30-min call with RobBook a 30-minute call with Rob
Owners weighing an offer

// Field notes / Data licensing

Selling your company's data to AI companies: an owner's guide

By Rob Eason, founderPublished Updated

// Who buys

Who is buying company records, and what for

A new kind of buyer is writing to business owners. AI companies, and the data firms that supply them, want the records your company produced while it worked: the emails, chat threads, documents, tickets and CRM history that show how real work gets done.

Three programs publish their terms. Handshake AI says it buys operational data from businesses across every major industry. micro1 describes its top tier as highly unique, proprietary operational data with significant value for frontier AI development. Scale AI says the buyers behind its program are interested in modeling the personas within your organization.

It started with companies that had closed. The law firm Frankfurt Kurnit Klein + Selz wrote in October 2026 that startups shutting down have been selling Slack archives, email and Jira tickets to AI developers. The same article turns to operating companies, which have current staff, customers and partners to think about.

// What they want

The records buyers ask for

Buyers want records that show work happening over time, in tools your team already uses. The published lists overlap more than they differ.

History matters. micro1 says companies with several years of records across multiple business systems are generally a stronger fit, and that the data should be mainly in English. Scale AI asks for years of records in the tools your teams already run on.

  • Email and calendars. Scale AI names Gmail, Outlook and Google Calendar. Handshake AI lists Gmail and Outlook among its example sources.
  • Chat. Slack, Microsoft Teams and Google Chat on Scale AI's list, and Slack on Handshake AI's.
  • Documents and files. Google Drive, Dropbox and Notion appear on Handshake AI's list, and Scale AI asks for documents and files in general.
  • Work tools and tickets. Jira, Confluence, Asana and Linear for Scale AI. Support tickets, project histories and SOPs for micro1.
  • Business systems. CRM, HR, ERP and support records from systems such as Salesforce, HubSpot, Workday and Zendesk for Scale AI. NetSuite, QuickBooks, ServiceTitan and Paychex on Handshake AI's list.
  • Whole departments. micro1 groups what it wants by function: software and engineering, finance and accounting, customer support, sales and CRM, legal and contracts, and operations and logistics.

// What they pay

Published price ranges, and what moves an offer

Across the three programs, the published ranges run from five figures to over $1 million per deal. Each is a range on a marketing page, not a quote for your records.

The descriptions show what moves the number: volume, how many teams and years the records cover, whether you keep supplying data over time, and how unusual the records are. micro1 ties its middle tier to large-scale datasets or ongoing participation across teams, and Scale AI says value scales with cadence.

Read when the money arrives, too. micro1, the one program of the three that publishes its steps, runs discovery and pricing, then contracting, then delivery and acceptance, then payment. It says it has 30 days to review data after you transfer it, and pays in full within 30 days of acceptance.

Here is how each program states its range, as of November 2026:

  • Scale AI: $10K to $1M+, which it calls the illustrative value per data partnership, scaling with cadence.
  • micro1: $100K to $500K for qualified enterprise data partnerships, $500K to $3M for large-scale datasets or ongoing participation across teams, and $3M+ for highly unique, proprietary data.
  • Handshake AI: $100K to $4M+ in payouts.

// Smaller companies

If you have fewer than 20 people

The programs set floors. Handshake AI asks for 20 or more full-time employees and 3 or more years in operation. micro1 asks for 30 or more full-time employees, mature operations and substantial historical data. Scale AI calls 40 or more employees its preferred starting point, though it says it assesses every partnership case by case.

So a 10-person firm sits below every published line. That doesn't make its records worthless. It does mean the published ranges weren't written for it, and an offer may be smaller, or may not come.

Location matters as well. micro1 says it's prioritizing US companies, then other Western markets, with demand strongest in the United States, the United Kingdom and Canada.

If you're small and an offer arrives out of the blue, ask three things before you share anything: who the end buyer is, how the price was set, and what happens to the records after delivery.

// What can stop a sale

Holding the records isn't the same as the right to sell them

The first legal question is whether you may license the records at all. Frankfurt Kurnit's article puts it plainly: a company that holds data doesn't necessarily have the right to sell it. Email and chat contain other companies' confidential information, often covered by NDAs and commercial agreements.

Your old privacy promises still bind you. The same article says companies must comply with the privacy statements in effect when the data was collected, and are bound by those commitments unless they get consent. The FTC warned in February 2024 that starting to use data for AI training, and telling people only through a quiet, retroactive change to your terms or privacy policy, may be unfair or deceptive.

State privacy law can apply too. Frankfurt Kurnit reads a transfer of workplace data for money as a sale of personal information under California's privacy law, and the California Attorney General's office lists the right to opt out of the sale or sharing of personal information among the rights that law gives Californians.

Then there are the checks that are easy to miss:

  • Regulated records. Health, financial and children's information can fall under HIPAA, GLBA and children's privacy laws, the article notes.
  • Platform terms. It points out that Slack's API terms prohibit third-party app providers from using API data to train a large language model. Read the terms of every tool you'd export from.
  • Employees. Staff objecting together may be protected concerted activity under Section 7 of the National Labor Relations Act, and a union contract may restrict the sale.
  • Customer and partner contracts. Look for confidentiality and data-use clauses in your biggest agreements before you list what you'd share.

// The steps

From first offer to payment, in order

Treat a data license like any contract that carries risk. These steps keep you in control of what leaves and on what terms. None of this is legal advice, and every step ends with your lawyer.

  • List the systems. Write down every tool that holds records, how many years each covers, and whose name the account is in.
  • Read what you promised. Pull every version of your privacy policy and customer terms, and every NDA or contract with a confidentiality clause.
  • Decide what stays out. Health and financial details, other firms' confidential material, HR files and anything under a no-sharing clause are the usual candidates.
  • Ask where personal details come out. Scale AI says it anonymizes everything inside your environment before anything is extracted. micro1 says personal information is removed or transformed before your data goes to third parties. Know which happens, and who checks it.
  • Approve each package. Scale AI offers your sign-off on every data package. Ask for the same wherever you go.
  • Negotiate the terms. Frankfurt Kurnit says the agreement should restrict how the buyer can use the data, allocate privacy risk, and say whether the data must be returned or deleted. Ask too about resale, exclusivity, how long the license runs, and when you're paid.
  • Have your lawyer read it before you sign. A privacy lawyer for the records, and your M&A lawyer as well if a sale of the business is anywhere in view.

// Staff, customers, a sale

Your staff, your customers, and a later sale

Frankfurt Kurnit's warning for operating companies is about goodwill: your employees, customers and partners may be upset. Their messages are part of what you'd be licensing, so tell staff before anything is exported, and decide what you'll say if a customer asks.

A license also outlives the moment you sign it. If you sell the business later, the buyer will read it in due diligence, along with the privacy promises it relied on. Our note on licensing before a sale covers what they'll ask.

And the same records have a use at home. The email, tickets and CRM history a buyer wants are what an AI system inside your own company would learn from too. If you license, check that the terms leave you free to use your own records that way.

// Questions

Before you answer the offer

How much do AI companies pay for company data?
Three programs that publish ranges quote from five figures to over $1 million per deal: Scale AI $10K to $1M+, micro1 $100K to over $3M, and Handshake AI $100K to $4M+. Volume, years of history, ongoing supply and how unusual the records are move the number. A published range is not a quote.
Is it legal to sell company emails for AI training?
Sometimes, and only after checking. Emails carry other companies' confidential information and the personal details of staff and customers, and you're bound by the privacy promises in effect when they were collected. State privacy laws, platform terms and union contracts can apply too. Ask a privacy lawyer before you agree to anything.
Can a small business sell its data to AI companies?
The programs that publish floors ask for 20, 30 or 40 or more employees, so most firms under 20 people fall below their lines. Scale AI says it assesses each partnership case by case. Expect smaller offers, or none, and check any offer that arrives unasked.
Do we keep ownership of our data if we license it?
That depends on the license. micro1 says companies retain ownership of their underlying data, and Scale AI asks for your approval of each data package. Read who may resell the records, whether the license is exclusive, how long it runs, and when raw data is deleted.
Rob Eason

Rob Eason, founder

// Book a call with Rob

Weighing an offer for your company's records?

A 30-minute call with Rob. Bring the offer, or just the question, and he'll walk through which of your systems hold what, what to ask the buyer, what to take to your lawyer, and what the same records could do inside your own company.

  • Co-owned a 60-employee financial services firm that served more than 7,000 customers and has a 5-star rating across more than 600 customer reviews, and sold his share to his partner.
  • Ran GOSO Technologies, a Boston solar company that Forbes Home named one of the top 10 solar companies in Boston, rated 5/5 across 24 Google reviews.
  • Spoke on the construction panel at Build Connect 2026 in Charlotte, sponsored by Choate Construction.
  • HighLevel Certified Admin with the second-most reviews in HighLevel's Certified Admin Directory worldwide: 5/5 across 83 reviews.
  • Won HighLevel's Top New Subaccounts via AI Employee award in its 2025 Summer of AI Awards.
  • 4.3/5 from 6 Google reviews of Massmarket AI.
  • Every directory and Google review, critical ones included, is shown word for word at massmarket.ai/reviews.
  • Takes every proof-of-concept call himself.