The market

Somebody wants to pay you for your old Slack messages. Sounds too good to be true.

Keep reading. It is real. It also has catches. Here are both.

Why anyone would pay for this

AI companies trained their models on the public internet, and that well has run dry. What they need now is something the internet never had: how real work actually gets done inside a real company. The ticket, the thread about the ticket, the fix, the follow-up, and how the decision was made. That only exists inside businesses like yours.

The receipts

Google bid for the internal company data of the defunct Spirit Airlines. Two AI data companies bid against it.
$10 million from Google. $7.5 million and $12.5 million from the other two.
Buyers publicly advertise what they pay operating companies for operational data.
$100k to $2M+, in published tiers starting at $100k+, $500k+, and $1M+
A firm that winds down startups sold their leftover data to AI companies.
Nearly 100 deals in a year, typically $10,000 to $100,000 each
One 13-year-old company sold its archive.
Hundreds of thousands of dollars

Look at the first row again. Three buyers looked at the same data and came in at $7.5 million, $10 million, and $12.5 million. That spread is why you do not take the first offer.

Sources: Fast Company · Forbes, Aug 2026 · Forbes, Apr 2026 · Published buyer pricing, 2026

The catches

1. Not every company qualifies.

Buyers want size and history. A common public bar is 20 to 30 employees or more, with US, English-language companies first in line. Small or young companies may not get offers.

2. Privacy risk is reduced, not erased.

Buyers strip or replace personal details before the data is used. That technology is getting better, and it is not perfect. When Spirit's data went up for sale, the flight attendants' union objected that even scrubbed data could reveal sensitive details, and privacy experts made the same point. Some buyers have human QA in this process, but there are no formal guarantees of perfection. That is why you control what goes in.

3. You have to actually have the right to sell it.

Customer contracts, employee agreements, and your own privacy policy can limit what you can license. The FTC has warned that quietly changing your terms so you can share customer data for AI training can be unfair or deceptive. That is a question for your attorney, and we make sure it gets asked before anything moves.

4. The price is not guaranteed.

Published ranges are what buyers advertise. Your number depends on your size, your systems, how connected your data is, and how many buyers want it.

5. This market is new.

Terms differ wildly between buyers: one-time purchase or license, exclusive or not, how long they keep it, who they can resell it to. Two offers at the same price can be very different deals.

So what do you do with that?

Find out whether you qualify. Get your attorney's read on your rights. Make the buyers compete. That last part is our job.

Take the assessment