For AI labsPartnersTrustResearch

For founders, operators & liquidators

Turn your company's history into revenue.

Martal Group licenses the Slack threads, tickets, docs and code that show how your team actually worked, anonymized in-house and delivered to vetted AI labs.

Talk to a specialist about what you have, or get a range in under a minute.

Dataset evaluation

IN PROGRESS

Connection depth across systems

  • Slack92%
  • Gmail78%
  • Jira70%
  • GitHub64%
  • HubSpot41%
6 systems connectedPII scrubbed5+ years of history
Typical licence
$20K–$400K+
Time to payment
2–4 weeks
Your time
~1 hour
Raw data deleted
30 days

What we license

Two kinds of data. Both worth real money.

Public text and code have already been scraped. What AI labs can't get is how real teams worked: the conversations, decisions and code behind every outcome. Your company has both.
SGNJ
6 workspaces48 unique contributors5+ years of dataMapping identities…Total data size

Workspace data

The operational record of how your team worked: chat, email, documents, tickets, CRM and support, across Slack, Google Workspace, Notion, Jira, HubSpot and more.

def reconcile_invoices(batch):

# retry failed rows, see BILL-482

for row in batch.failed():

ledger.post(row, idempotent=True)

return batch.summary()

3,400+ commits9 repositories62 contributors

Source code

Private repositories, pull requests, reviews and the tickets behind them. Production code written by a real team to ship to real users.

Are you an AI lab?

License anonymized, multi-tool company data for agent training, RL environments and evaluations.

Explore datasets

RL environments

Post-training

Evaluations

You built something real.
Your data should pay you back.

Years of operational history have real earning potential. Martal values, packages and licenses it to AI buyers, so the company gets paid for work it has already done.

Preliminary range

$85K – $180K

Firm number after evaluation

Data valuation

We review what you have, count it, and tell you what it's worth before you commit to anything.

  • Frontier AI labs
  • RL environment builders
  • Enterprise AI teams
  • Evaluation & benchmark teams

Vetted buyer network

Frontier AI labs, RL environment builders and enterprise AI teams looking for real-world work data.

PERSON_014 flagged the COMPANY_221 renewal. Card ending 4417.

Same person → same token, in every tool.

Personal dataCredentials removedScrubbing complete

Anonymization and PII scrubbing

Identities are tokenized consistently across tools; credentials, card numbers, faces and signatures are removed.

Licence agreement

SIGNED
  • ✓Training-only licence
  • ✓Exclusions annexed
  • ✓Wire to company entity
  • ✓Raw deletion certificate

Licensing and payment

A training-only licence your counsel reviews, exclusions annexed, and payment wired to the company entity.

How it works

Five steps. About an hour of your time.

From first call to payment, we do the work. You decide what's in scope and approve what leaves.
  1. 1

    Intro call

    We map which systems you used and how they connect, and give you a rough range.

    15 min
  2. 2

    Set the scope

    You pick systems and date ranges. Exclusions are annexed to the contract before any data moves.

    You decide
  3. 3

    We evaluate

    Read-only access lets us count volume, coverage and connection depth. Then you get a firm number.

    < 1 hour
  4. 4

    We anonymize

    Every person, company and account becomes a consistent token. You review a sample before delivery.

    Our work
  5. 5

    You get paid

    Payment to the company entity. Raw archive deleted within 30 days, with a certificate.

    2–4 weeks

Estimate

What is your data worth?

Three questions for a preliminary range. Connected data is worth more than big data.
How many people worked at the company?
How many years of history do your systems hold?
Which systems did the team work in?

Preliminary range

$50K – $105K

  • →Connected beats big. The same project traced across chat, tickets and code is worth more than any one tool on its own. Your mix qualifies for a connected-data uplift.
  • →Longer history and more people mean more complete, multi-step workflows.
  • →This is not an offer. We count before we quote — the firm number comes after evaluation.
Get a firm number →

Anonymization

Anonymization isn't a step. It's the product.

One identity map across every tool keeps workflows intact while identities disappear. Faces, logos and signatures are blacked out; credentials and card numbers are removed. How our pipeline works

Same person → same token, across every tool.

# eng-billing · Slack

PERSON_014

@PERSON_027 the COMPANY_221 invoice batch failed again — 1,284 rows. Customer contact is PERSON_390 (EMAIL_390).

PERSON_027

Rolling back PR #2213. Opening BILL-482 and I'll call her at PHONE_390 before EOD.

Re: March invoices · Email

From: PERSON_390 <EMAIL_390> To: PERSON_014 Hi PERSON_014 — our AP team flagged duplicate charges on the March invoices. Card ending 4417. Can you confirm a credit before our close on the 5th? — PERSON_390, COMPANY_221 ADDRESS_118

Dataset profiles

Every company has a different shape.

What matters is how many systems your team used and how they connect: people, years of history and the workflows that run through them.

DATASET KESTREL

Logistics SaaS

People
180
Years
5
Tools
9

Customer escalations traced from support ticket to code fix to account update.

  • Slack
  • Gmail
  • Jira
  • GitHub
  • HubSpot
  • Zendesk
  • Notion
  • Sheets
  • AWS

DATASET HERON

Construction tech

People
220
Years
4
Tools
7

Field-to-office workflows: RFIs, change orders, scheduling and billing.

  • Google Workspace
  • Slack
  • Asana
  • Airtable
  • GitHub
  • Salesforce
  • Procore

DATASET WREN

Fintech (wound down)

People
65
Years
6
Tools
7

Compliance reviews, month-end close and incident postmortems.

  • Slack
  • Outlook
  • Linear
  • GitLab
  • Confluence
  • QuickBooks
  • Intercom

Our commitments

In the contract, not the marketing.

✓

Training-only licence

Buyers may train and evaluate models. Never market to, profile or contact anyone in the data.

✓

No third party touches raw data

Anonymization runs in our own environment. Labs only receive the code-named dataset.

✓

You approve a sample first

Nothing is delivered until you've reviewed anonymized output from your own data.

✓

Raw deletion in 30 days

Written into the contract, with a deletion certificate when it's done.

✓

Exclusions honored

Anything you scope out (a client, a channel, an HR folder) never leaves your systems.

✓

Employee privacy by design

HR, health, payroll and personal channels are excluded by default.

Who works with us

Operating, winding down or closed.

Any company with real operational history can license it. For closed companies, an authorized director, liquidator or assignee signs on the entity's behalf.

Operating companies

License a slice of your history as a new revenue line. Keep using every tool as normal.

Companies winding down

Turn systems you're about to delete into cash for creditors and shareholders.

Closed companies

Directors, liquidators and assignees can sign for the entity. We ask for authority up front.

Advising a company that's shutting down? See our partner program

FAQ

Your questions answered.

Am I allowed to sell this data?
In most cases, yes. The company owns its operational records. We license rather than sell, for training and evaluation only, and your counsel reviews the agreement before signing. Customer contracts with confidentiality terms are checked during scoping and excluded if needed.
Can a buyer tell the data came from us?
Buyers receive a code-named dataset with every person, company, address and account replaced by consistent tokens. You review a sample before delivery, and the licence prohibits any attempt at re-identification.
Who sees our raw data?
Only our anonymization team, inside our own environment. No buyer and no subcontractor ever receives raw records.
We're winding down. Who can sign?
A director, officer, liquidator or assignee with authority over company assets. We ask for the authority document up front, and payment goes to the company entity so it can flow to creditors and shareholders properly.
How much will we get, and how is it set?
Deals run from $20K to $400K+. The number depends on how many systems you used, how well they connect, years of history and headcount. We don't quote a firm number until we've counted.
How much of our time does this take?
About an hour in total: an intro call, granting read-only access, and reviewing the anonymized sample. No engineering work.
What about our employees' privacy?
Personal and sensitive categories (HR, medical, payroll, DMs you exclude) are out of scope by default, and every remaining identifier is tokenized. We're glad to share our employee data standard with your team or board.
What data are you looking for?
Operational data that shows how work gets done: chat, email, docs and wikis, tickets and projects, code and pull requests, CRM, support and finance workflows. The more of these that connect, the better.

Find out what your data is worth.

Connect with a Martal specialist. No commitment, and we handle the rest.