For owners and finance leads of companies with roughly 20 to 500 people. Eight steps, from listing your systems to getting paid, with what you do and what the buyer does at each one. Buyer terms are quoted as each program publishes them.
AI companies want records of how real work gets done: the ticket, the discussion, the decision, the fix. You do not sell your company or hand over your systems. You license a defined copy of certain records, for a defined use, under a written contract.
A written agreement in which your company lets a buyer use an agreed copy of specific records, for agreed purposes, in exchange for payment. Your contract should state that ownership and the originals stay with you.
Mode states, as published, that it buys “an agreed copy” and that originals stay with the company. micro1 states that scope is agreed in writing and the company keeps ownership of its underlying data.
micro1's page lists SOPs, knowledge bases, internal documentation, CRM data, project histories and QA processes, plus “decision-making patterns.” See what data AI labs want.
Discovery, contract drafting, export, de-identification and payment sit on the buyer's side. Your time goes into scoping, internal approvals and legal review.
Do these three before you contact any buyer. They decide which programs you fit and what you can safely offer, and they keep the buyer's questions from setting your boundaries for you.
Write a one-page list. Buyers screen on a small set of facts, and their published rules are specific about each one.
Gmail, Outlook, Google Drive, SharePoint, Slack, Microsoft Teams, Notion, Confluence, Dropbox, DocuSign, Zoom recordings and transcripts.
Jira, Asana, Monday.com, Zendesk, ServiceNow.
Salesforce, HubSpot, QuickBooks, Xero, NetSuite, Paychex.
GitHub, GitLab and Bitbucket repositories with history; AutoCAD, Figma, ServiceTitan.
Systems named on buyer pages, as published, checked 7 October 2026.
Scope is the list of systems, date ranges and record types you are willing to license. The exclusions list matters more: it names what will never leave, whatever the offer. Write both before the first conversation.
Also decide who signs off internally (the role table below splits the decisions) and how you will tell employees whose messages are in scope. Our guide to preparing your data for sale covers inventory, exclusions, retention and a secrets scan in more depth.
Match your inventory against what each program publishes. Eligibility comes first. A payout range is the program's own description of past or possible deals, not a quote for your data.
| Program | Published payout | Published eligibility | What to note |
|---|---|---|---|
| micro1 Enterprise Data Partnership | “$100k+ qualified”, “$500k+ large-scale”, “$1M+ highly unique”; referral page: “$100K-$2M+ for approved data packages” | 30+ employees, mature operations, documented processes, modern software tools, primarily English; US prioritized, then other Western markets | Scope agreed in writing; originals deleted after processing; company keeps ownership (its statements) |
| Mode | “$100K-$5M” | 20+ full-time US office employees; accounting firms 10+; law firms 6+; several years of records the company owns; US-based teams strongest fit | Buys “an agreed copy”; originals stay with the company; de-identifies before onward delivery |
| Grepped | “$20K-$5M”; “get paid in 7 days” | Any vertical; also pays individual professionals for expertise | Ask when the 7-day clock starts and what must happen first |
| Miro Advisory | Operating datasets “$100K-$1M+”; private codebases “$10K-$1M+” (indicative) | Businesses; software companies | Indicative ranges; no apply link on this site |
Last checked: 7 October 2026. Sources: micro1.ai/data-partnerships, micro1.ai/company-referral, data.mode.inc, grepped.ai, miroadvisory.com, as published. Figures are each program's own wording, not offers.
How to read it: a 15-person accounting firm meets Mode's published 10+ rule and not micro1's 30+. Some labs also run direct intake (Google publishes one; OpenAI has a data partnerships page), but those routes are for large or unique datasets. For the full side-by-side, see buyer programs compared or the eligibility checker.
In a 20 to 500 person company, licensing records to an AI company is nobody's day job. Name one internal owner, usually the owner or the finance lead, and give each role a clear decision before you apply.
| Role | What they decide | Steps |
|---|---|---|
| Owner or CEO | Whether to sell at all, which programs to approach, the final exclusions list and the signature. | 136 |
| Finance lead | The inventory facts buyers screen on (headcount, years of records), payment structure, milestones and invoicing; tax treatment with your accountant. | 168 |
| Outside lawyer | Client contracts, the NDA, the agreement clauses, the consent statements you would give, and employee notice. | 246 |
| System admin | What each system can export, scoped and time-limited access, a scan for passwords and keys, and the export log. | 17 |
| HR or people lead | The employee notice plan, which channels and mailboxes stay out, and answers to staff questions. | 27 |
Steps refer to the eight steps on this page. In smaller firms one person often holds two roles; the decisions still need an owner.
Once you apply, a deal moves through these eight stages. Practitioners cite 60 to 90 days from first contact to close. Treat that as a range, not a promise.
What slows each stage, and why short timelines are rare, is covered in how long an AI data deal takes.
These are questions every seller should check in any data agreement. They are generic. They are not claims about any program named on this page.
Is the license non-exclusive, time-limited or perpetual? Can you license the same records to anyone else?
Who else may receive the data? May the buyer resell it, and to whom?
Training only, evaluation, or both? For which models or customers?
Who pays if de-identification misses something? Is liability capped, and when does it expire?
What are you promising about employees, customers and clients, and can you stand behind it?
Does anything in scope breach a promise you made to a client in law, accounting, M&A or healthcare work?
One-off or recurring? Which milestones? What exactly are the acceptance criteria?
Can you verify de-identification? When are originals and the copy deleted? What survives termination?
Each one happens on the seller's side, before a buyer drafts anything, so each one is in your control.
Without an inventory and an exclusions list, the buyer's first questions end up setting your boundaries.
Fix: finish steps 1 and 2 before you fill in any intake page.
A copy that has left cannot be recalled, and it removes your reason for a buyer to compete on terms.
Fix: manifest and cleaned samples only, as practitioners advise, until there is a price and a signed agreement.
Confidentiality terms you signed with clients still bind you after a buyer de-identifies the data.
Fix: list every client with an NDA or professional-secrecy duty and have your lawyer clear the scope against it.
Buyer requests sit between IT, finance and the owner, and each stage waits on the last person who saw it.
Fix: name one project owner and use the role table above.
Employees whose messages are in scope hear about it after the export. That is a trust problem even where notice is not required.
Fix: agree the notice plan with HR and your lawyer before step 4.
A higher figure tied to perpetual exclusivity, uncapped liability or vague acceptance criteria can be worth less than a lower, cleaner offer.
Fix: compare scope, exclusivity, liability, payment timing and deletion side by side.
Example Logistics Co. is invented to show how steps 1 to 3 fit together. It has a US office, works in English, and has used Slack since 2018, Jira since 2019, Zendesk since 2020, Confluence for SOPs, QuickBooks and Paychex.
Next: manifest and cleaned samples to more than one program, NDAs, then compare proposals on scope, exclusivity and payment terms with a lawyer. No price is shown on purpose: nobody can price records without seeing them.
Sometimes. As published and checked on 7 October 2026, Mode lists 20+ full-time US office employees in general, but 10+ for accounting firms and 6+ for law firms. micro1 lists 30+ employees. Grepped lists any vertical and also pays individual professionals for expertise. Below those lines nobody publishes a promise. Run the eligibility checker to see which published rules you meet.
Two programs publish a statement on this. Mode says it buys an agreed copy and the originals stay with the company. micro1 says scope is agreed in writing, originals are deleted after processing and the company keeps ownership of its underlying data. Your contract should say the same thing in plain words, including what happens to the copy.
Weeks to months. A deal moves through inquiry, NDA, review, agreement, export, de-identification, acceptance and payment, and practitioners cite 60 to 90 days to close. Client approvals, legal review and export work are common sources of delay. No one can promise a date, and this site does not.
Practitioners advise getting more than one offer and never sending a full dataset before there is a price: share a manifest and samples instead. Before you apply widely, read each NDA and check whether any term would stop you from licensing the same records to someone else.
A manifest is a short written description of what you could license: the systems, date ranges, rough record counts, languages, record types and your exclusions list. Practitioners advise sharing a manifest and small cleaned samples, not a full dataset, until there is a price. The buyer can judge fit, and you keep the ability to compare offers.
Plan on it. Chat, email and tickets carry employees' names and words, and notice rules differ by state and country; in the EU and UK, GDPR and works councils can add steps. Decide early what is in scope, exclude private messages and HR channels where you can, and ask your lawyer what notice or consent applies before any export.
Yes, before you sign. A data license touches client contracts, employee notice, personal-data law and sector rules such as HIPAA or GLBA. This page is general information, not legal advice. Talk to your own lawyer before you sign.
Start with the eligibility checker: it compares your headcount, sector, systems and years of records with each program's published rules. Nothing you enter leaves your browser.
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.