Real Business Processes for AI Labs
We source licensed, de-identified records of how real businesses work for frontier labs, by sector and region, from firms in the US, Canada and the UK. Every business approves in writing what it shares.
Where the Data Comes From
Our clients are businesses across many industries with years of records that show how their work gets done, from the first request to the finished job. We represent the businesses, so your request reaches each owner through a firm they already work with, and we handle consent, scope and de-identification checks with them.
Why This Data Is Hard to Get
- 01
Provenance is hard to show
When records come second-hand, it's hard to say who agreed to their use or what they agreed to. You need a clear line back to the business that approved them.
- 02
Owners hesitate to sign
A licensing request from a company the owner has never dealt with can stall, because they want someone they already trust to explain the scope, the de-identification and the terms.
- 03
You need records from outside the US
Canadian and UK firms keep their books under their own tax rules and write in their own spelling. Their records show your models how businesses outside the US run.
How Sourcing Works
A dataset starts with your request and goes ahead only with each owner's written approval. Typical time from request to delivery: typical timeline.
- 1
You send a request
Tell us the sectors, regions, processes, date ranges and volumes you need. We'll tell you which of them our clients can supply.
- 2
We approach businesses that fit
We explain the request to each owner in plain terms and answer their questions. A business takes part only if its owner approves in writing.
- 3
The owner sets the scope
Each owner chooses which systems and date ranges are in and what stays out. The scope and the approval are written down before anything moves.
- 4
Checked, then licensed to you
We check the de-identified copy with the business against the approved scope. You receive it under a written licence the business has signed.
Records From Three Markets, Prepared With Consent
Our clients hold years of records that show how their businesses run, across the US, Canada and the UK. With their written approval, our data partners prepare de-identified copies with client details taken out.
What You Can Request
- Records of real business processes: requests, handoffs, drafts and approvals
- Client names, contact details and other personal information taken out
- Records from firms in the US, Canada and the UK, across many industries
- Written approval from every business before anything is shared
- A written scope for each business: systems, date ranges and exclusions
- Sectors, regions and volumes confirmed with you for each request
Common Questions
Still deciding? On a free 30-minute call we'll tell you what fits.
Do you license data to labs directly today?
Today our clients' deals run through our data partners, and we represent each business in that process. Over time we plan to work with labs directly, which is why we'd like to hear what you need now.
Is client data part of the dataset?
No. What we source shows how work gets done. The data partner on each deal takes out client names, contact details and other personal information, and each owner can leave whole systems out of the scope.
How is consent handled?
Each business approves in writing before anything is shared, covering which systems, which date ranges and what stays out. We take the owner through the request first, so the person approving knows what the copy will contain.
Who does the de-identification?
The data partner on each deal de-identifies the copy, and the standard is written into the licence. On the business side, we check the de-identified copy against the scope the owner approved before it's delivered.
Can you source a specific sector or region?
Yes, within our client base, which covers many industries across the US, Canada and the UK. Send us the details and we'll tell you which parts we can supply and how long that would take.
Can people from these industries review or label data?
Possibly. People who work in a sector know its records and could review or label them, with their employer's approval. Tell us if that would help your team.
Tell Us What You Need
Email the sectors, regions, processes and volumes you're after to support@mccarthys.ai, or book a call. We'll come back with what our clients can supply and a realistic timeline.