01
Tell us what the model is missing.
Describe what the model needs to learn, how much data it takes and when you need it. A call or a page of notes is enough.
- Modality and volume
- Coverage: regions, languages, edge cases
- Format, rights and deadline
Tell us the data you need. We'll tell you where it is and what it costs.
We scope your dataset for free and send a brief with every source, a sample of each and the full price. Approve it, and we buy the data, clean it and deliver it as one dataset.
We get AI labs the data the internet doesn't have.
Teams training new models need data nobody has published. It sits inside companies, or it hasn't been recorded yet. We find it, buy it from every holder, record what's missing and deliver it to you as one licensed dataset.
From companies
Logs, records, video and audio held by businesses that have never sold them.
Recorded for you
New first-person video from people wearing cameras and smart glasses, filmed to your spec.
02
Get a priced brief, free.
We come back with every source we found, a sample from each one, the total cost and our fee. You pay nothing to see it.
- Every source, named and priced
- Sample data from each source
- Our fee on its own line
03
Approve it. We buy the data.
You sign one contract with us. We negotiate a license with every seller and film anything that doesn't exist yet on cameras and smart glasses.
- One contract and one invoice
- We license from each company
- We record what doesn't exist yet
04
Receive one clean dataset.
We merge the data from every source, remove duplicates and personal data and check it against your spec. Then we deliver it in your format.
- Duplicates and personal data removed
- Checked against your spec
- Source and license for every record
See the full price before you commit.
Every scope ends in a one-page brief like this. Each line comes with a sample of the data. If the numbers don't work for you, you walk away and owe nothing.
Example · first-person video of warehouse picking
Placeholders| Source | What you get | Volume | Cost |
|---|---|---|---|
| [N] logistics companies | Existing picking footage and logs | [N] h | $[N] |
| New recordings, smart glasses | Workers filmed on [N] sites | [N] h | $[N] |
| New recordings, head cameras | Close-up hand and shelf views | [N] h | $[N] |
| Data cost | $[N] | ||
| Indri fee | $[N] | ||
| Total | $[N] |
Data that already exists, and data we record for you.
- First-person videoRecorded on smart glasses and head cameras
- Operational logsTelemetry, transactions and system records
- Calls and meetingsAudio with transcripts
- DocumentsContracts, forms and correspondence
- Sensor and robotics dataMachines, fleets and teleoperation
- Clinical recordsEncounters and notes, de-identified
Rebooting a freight network.
Say a mobility company is relaunching its freight business and needs data to train dispatch and pricing models. That data sits with thousands of independent truck carriers, each in its own format. We license records from every carrier and deliver them as one dataset.
2,400Carriers mapped
312Licensing agreements
11 wkScope to first delivery
Hold data an AI lab would pay for? We'll find the buyer.
Sell your dataTell us what your model can't learn yet.
Send us the scope. We'll come back with the sources and the full price, free.