Before preparationAn 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.
Business data licensing
Turn your existing business data into a new source of revenue.
AI companies pay for real examples of how businesses work. Maryah licenses the emails, sales records and support tickets you choose to share, with identifying details removed and terms you approve.
$20M+Paid to companies
100+Buyers right now
4×Average sales per dataset
How much is your data worth?
Lower range$800KtoUpper 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.
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 TeamsTeam conversations
Questions your team discusses, the replies and the decisions they make.
Message threads · Project discussions · DecisionsSee a real-world exampleClose example
What a record could look like
- Question
- A customer cannot sign in. What should we check?
- Discussion
- Check their account settings and reset access.
- Result
- Access restored. The team records the fix.
Gmail · OutlookBusiness email
The back-and-forth behind a customer request, an approval or a project.
Email threads · Follow-ups · ApprovalsSee a real-world exampleClose example
What a record could look like
- Request
- Can we move the delivery to Friday?
- Reply
- The team checks stock and confirms the date.
- Result
- The customer accepts the new delivery plan.
Salesforce · HubSpotCRM & sales records
How a sale happened: what the customer needed, your follow-ups and why they bought.
Sales notes · Deal stages · Won / lost reasonsSee a real-world exampleClose example
What a record could look like
- Sales note
- Company 12 needs a system for three offices.
- Follow-up
- Demo completed. Customer asks for a smaller plan.
- Deal stage
- Enquiry → Demo → Proposal → Won
- Outcome
- Customer chose the smaller plan after a revised offer.
Zendesk · IntercomCustomer support
A customer’s problem, the steps your team took and how it was fixed.
Support tickets · Replies · ResolutionsSee a real-world exampleClose example
What a record could look like
- Problem
- An order arrived with one item missing.
- Team response
- Check the order and arrange a replacement.
- Resolution
- Replacement delivered. Ticket closed.
Notion · Google DriveDocuments & guides
The instructions and knowledge your team uses to get work done.
How-to guides · Checklists · Project handoversSee a real-world exampleClose example
What a record could look like
- Guide
- How to onboard a new customer
- Steps
- Confirm their needs. Set up the account. Run a walkthrough.
- Check
- Make sure the customer can complete their first task.
Zoom · Approved business callsCall transcripts
Written conversations from business calls: the questions, answers and agreed next steps.
Sales calls · Support calls · Next stepsSee a real-world exampleClose example
What a record could look like
- Customer question
- Can this work with the tools we already use?
- Team answer
- Here is how the setup would work.
- Next step
- Send a setup guide and arrange a demonstration.
The useful work stays. Identifying details come out.
The agreed preparation process removes or replaces identifying details before AI training. Messages and documents also need review for sensitive information beyond names.
After preparationThe 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
Jim Smith becomes Person 14 consistently across the dataset. Company names and email addresses get labels too.
Remove sensitive information
Passwords, payment credentials and other private fields are excluded, not 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 may still identify someone. The agreed process checks identifying context and sensitive fields before AI training.
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, under uses you agree to in the licence.
Train AI
Show AI how people solve problems, make decisions and complete real tasks.
For exampleA 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 exampleSales records help an AI practise 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 exampleA delivery issue inspires new practice scenarios with different details.
What happens when the licence 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.
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 workloadAbout 2 hours
Maryah guides you. Allow about two hours for verification. Older systems or several data sources may need more time.
01
Assess the fit
Tell us about your company and the systems you use. The initial assessment needs no raw data upload.
02
Verify the counts
Your team confirms record counts with screenshots. Your business records stay in your systems.
03
Agree the scope
Choose the systems, folders and records to include. Set the permitted uses and buyer restrictions in the agreement.
04
Find a buyer
The opportunity is presented to buyers. A listing is not a guaranteed sale.
05
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.
Ready to take the next step? Review the licensing agreement
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
- Maryah guides the process. Allow roughly two hours for verification; requirements vary.
- What you approve
- The data scope, uses and restrictions in your agreement. Check how approvals work before signing.
- When you get paid
- The schedule and acceptance conditions are set in your agreement. A signed licence does not guarantee a sale.
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.
Can a logistics company take part?
Potentially. A delivery problem, your team’s response and the outcome may be useful, alongside sales, support and operational guides. 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 which records are offered for licensing.
What information should stay out?
Do not submit passwords, API keys, payment details or other sensitive records through this website. Tell us what systems you use first so we can agree suitable exclusions and handling.
What do we provide during onboarding?
Information about your systems, record counts and screenshots for verification. You do not upload raw business records at this stage. Data moves only after a buyer deal is agreed.
Does replacing names make data anonymous?
Not by itself. A message or a combination of details can still identify someone. The review must go beyond names and email addresses. Replacing names is not a guarantee that every privacy risk is removed.
Can we restrict buyers and uses?
Raise excluded buyers, competitors, customers or uses before signing. These restrictions belong in the agreement. Do not assume every sale needs a new approval: check the terms that apply to your dataset.
Do we keep ownership?
Licensing grants specified rights to use the data, not ownership. You must have the rights to license it, including any necessary third-party permissions. We explain any exclusivity terms before you sign.
Where does our data go?
After a buyer deal is agreed, data goes to the de-identification partner or the buyer’s controlled environment, as specified in the agreement. Preparation happens before AI training. Buyers do not get access to your live business systems.
How much work is involved?
Allow roughly two hours for verification as an initial guide. Older systems, several data sources or extra privacy checks may take longer. We explain the steps, responsibilities and any costs before you commit.
How much is our data worth?
Value depends on useful history, volume, quality, context, rights and buyer demand. Employee count alone cannot set a price. 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.
Maryah · Al Maryah Island, Abu Dhabi, UAE
Questions before a call? info@maryah.ai