The catalyst for physical intelligence in healthcare.
AI models have never seen the inside of a hospital.
Calador drives progress by helping AI teams train models on grounded, real-world data.
- Sterile processingDecontamination, tray assembly, sterilization
- Operating roomsSurgical video, instrument counts, case timing
- Intensive careBedside monitoring, ventilation, rounds
Their models learned from the open internet, where the most critical healthcare data does not exist.
The OR, the ICU, sterile processing. Every floor runs work no model has watched.
Calador de-identifies, evaluates and provides critical case data.
Their models learned from the open internet, where the most critical healthcare data does not exist.
The OR, the ICU, sterile processing. Every floor runs work no model has watched.
Calador de-identifies, evaluates and provides critical case data.
Calador helps the teams building physical AI train and evaluate their models on de-identified clinical video from hospitals and surgery centers. Partners keep ownership and set the terms.
How it works.
Partner institutions
Cases
Data already on your servers.
Calador
De-identification
Faces, voices, on-screen text and metadata removed.
Model makers
Model training
Licensed to approved model makers for training and evaluation.
Partner institutions
Revenue
A revenue share on every license.
For hospitals and for AI labs
What a record contains
- Video
- Aligned to one clinical timeline
- Context
- Op notes, imaging, labs and medications
- Removed
- Faces, voices, on-screen text and metadata
- Use
- AI training and evaluation only
For AI labs
Grounded, real‑world clinical data.
Train and evaluate your models on de‑identified clinical video, aligned with op notes, imaging, labs and medications.
For hospitals
Your data. Your terms.
Earn extra revenue on the data sitting on your servers. Never lose ownership.
The Calador relationship
- Full legal and compliance reviews
- HIPAA and SOC 2 compliance, with full textual and video de‑identification
- Complete discretion
- Opt out anytime
- Control over who sees your data
Why the data matters
The gap is data.
A small model trained on surgical video finds the exact instrument in frame more than twice as often as a leading general-purpose model.
A fraction of the size. Twice the accuracy.
- Leading general-purpose model
- 23%
- Small model trained on surgical video
- 52%
60%+
of US surgeries are now done through a camera. Almost none of that footage has reached the people building AI.
Benchmarks built by clinicians. Run against the frontier.
Labs need clinicians to define tasks and grade results. Calador builds the benchmarks that show whether physical AI is ready for the hospital.
Evaluating models from
We’ve done this before.
Our team of surgeons and healthcare data leaders has spent years putting data agreements in place with hospitals and health systems around the world.
Discretion by default. We never name a partner institution without its written permission.
60+
healthcare institutions
12+
countries
10,000+
hours of healthcare video
Questions partners ask first.Can't find your answer? Join the waitlist and we'll get back to you.
Is it secure?
Yes. Video is de-identified before licensing, encrypted at rest and in transit, and handled under HIPAA with a BAA. Our controls are audited under SOC 2 Type II. Only approved model makers can license data, and they are contractually barred from re-identifying, linking, redistributing or publishing it.
Is this selling patient data?
No. Video is de-identified before licensing, and you grant limited rights for AI training and evaluation. You keep ownership.
What gets removed during de-identification?
Faces, voices, on-screen text and metadata are removed before anything is licensed. You sign off on the process.
Does anything leave our building before we approve it?
No. Before anything leaves your building we agree on which procedures, dates and surgeons are included, permitted uses, anything out of bounds, and retention and deletion terms.
Do our IRB, compliance and legal teams get a say?
At every step. We start with your requirements and build the program around them.
Who sees our name?
No one, unless you choose to be credited on the benchmarks you help build. We never name a partner without written permission.
What if we want out?
Retention and deletion terms are agreed up front. Leaving is as clear as joining.
What does it cost us?
Time to scope and review the program. Curation, de-identification, enrichment and contracting are on us.
How do we get paid?
You earn a revenue share on every license. The same dataset can be licensed again. Your surgeons can also be paid as expert reviewers for clinical evaluation work.
Who can license the data?
Only approved model makers, for defined uses: AI training and evaluation.