If your company is closing, its Slack, email, tickets and code may still have value to AI buyers. The order of operations matters: export before you cancel anything, settle who can approve a sale, and only then talk price.
Last checked: 7 October 2026. Prices are attributed to whoever cited them.
Two public reference points, and one gap you should know about.
In 2026 Forbes ("AI's New Training Data: Your Old Work Slacks And Emails," April 16, 2026), Fast Company and Gizmodo reported that shut-down startups were selling their work archives for AI training. Troveo cites roughly $10,000 to $100,000 per archive deal and about $5,000 per code repository in that closure market. Those are one participant's figures, not a price list. Our case notes on shut-down startups cover the coverage and the objections in more detail.
At the other end, Spirit Airlines' internal records drew a $10 million winning bid from Google in its bankruptcy auction in August 2026, as reported by ABC News, TIME and others, and micro1 later made a $12.5 million rival offer. That sale needed a bankruptcy judge's approval, and we could not confirm the final ruling as of 7 October 2026. See the Spirit Airlines case file.
The gap: the buyer programs that publish six- and seven-figure ranges describe running companies. micro1 asks for 30+ employees, "mature operations," documented processes and modern software tools. Mode asks for 20+ full-time US office employees and "several years of records the company owns." Neither page we checked says how it treats a company that is winding down. micro1's referral terms distinguish workflow partnerships from corpus partnerships; a closed company can only offer a corpus, because there is no ongoing workflow left. So ask each program directly whether it will review a company in wind-down before you count on it.
Archives disappear when subscriptions end. Do this before the cancellation emails go out.
Pause any automatic retention or deletion rules, and list every SaaS tool with its renewal or cancellation date. Read each vendor's terms for what happens to your data after cancellation and how long you have to export.
Communication: Gmail or Outlook, Slack or Microsoft Teams, Google Drive, SharePoint, Notion, Confluence, Dropbox. Work tracking and support: Jira, Asana, Monday.com, Zendesk, ServiceNow. Sales and finance: Salesforce, HubSpot, QuickBooks, Xero, NetSuite. Code: GitHub, GitLab or Bitbucket with full history, not just the latest snapshot.
Buyers value connected histories: the ticket, the Slack thread about it, the commit that fixed it. Keep IDs and timestamps intact so those links survive. Our page on codebases and git history explains why commits, reviews and issues together are worth more than code alone.
A one- or two-page inventory: each system, date range, approximate volume, number of users, and what is in it. This is what you share with buyers first, not the data.
Direct messages, HR and payroll channels, legal advice, customer personal data, anything covered by a client confidentiality clause, and secrets such as API keys in code. Run a secrets scan on repositories. The fuller checklist is in prepare your data for sale.
Encrypted storage, a short list of people with access, and a record of who touched what. If a sale never happens, you will still need to delete or retain this copy under your own policies and contracts.
A closing company has more stakeholders, not fewer.
Described by what they publish. We have no relationship that changes what they offer you.
micro1 publishes "$100k+ qualified," "$500k+ large-scale" and "$1M+ highly unique." Mode publishes "$100K-$5M." Grepped publishes "$20K-$5M" and says it covers any vertical. All describe operating businesses, so confirm wind-down eligibility first.
Miro Advisory publishes indicative ranges of "$10K-$1M+" for private codebases and "$100K-$1M+" for operating datasets. Code with history, reviews and issues is the easiest asset to scope after a closure.
The 2026 coverage describes buyers of closed-startup archives; Troveo is the participant whose prices were cited. Expect per-archive or per-repository pricing.
Google has a content intake form and OpenAI has a data partnerships page. Direct deals tend to suit large or unique datasets; most small archives go through a data company.
How to approach them when you are closing. Be upfront that the company is winding down, and say when. A buyer that would have accepted you as a running company may still review a closing one, may want a different structure, or may decline; you want to know which in the first exchange, not after weeks of review. State who has authority to sign, and confirm that the board has approved exploring a sale.
Send the same manifest to each buyer so their answers are comparable. Say what you have already excluded and why. Ask each one three things in writing: whether it reviews companies in wind-down, what de-identification it applies and who does it, and what it needs from you after the company has closed, such as a contact for questions or a signature on acceptance.
Finally, ask how payment works if the company is dissolved before acceptance. The answer affects how long the company must stay in existence, which bank account receives the money, and how proceeds reach creditors or shareholders. This is a question for your lawyer as much as for the buyer.
If no buyer is interested, that is an answer too. Delete or retain the export according to your own policies and customer contracts, and record what you did. An unsold archive sitting on a former founder's drive is a privacy risk with no upside.
The timing point is the one founders miss. A wind-down runs on weeks; a data deal can take months, through NDA, buyer review, agreement, export, de-identification and acceptance. Someone with authority has to stay available to sign, and the data has to stay under control the whole time.
General guidance. Each one is avoidable if it is on the list before the shutdown plan is approved.
The fastest way to lose the asset. Subscriptions get cut to save cash, and the archive goes with them. Put "export complete" on the checklist before any cancellation.
Separate dumps of chat, tickets and code lose the connections between them. Buyers value connected histories, so keep IDs, timestamps and references intact.
The records belong to the company. A founder who sells them personally, or keeps the proceeds, creates problems with the board, investors and creditors.
Many service agreements require return or deletion of customer data at the end. Those obligations do not disappear because the company is closing.
Practitioners say never send a full dataset before price. A manifest and a small redacted sample are enough for an offer.
Closure deals are small and quick, which makes the first offer tempting. Ask at least two buyers. The Spirit auction moved from a $5 million opening bid to $10 million, as reported by Stretto, because more than one bidder was in the room.
Deals take weeks to months. If every officer has left by the time a buyer sends the agreement, the deal stalls. Name who stays available, and for how long.
Put the data decision into the wind-down plan itself, with an owner, a budget for legal review, and dates for export, manifest, outreach and decision.
Size changes what you have to sell, who must approve it, and who will object.
A typical venture-backed startup closure. Your realistic market is closure-style buyers and codebase buyers, plus any program that confirms it will review a company in wind-down. The board and lead investors are usually the approvers. Keep the export small and clean: public channels, project tools and repositories, with personal and customer data excluded.
More systems, more years, more client contracts. The archive may interest data company programs if they accept a closing company, and the extra history adds value. It also adds work: more contracts to check, more former employees in the data, and possibly lenders with a claim on company assets. Budget real legal time.
Closures at this size often run through a formal insolvency process. Then a trustee, assignee or court controls asset sales, as in Spirit's case, and objections are heard in public. Expect privacy review, employee or union input, and a longer timeline. The data may be valuable; the process will be visible.
The single biggest planning problem: they run at different speeds.
Add the data question to the wind-down plan on day one: who owns it, what budget, and what the deadline for exports is.
The people who know the systems are still there. Use them to export, map the systems and mark exclusions. After they leave, nobody can explain the data.
Exports complete and verified, stored under control, manifest written.
Manifest and redacted samples to two or three buyers in the same week, under NDA. Ask for written offers with a validity date.
The long tail. Practitioners cite 60 to 90 days from inquiry to close. Someone with signing authority, and access to the data, has to remain available until payment.
Yes, this happened in 2026: Forbes, Fast Company and Gizmodo reported on closed startups selling work archives for AI training. Whether your company can do it depends on who controls the assets during the wind-down, your contracts with customers and investors, and privacy law. Ask your lawyer before you agree to anything.
Troveo cites roughly $10,000 to $100,000 per archive deal and about $5,000 per code repository in the closure market. Buyer programs publish higher ranges, but they describe running companies with minimum team sizes. Nobody can price your archive without seeing a manifest and samples.
Before you cancel subscriptions, take admin-level exports of chat, email, shared drives, ticketing, CRM, accounting and code repositories with full history. Check each vendor's terms for what happens to data after cancellation. Store the exports securely and record what you have in a manifest.
Usually the board, and possibly investors whose financing documents give them consent rights. If the company is in a formal insolvency process, a trustee, assignee or court may control the sale. The Spirit Airlines data sale, for example, needed a bankruptcy judge's approval.
Their published pages describe operating companies: micro1 asks for mature operations and documented processes, Mode for several years of records the company owns. Neither page we checked says how it treats a company in wind-down, so ask the program before you rely on it.
It is worth discussing with your lawyer. The objections in 2026 coverage, including a union objection that delayed the Spirit sale, show that employees care. Excluding direct messages and HR channels, and requiring de-identification, reduces the risk.
Often, yes, and code can be the easiest asset to scope after a closure. Troveo cites about $5,000 per code repository in the closure market, and Miro Advisory publishes indicative ranges of $10K to $1M+ for private codebases. Check open-source licenses, client ownership of code and secrets in the history first.
Longer than most wind-downs plan for. Practitioners cite 60 to 90 days from inquiry to close, through NDA, buyer review, agreement, export, de-identification and acceptance. A court-supervised sale can take longer: Spirit's data auction was on August 14, 2026, and the sale was still awaiting a ruling weeks later.
Check which published rules you meet, then ask each program whether it reviews companies that are winding down.
Independent site. Some links are referral links: if your company signs with a buyer through them, the buyer may pay us a fee. You are not charged, and we never see your data.