsheet is the archive. Ten years of support conversations, project records and CRM history sit in systems that are about to be switched off, and the people who understand them are working their notice. For a growing number of companies, that archive may deserve a licensing conversation before the lights go out.

AI developers have already used much of the openly available text on the internet. To keep improving, they need real-world data that shows how businesses operate: how problems get solved, how decisions get made, how work actually gets done. Most of that material is not public. It lives inside companies, and many developers prefer to license it with the owner’s agreement.

SourceX licenses that kind of business data directly from companies for AI training and evaluation, and pays them for it.[1] The company signs with SourceX, not with an AI lab, and SourceX manages the agreement, data preparation and payout. For finance teams, the important detail is that a business does not have to be trading to take part. Acquired and wound-down businesses can be considered alongside active ones, provided they meet the profile and can show they own the records.

The headline offer is up to $4M for a licensed dataset. That is an upper bound for the largest enterprise datasets, not a typical or guaranteed outcome, and it belongs in a deal model only as a contingent scenario.

The opportunity tends to vanish for boring reasons. A system administrator leaves. A software contract lapses and the export becomes unreadable. A buyer acquires certain assets while the seller keeps others, and suddenly nobody is sure which entity can authorize a new use of the records.

For illustration, take a services firm with 120 staff at its peak, preparing to close. Its support history links customer questions to resolutions across eight years. With a week of attention from the right people, the company could describe that collection, confirm who owns it and decide whether to explore a license. Three months later, the knowledgeable staff are gone and the same exercise costs far more.

None of this changes existing duties. Retention schedules, deletion obligations and confidentiality commitments still apply. The point is to make a decision while a decision can still be made, and to record who made it.

Before access disappears. Concept illustration.

CFOs, fractional finance leads, accountants and M&A advisers are often the first to see that a company is approaching a transition. That makes them well placed to raise the question, and SourceX’s referral program is built for exactly this kind of introduction.[2] Its own pitch to partners is short: one introduction is all we need.

The adviser does not value the archive, collect samples or act as the data owner. The job is to connect a willing decision maker with SourceX and let the company describe its own systems. The fit check is a few quick questions, with no uploads and no commitment, so the introduction asks very little of the client.

A useful first note covers the entity name, its peak headcount, the main systems and the approximate years of data, plus the person who can speak to rights. Customer ledgers and email exports stay out of it entirely.

Under the standard published terms, an approved partner earns 25% of the actual supplier payout, up to $100,000 cumulatively per company.[3] The actual supplier payout means the compensation SourceX actually pays the referred company, excluding separately itemized taxes and expense reimbursements. SourceX pays the reward separately, so it never reduces what the client receives.

The arithmetic is straightforward. If a referred company’s actual payout were $200,000, the standard reward would be $50,000. If the payout were $500,000, the reward would reach the $100,000 cap. These are calculations, not forecasts.

Before making any introduction, the partner needs an approved account and a signed Channel Partner Agreement. Attribution runs on a 90-day window, and only one partner is credited per company. An existing application or relationship with SourceX can block credit, so check before mentioning the program to a client.

A submitted lead, a meeting or even a signed supplier agreement does not trigger payment on its own. The reward becomes payable after SourceX has paid the referred company and the remaining conditions are met: acceptance and payment by the acquisition partner, finance approval, and the partner’s tax and payment details. The terms set a deadline of 60 calendar days after the later of final acceptance and completed handoff. Signed exceptions, set out in a Schedule A or addendum, take precedence over the standard terms.

Accountants and other licensed professionals should also check their own rules on referral fees and disclose any commercial interest where required. The client should always know why they are being introduced.

The practical change is small. Alongside receivables, contracts and equipment, add a line for company records: who owns them, when access changes and whether a no-upload fit check should happen before that date. The answer may be a documented decision not to proceed. That still has value, because the question was asked while someone could answer it.

If you advise a company that fits, review the SourceX referral program, get your partner account approved and sign the agreement before you make the introduction. The best time to ask about the data is before the people who understand it walk out of the door.

About the founder

Noah Loul is a four-time founder and the founder and CEO of SourceX, which licenses business data directly from companies for AI training and evaluation. He previously founded B2B Rocket, acquired by Blackpearl Group. His work focuses on building permission-based data infrastructure for AI.