Get paid in 7 days

How much is your data worth?

Lower range$800K to Upper range$1M
LowHigh
100 employees
201002505001,000+
How long has the company been around?

Illustrative estimate in USD, not an offer. Actual value depends on your data and buyer demand. No sale or payment is guaranteed.

Get a detailed assessment Book a 30-minute call

What business data means

What data are we talking about?

Useful data shows how your business handles a request, makes a decision, and gets a result. Think email threads, sales histories, and solved problems — not a contact list.

Choose a category to see an example.

Slack · Microsoft Teams

Team conversations

Questions your team discusses, the replies, and the decisions they make.

Message threads · Project discussions · Decisions See a real-world exampleClose example

What a record could look like

Question
A packaging line is down. Do we have the replacement part in stock?
Discussion
Maintenance checks inventory and confirms the part is on shelf 4.
Result
Line restarts in 40 minutes. Downtime is logged.

Made-up records, not real customer data

Gmail · Outlook

Business email

The back-and-forth behind a client request, an approval, or a project.

Email threads · Follow-ups · Approvals See a real-world exampleClose example

What a record could look like

Request
Can we move the closing to Friday?
Reply
The firm checks the calendar and confirms the new date.
Result
The client accepts the rescheduled closing.

Made-up records, not real customer data

Salesforce · HubSpot

CRM & sales records

How a sale happened: what the customer needed, your follow-ups, and why they bought.

Sales notes · Deal stages · Won / lost reasons See a real-world exampleClose example

What a record could look like

Sales note
A fleet customer needs six vans before quarter-end.
Follow-up
Demo completed. Customer asks about service contracts.
Deal stage
Enquiry → Demo → Proposal → Won
Outcome
Customer signed after a revised service offer.

Made-up records, not real customer data

Zendesk · Intercom

Customer support

A customer's problem, the steps your team took, and how it was fixed.

Support tickets · Replies · Resolutions See a real-world exampleClose example

What a record could look like

Problem
A customer's AC failed during a heatwave.
Team response
Dispatch prioritizes the call and sends a tech same-day.
Resolution
Compressor replaced. Ticket closed with a 5-star review.

Made-up records, not real customer data

Notion · Google Drive

Documents & guides

The instructions and knowledge your team uses to get work done.

How-to guides · Checklists · Project handovers See a real-world exampleClose example

What a record could look like

Guide
How to close the kitchen at night
Steps
Lock the registers. Clean the fryers. Set the alarm.
Check
Manager verifies the checklist before sign-off.

Made-up records, not real customer data

Zoom · Recorded calls

Call transcripts

Written conversations from business calls: the questions, answers, and agreed next steps.

Sales calls · Support calls · Next steps See a real-world exampleClose example

What a record could look like

Customer question
Do you take our insurance plan?
Team answer
Front desk verifies coverage while the patient waits.
Next step
Appointment booked with a cost estimate sent first.

Made-up records, not real customer data

Where it lives

Your business already has
the data frontier AI needs

HubSpotSalesforceServiceNow Dynamics 365ZendeskJira QuickBooksSlackAsana DocuSignPipedriveXero HubSpotSalesforceServiceNow Dynamics 365ZendeskJira QuickBooksSlackAsana DocuSignPipedriveXero

Every day, your teams generate operational data across sales pipelines, pricing, invoicing, forecasting, and finance — data that captures workflows highly valuable for training frontier AI models.

Privacy, made simple

The useful work stays.
Identifying details come out.

Before anything is used for AI training, identifying details are removed or replaced under the process we agree with you. Messages and documents are also reviewed for sensitive information beyond names.

Before preparation An example from a work email

Jim Smith resolved a delivery issue for Northwind Ltd. Send the revised plan to jim@example.com.

Jim Smith confirmed the customer accepted the new date.

After preparation The same work, with replacement labels

Person 14 resolved a delivery issue for Company 08. Send the revised plan to Email 14.

Person 14 confirmed the customer accepted the new date.

Replace identifying details

Names become consistent labels across the dataset. Company names and email addresses get labels too.

Remove sensitive information

Passwords, payment credentials, and other private fields are excluded — never offered for sale.

Keep the useful story

The problem, the steps taken, and the result stay connected. That is what helps AI learn.

Made-up example. Replacing names is only one step: other details can still identify someone, so the agreed process checks identifying context and sensitive fields before anything is used for training.

Why buyers want your data

Your data helps AI learn.

AI developers and research teams look for the problem, the response, and the result. Your records can help in three ways — only for the uses you approve in the license.

Train AI

Show AI how people solve problems, make decisions, and complete real tasks.

For example: a support conversation shows a customer's problem, the team's response, and the fix.

Improve AI

Help an existing AI get better at a specific job — this is called fine-tuning.

For example: sales records help an AI practice writing more useful follow-up messages.

Create practice data

Use patterns from real work to create new, made-up training examples — called synthetic data.

For example: a delivery issue inspires new practice scenarios with different details.

What happens when the license ends?

The agreement sets when the raw dataset must be deleted and when future licensing can stop. Training permission does not automatically allow public release or resale of the raw data. Models already trained, their outputs, and derivatives may remain — deleting the dataset does not reverse completed training.

The process

From your tools to a signed deal.

Your records stay in your systems during onboarding. Start with an assessment, then verify counts with screenshots. Data moves only after a buyer deal is agreed.

Your team's workload: about 2 hours. We guide you — older systems or several data sources may need more time.

  1. 01

    No data upload

    Assess the fit

    Tell us about your company and the systems you use. The initial assessment needs no raw data.

  2. 02

    Screenshots

    Verify the counts

    Your team confirms record counts with screenshots. Your business records stay in your systems.

  3. 03

    Your terms

    Agree the scope

    Choose the systems, folders, and records to include. Set the permitted uses and buyer restrictions in the agreement.

  4. 04

    No guarantee

    Find a buyer

    The opportunity is presented to buyers. A listing is not a guaranteed sale.

  5. 05

    After a deal

    Prepare and deliver

    After a buyer deal, data moves to the agreed controlled environment. Identifying details are removed or replaced before AI training. You get paid for every approved sale under the agreement.

Before you decide

Your control. Your terms.

What you share

Counts and screenshots first. The agreed records move only after a buyer deal.

What you keep

Your business, your original records, and ownership of the data.

What stays out

Passwords, payment credentials, and excluded sensitive records. Identifying details need review before training.

Who does the work

We guide the process. Allow roughly two hours for verification; requirements vary.

What you approve

The data scope, uses, and restrictions in your agreement.

When you get paid

The schedule and acceptance conditions are set in your agreement. A signed license does not guarantee a sale.

Security

How we secure your data

Before any participation, together we define:

Approved data scope and clear boundaries

  • Original datasets are retained only for processing purposes.
  • Original datasets are deleted after processing.

Security and customer confidentiality standards

  • No data is publicly shared.
  • No customer information is exposed.

Encrypted, read-only access

  • Data is encrypted in transit and at rest.
  • Access is read-only and revoked when the project ends.

Written terms you sign off on

  • Data use, retention, and payment are set out in a signed agreement.
  • You approve every source before anything is shared.

Straight answers

You decide exactly what you share.

Start with counts and screenshots, not a data upload. Your company chooses the scope. These answers explain what to check before you commit.

Who buys the data?

We license de-identified business workflow data to AI labs and research teams training frontier models. Your data is never resold to competitors or published.

What kind of data qualifies?

Operational records from the tools your team uses daily: CRM pipelines, support tickets, contracts, invoices, project boards, SOPs, and internal docs. Larger teams and more years of history generally mean a higher payout.

Can a company outside tech take part?

Potentially. A delivery problem, your team's response, and the outcome can be useful — alongside sales, support, and operational records. We assess the records and your right to share them, not just your industry.

Do we have to share everything?

No. You can exclude specific systems, folders, customers, or sensitive workflows. The agreed scope defines exactly which records are offered.

What information should stay out?

Passwords, payment credentials, API keys, and other sensitive records. Tell us what systems you use first so we can agree on exclusions and handling before anything moves.

Will my customers’ information be exposed?

No. Personal and customer-identifying information is removed or replaced during processing, and original datasets are deleted once processing is complete.

Does removing names make data anonymous?

Not by itself. A message or a combination of details can still identify someone. Our review goes beyond names and email addresses — replacing names is not a guarantee that every privacy risk is removed.

Can we restrict buyers and uses?

Yes — raise excluded buyers, competitors, or uses before signing. These restrictions belong in the agreement. Check the terms that apply to your dataset rather than assuming every sale needs a new approval.

Do we keep ownership?

Yes. Licensing grants specified rights to use the data, not ownership. You must have the rights to license it, including any necessary third-party permissions.

Where does our data go?

After a buyer deal is agreed, data goes to the de-identification step or the buyer's controlled environment, as specified in the agreement. Buyers never get access to your live business systems.

How much work is involved?

Allow roughly two hours for verification as a starting guide. Older systems, several data sources, or extra privacy checks can take longer. We explain the steps, responsibilities, and any costs before you commit.

How is the payout calculated?

The estimate is based on your team size and how long you have been in business. The final amount depends on data sources, volume, quality, access terms, and due diligence. The calculator is an illustration — a detailed assessment gives a better-informed estimate, not a guaranteed offer.

Find out if your data qualifies.

In 30 minutes, we will discuss the tools you use, what you could share and how the process works. No data upload needed for the call.

Know another company that’d be a good fit?