Discover without taking custody
Review a resource description and its limitations. A catalog entry is neither an available dataset nor permission to access it.
THE FIRST PRODUCT WORKFLOW / PLANNED
A research question.
A responsible way forward.
Can the right institution help answer your question? Start by assessing resource suitability and permission, before a study commits to data access or recruitment.
RESEARCH FEASIBILITY
Answer research questions together.
Keep patient identity at its source.
Make every release accountable.
ONE BOUNDED ENGAGEMENT
- Start
- A defined research question
- Work
- Resource discovery, scoped task, review
- Return
- An answer with its evidence and limits
A PRACTICAL FIRST USE
Determine suitability.
Before widening the work.
A life sciences research team or academic collaborator needs to understand what a resource could support. An institution needs to control what is described, computed and released. The proposed pilot brings those decisions into one visible process.
Approve the exact task
Bind purpose, recipient, artifact version, location and expiry. Deposits and bounded proofs support the decision; they do not replace it.
Review what comes back
Assess the output for privacy and scientific interpretation before release. Retain the reason for a refusal as carefully as an approval.
WHAT DISCOVERY IS AIMING TO CONNECT
Useful foundations.
One accountable workflow.
The intended distinction is the complete custody-to-release process. Federation, encryption or AI alone is not a new category or a privacy guarantee.
PLANNED DIVISION
Vault
Describe a resource, preserve custody and version the approved research view.
Explore the divisionPLANNED DIVISION
Clinical
Bring a precise evidence gap and accountable professional interpretation.
Explore the divisionPLANNED DIVISION
Trials
Inform a future study discussion without identifying or contacting participants.
Explore the divisionPLANNED DIVISION
Discovery AI
Constrain a defined task and inspect its inputs and outputs; AI execution is not part of this public demonstration.
Explore the divisionMEASURE BEFORE MAKING CLAIMS
Agree the baseline.
Show the evidence.
These are proposed evaluation measures. Targets would be agreed with each pilot team; no performance results or reductions are claimed today.
Time to an approved answer
From an agreed, complete request to a recorded reviewed answer; include waits and report their causes.
BASELINE AND TARGET: NOT YET SETReviewer effort
Active minutes by research, custodian and privacy roles; report separately from elapsed time.
BASELINE AND TARGET: NOT YET SETReproducibility
Repeat the same versioned task and reconcile the artifact and decision evidence.
BASELINE AND TARGET: NOT YET SETProhibited-request denials
Test every agreed negative case and retain the result. Any unauthorized release stops acceptance.
BASELINE AND TARGET: NOT YET SETTAKE A PRECISE NEXT STEP
Bring the question.
Help define the pilot.
Discuss scope with a research owner, custodian and reviewers. Share only a non-sensitive outline.