For companies, not individuals. AI companies and labs now pay businesses for records of real work: email threads, chat history, support tickets, SOPs, CRM notes and code with its history. Nobody can price your data without seeing it. What you can see is what buyers publish, whether you meet their rules, and what to check before you sign.
Ranges are as published by each program, not offers.
Our index covers 102 million domains, 99.99% of the active internet. We score every company from 0 to 100 for the data it likely holds, and we keep the history to see which companies were active before and are not anymore.
Factors we use: history, scale, knowledge assets, operational systems, customer systems, organization, industry value, expertise and activity status. The weights are our own.
The public web is open to anyone who crawls it. The record of how a company actually works is not. That record is what labs and the data companies that supply them have started to buy.
A model that has read every public page still has not watched an accountant close a month or a support team trace a bug from the first ticket to the fix. Those sequences sit inside companies: in email, chat, tickets, CRM and Git history. Buyers describe what they want in similar terms. micro1's page names SOPs, knowledge bases, internal documentation, CRM data, project histories, QA processes, "decision-making patterns" and "AI performance feedback".
Three public signals show the market is real, and also how uneven it is.
Google agreed to pay $10 million in Spirit Airlines' bankruptcy proceedings for the airline's internal data (emails, Teams messages, spreadsheets and operations files) to train AI, as reported by ABC, TIME and others. micro1 then made a $12.5 million rival offer.
micro1 later held a session with former Spirit employees and said it pursues "non-sensitive, non-consumer data" with third-party de-identification. Read the Spirit case.
Forbes ("AI's New Training Data: Your Old Work Slacks And Emails", 16 April 2026), Fast Company and Gizmodo covered startups selling Slack and email archives after closing.
Troveo cites about $5,000 per code repository and roughly $10,000 to $100,000 per archive deal in that closure market. Read the closure-market case.
micro1, Mode and Grepped publish company data programs with payout ranges and eligibility rules. Miro Advisory publishes indicative ranges. Google runs its own intake page, and OpenAI has a data partnerships page.
The rules differ a lot by size, country and industry. Compare the programs side by side.
Spirit was a large airline in liquidation with no business left to protect. A running company with 40 staff is a different seller. Use these figures to understand the market, not to price your own records.
Each program publishes its own minimums. A 12-person law firm and a 120-person software company can both fit, but not the same programs. As published, checked 7 October 2026.
This site is not for individuals who want to sell their personal data, browsing history or app activity. Consumer "get paid for your data" apps are a different market with different buyers, and we do not cover them. Everything here is about a company licensing its own business records, with its lawyers and its staff in the loop.
Not sure where your company stands? The eligibility checker compares your answers with each program's published rules, in your browser.
Buyers list the tools they can take exports from. The value is rarely one file. It is the connected history: the ticket, the thread about it, the fix and the reply.
Gmail, Outlook, Google Drive, SharePoint, Slack, Microsoft Teams, Notion, Confluence, Dropbox, DocuSign, Zoom recordings and transcripts.
Slack and TeamsTwo rules in the published eligibility lists tell you what makes records sellable: depth (Mode asks for several years of records the company owns) and structure (micro1 asks for documented processes and modern software tools). A few months of scattered files on a shared drive rarely meets either. The guide to what data AI labs want goes through it system by system.
Exports of email, chat, tickets or code. Practitioners say this is the cheapest form.
Tests built on your data that check whether a model does the work correctly. Practitioners put these at roughly 10x raw.
Full simulated workflows. Practitioners say these reach 6 to 8 figures but need heavy engineering. Compare the tiers.
Quoted as each buyer publishes it on its own pages. These are ranges, not offers or averages, and nothing here is a quote for your company.
| Program | Company payout (as published) | Eligibility (as published) | Referral link here |
|---|---|---|---|
| micro1 Enterprise Data Partnership | "$100k+ qualified", "$500k+ large-scale", "$1M+ highly unique"; referral page: "$100K-$2M+ for approved data packages"Source: micro1.ai/data-partnerships and micro1.ai/company-referral | 30+ employees (referral posting: 30 to 200), documented processes, modern tools, primarily English, US prioritized | Yes |
| Mode company data | "$100K-$5M"Source: data.mode.inc | 20+ full-time US office employees; accounting firms 10+; law firms 6+; several years of owned records | Yes |
| Grepped | "$20K-$5M"; also publishes "get paid in 7 days"Source: grepped.ai | Any vertical | Yes |
| Miro Advisory | Operating datasets "$100K-$1M+"; private codebases "$10K-$1M+" (indicative)Source: miroadvisory.com | Businesses; software companies | No |
| Direct to labs | No published rangeGoogle intake page; OpenAI data partnerships page | Large or unique datasets | No |
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. How referrals work.
This site explains, compares and links out. The buyer runs everything from the first call to the payment. We never touch your data.
Answer seven questions in the eligibility checker: country, headcount, industry, years of records, systems, language and whether client-confidential data is involved. It runs in your browser and sends nothing.
List which systems you would offer and which you would never include: client files, patient records, HR cases, anything under a client NDA. The preparation guide has an inventory you can work through.
An apply button here goes through our /go/ link and forwards you to the program's own application. You fill in their form. Applying to more than one program is how you get more than one offer.
The buyer contacts you, runs its own call, signs an NDA and reviews what you have. Practitioners advise sharing a manifest and samples at this stage, never the full dataset before a price is agreed.
You negotiate the agreement directly with the buyer, with your own lawyer. Scope of use, exclusivity, warranties, indemnities, payment milestones and deletion are decided here. See the licensing agreement outline.
The buyer handles the export and de-identification as the contract says, checks the data against the acceptance criteria and pays. Deals take weeks to months; practitioners cite 60 to 90 days to close. See how long a deal takes.
No program publishes a scoring formula. Their published rules and privacy statements point to six things they review, usually from a manifest and samples shared under NDA.
Years, not months. Mode asks for several years of records the company owns. A long history shows how decisions played out, not only how they started.
Records that link up: the ticket, the chat thread about it, the commit that fixed it and the reply to the customer. micro1 names project histories and "decision-making patterns".
micro1 asks for mature operations and documented processes. SOPs and wikis let a buyer compare what the records show with how the work is meant to run.
You can only license what the company owns. Client deliverables, content licensed from others and data held for a third party may not qualify.
micro1 says sensitive and confidential information is scrubbed and no customer information is exposed; Mode says it de-identifies before onward delivery. The more personal data there is, the more work that means.
micro1 lists primarily English and prioritizes the US, then other Western markets. Mode names US-based teams as the strongest fit. Other regions face a smaller pool of programs.
These apply to any buyer. They are questions to put to the contract and to your lawyer, not claims about any program.
Is the license exclusive, time-limited or open? May the buyer resell your data, and to whom? Selling the same records twice can breach the first contract. Exclusivity explained.
Training only, evaluation too, or anything? Which downstream buyers can receive it?
Who pays if de-identification misses something? Is your liability capped, and does it expire? Indemnities explained.
What will you be asked to promise about consent from employees, customers and clients? Employees and data sales.
Law, accounting, M&A and healthcare firms hold secrets that belong to clients. Some records can never be included. Client confidentiality.
One-off or recurring? Paid on milestones or on acceptance? What exactly are the acceptance criteria?
Can you verify how names and identifiers are removed, and is there any audit right? De-identification guide.
When are originals and the delivered copy deleted? What survives termination?
General information, not legal advice. Talk to your own lawyer before you sign. Laws that often come up: GDPR, CCPA/CPRA, HIPAA, GLBA and attorney-client privilege. Start with is it legal to sell company data.
Every guide quotes buyers only from their own published pages, with the date we checked them.
The core guides for a first decision.
What each system holds and what to leave out.
Published minimums and sector rules for your field.
Questions for your lawyer, by topic.
Public deals, with sources and dates.
Getting better terms, and how intermediaries are paid.
Some companies can. micro1, Mode and Grepped publish programs that pay businesses for internal records, with published payout ranges and eligibility rules (as published, checked 7 October 2026). Whether your company qualifies depends on its size, country, years of records, the systems it uses and what the data contains. No buyer promises acceptance, and neither do we.
Nobody can price it without seeing it. What you can compare are the published ranges: micro1 lists “$100K-$2M+ for approved data packages”, Mode lists “$100K-$5M” and Grepped lists “$20K-$5M” (as published, checked 7 October 2026). These are ranges a program publishes, not offers, and the top of a range is not the typical result.
Most companies deal with data companies that collect, clean and de-identify records and then supply AI labs. Some labs also run their own intake: Google has an intake page and OpenAI has a data partnerships page. Direct intake usually suits large or unique datasets.
No. This site never receives, stores, views or transfers any company’s data. The eligibility checker runs in your browser and sends nothing. If you apply, you deal with the buyer directly: the buyer runs the call, the contract, the export and the payment.
Weeks to months. A typical path is NDA, buyer review, agreement, export, de-identification and acceptance before payment. Practitioners cite 60 to 90 days to close. Treat any faster figure as a best case, not a plan.
It depends on who owns the records, what your client contracts say, whether employees were told, what personal data is inside, and sector laws such as GDPR, CCPA/CPRA, HIPAA, GLBA and attorney-client privilege. General information, not legal advice. Talk to your own lawyer before you sign.
Not according to what two programs publish. micro1’s page says the company keeps ownership of its underlying data and that originals are deleted after processing. Mode says it buys “an agreed copy” and that originals stay with the company. Your own contract decides the details, so read the license, deletion and survival clauses.
Only if your contracts allow it. An exclusive license, even a time-limited one, can stop you licensing the same records again, and selling them twice can breach the first agreement. Practitioners advise collecting more than one offer before you sign anything, then reading exclusivity and resale terms side by side.
Check your company against each program's published eligibility in about two minutes, then compare the programs that fit. Nothing you enter leaves your browser.