Sell your company email archive to AI

What AI buyers want from email

Email captures work that crosses team and company lines: vendor negotiations, approvals, customer escalations and project hand-offs. AI labs building agents want these multi-step exchanges. enterprise buyers, for example, list project histories and decision-making records among the data they seek.

• Threads with a clear request, back-and-forth and outcome
• Approval chains and sign-offs
• Attachments that show work product, such as quotes and reports

Newsletters, notifications and spam have little value. Filter them out first. Email also links to other systems through attachments, calendar invites and references to tickets or deals. Those links raise "data richness," which Sunset uses in pricing (Forbes).

What email archives have sold for

SimpleClosure launched a WorkspaceData product and reports AI buyers up 9x versus 2025 (01net). It projects returning $10M+ to startups by the end of 2026. That is a projection.

Lessons from the Spirit sale

The Spirit case shows what regulators and courts may expect from a large email sale:

• Professional de-identification. The privacy ombudsman recommended approval only with conditions, including Tonic.ai de-identification of email and a certification filed before each transfer (The Next Web).
• No re-identification promise. A public commitment not to re-identify, flowing down to every later recipient.
• Employee objections. Unions objected over re-identification risk, and 120+ members of Congress urged excluding employee data (Senate Commerce Committee).
• Consumer data excluded. Customer and loyalty data was left out of the asset schedule.

Risks specific to email

• Privileged email. Messages with lawyers may be privileged. Frankfurt Kurnit recommends excluding privileged, HR, health and union data entirely (Frankfurt Kurnit).
• Outside parties. Email includes customers, vendors and job applicants. Under the CCPA, "consumer" includes employees and B2B contacts, and a transfer for money may be a "sale" unless the data is de-identified (Cal. Civ. Code 1798.140).
• Risk assessments. California's CPPA rules require a risk assessment before selling or sharing personal information, with compliance from Jan 1, 2026 (White & Case).
• Client confidentiality. Law firms face ABA Model Rule 1.6, and ABA Formal Opinion 512 requires informed client consent before client information goes into self-learning AI tools (ABA).
• EU contacts. GDPR purpose limitation may apply to emails with people in Europe (Hunton).

This is general information, not legal advice. See is selling company data legal?

How to prepare an email archive

1. Pick the mailboxes and date ranges in scope. Leave out HR, legal, executive-compensation and health mailboxes.
2. Filter out automated mail, newsletters and personal messages.
3. Run de-identification on names, addresses, phone numbers, signatures and account numbers, in both bodies and attachments.
4. Check the output. HHS notes, for health data, that de-identification risk "is not zero" (HHS).
5. Document chain of title. SimpleClosure runs multi-pass de-identification and chain-of-title checks before sale (01net).

Operating vs. shutting-down companies

Email sales so far have come from companies that closed or entered bankruptcy, such as Spirit and the startups SimpleClosure serves. An operating business can license a historical slice, but it should check its own privacy notices first. The FTC has warned that quietly changing terms to allow AI training could be unfair or deceptive (FTC). Closing down? Use the startup shutdown data checklist.

See what your archive may be worth: take the 60-second estimate.

Frequently asked questions