Independent representation for data sellers
Leading AI labs will pay $50K to $1M+ for the operational data you already have.
Slack, docs, Drive, email, CRM/ERP and SOPs. We package and de-identify it, get competing offers from multiple labs, and bring you the best one.
- No upfront cost
- No equity
- No commitment
- Nothing shared without approval
Get your offer
Step 1 of 3Labs want how work gets done: workflows, decisions and how work moves through systems. Not who said what.
Why the data is valuableHow it works
From rough estimate to signed offer
You stay in control at every step. We do the packaging, the outreach and the negotiating, the way a banker would run a sale process.
- i.
Share rough estimates
Tell us which systems you use, roughly how much is in each, years of history, headcount and years in business.
- ii.
We package & de-identify
We scope the dataset with you, then export, clean and de-identify it, or guide your team to do it. Exclusions are honored.
- iii.
Labs compete
We take a packaged description to multiple AI labs and run a competitive process, so you're not negotiating against a single buyer.
- iv.
You approve, you get paid
You see the buyer, the price and the terms. Nothing is shared until you say yes. If it doesn't work for you, walk away.
Indicative terms · Example
Summary of competing offers
- Licensor
- Your company
- Dataset
- Operations workspace, 6 years
- Scope
- Slack · Docs · CRM
- Privacy
- De-identified before delivery
- Structure
- License. You keep your data.
Offers received
De-identification
The workflow stays. The identities don't.
Names, emails, phone numbers, addresses, account numbers and other identifiers are removed or replaced with consistent placeholders. Labs keep the shape of the work. You approve the exact scope before anything is delivered.
Maria Chen9:14 AM
@Tom Alvarez can you approve the revised quote for Brightline Foods? They need it by Thursday.
Tom Alvarez9:21 AM
Approved. Send it from tom@hartwell-logistics.com and copy the account manager.
PERSON_309:14
@PERSON_7 can you approve the revised quote for CUSTOMER_22? They need it by Thursday.
PERSON_709:21
Approved. Send it from and copy the account manager.
Removed or replaced
- Names
- Emails
- Phone numbers
- Addresses
- Account numbers
- Customer names
- Excluded channels & folders
Where the data lives: shop floors, dispatch desks, file rooms and the systems behind them.
What qualifies
Data you already have, in tools you already use
The most valuable datasets show real work moving through real systems over several years. You choose what's included, system by system.
| System | What it shows a buyer | Typical sources |
|---|---|---|
| Slack & chat | Channels and threads that show how decisions actually get made. | Slack, Microsoft Teams |
| Docs & wikis | Specs, plans, retros and notes: how the company thinks on paper. | Notion, Confluence, Google Docs |
| Shared drives | The working files behind real projects: decks, sheets, drafts. | Google Drive, OneDrive |
| Request, approval and escalation patterns across teams and vendors. | Google Workspace, Microsoft 365 | |
| CRM & ERP | Pipelines, orders, tickets and the operational logic around them. | Salesforce, HubSpot, NetSuite, QuickBooks |
| SOPs & playbooks | Step-by-step procedures that encode how your business runs. | Runbooks, checklists, training docs |
Our commitments
Privacy, control, low effort, no strings
The four things owners ask about first, answered plainly. The specifics go in writing before you sign anything.
Privacy
Everything is de-identified. Labs want how work gets done (workflows, decisions, how work moves through systems), not who said what.
Control
You choose which systems are included. Nothing is shared until you approve the buyer, the price and the terms.
Effort
Our engineers can handle the export and cleanup, or your team can run it with our playbooks. Your call.
No strings
No upfront cost, no equity and no commitment to get an offer. If the number doesn't work, you walk away.
Who qualifies
A good fit if you can check most of these
Not sure? Submit rough estimates anyway. We'll tell you honestly whether there's a market for your data.
- US-based companyOur current buyer programs focus on US businesses.
- 20 or more employeesEnough people that real workflows and handoffs exist.
- Data in modern systemsSlack, Google Workspace or Microsoft 365, a CRM/ERP, wikis.
- A few years of historyLonger histories show how work evolves. That's valuable.
Value estimator
Get a feel for the range
Move the sliders and pick your systems. This is a simple illustrative heuristic to help you decide whether it's worth a conversation. It is not an offer. Real offers come from labs competing for your dataset.
- More systemsChat, docs and email together show a fuller picture of work.
- More historySeveral years show how processes change over time.
- Unique workflowsWell-documented, specialized work matters more than raw size.
We're focused on 20+ employees, but specialized datasets are considered case by case.
Illustrative only. This is a rough heuristic, not an offer, quote or guarantee. Real offers depend on lab demand, data quality and uniqueness, de-identification scope, and terms.
You approve the buyer, the price and the terms before anything is shared.
How much could my company get paid?
Offers for US companies with 20+ employees generally fall between $50K to $1M+. Where you land depends on which systems you include, how much history you have, how unique your workflows are, and how many labs are bidding. The estimator on this page is a rough illustration, not an offer. The real number comes from competing lab offers.
Who actually buys the data?
Leading AI labs that train and evaluate models on realistic business work. We run a competitive process across several labs and bring you the best offer. You know who the buyer is and what they're paying before anything is shared, and you can say no.
What do AI labs want from company data?
They want how work gets done: workflows, decisions, handoffs, and how work moves through systems like Slack, docs, email and CRM/ERP. They aren't after personal details or who said what, which is why everything is de-identified before delivery.
How is our data de-identified?
Names, emails, phone numbers, addresses, account numbers and other identifiers are removed or replaced with consistent placeholders, and free text is scanned for sensitive details. You can exclude channels, folders, mailboxes or record types entirely. The exact scope is agreed in writing, and you approve it before delivery.
Do we lose ownership of our data?
Most transactions are structured as licenses rather than outright transfers, so you keep your data and keep using it. The specific rights (scope, exclusivity, duration, permitted uses) are set in the agreement you review and approve. We recommend having your counsel review it.
What does it cost us?
Nothing upfront. There's no equity, no retainer, and no commitment to get an offer. How we're compensated is disclosed in writing before you sign anything.
How much work is this for our team?
As little as you want. Our engineers can handle the export, cleanup and de-identification under your supervision, or your team can run the export using our checklists. Most of your time goes into deciding scope and approving the final terms.
Can we choose which systems are included?
Yes. You decide system by system (and within a system, e.g. specific Slack channels or Drive folders). Anything you exclude stays out.
Is it legal to sell or license our company data?
Often yes, but it depends on your contracts, privacy notices, industry rules and the data involved. Customer data, health data and financial data can carry extra obligations. We help you scope around restricted data, and we recommend legal review before you sign. Our guides cover the main questions, but they aren't legal advice.
We're under 20 employees or outside the US. Do we qualify?
Our current focus is US companies with 20+ employees. Specialized datasets (real estate, medical, manufacturing, trading) are considered case by case, so it's still worth asking.
What do you need from us to get an offer?
Only rough estimates: which systems you use, about how much data is in each, years of history, headcount and years in business. No files are uploaded to get an offer.
What happens after I submit?
We review your estimates, schedule a short call to confirm scope, then take a packaged, de-identified description to multiple labs. You see the competing offers and decide. Nothing is shared until you approve.
Guides
Learn before you sell
How much is my company's data worth to AI labs?
What drives the value of operational company data for AI training (volume, history, uniqueness, quality and terms) and how to get a realistic number.
Read guide · 4 min
How to sell company data to AI companies
A practical, step-by-step guide to selling or licensing your company's operational data to AI labs, from what to inventory to how offers and approvals work.
Read guide · 5 min
What AI labs look for in enterprise data
The characteristics that make enterprise data valuable to AI labs (real workflows, decisions, history and context) and what makes a dataset less useful.
Read guide · 4 minFind out what your data is worth
Rough estimates are enough. No upfront cost, no equity, no commitment, and nothing is shared until you approve the buyer, price and terms.