HYBRID DISCOVERY
Sign in

ANSWER PRIVACY / SYNTHETIC CONCEPT DEMO

A useful answer.
A deliberate boundary.

Keeping a record private is one job. Deciding what an answer may reveal is another. Explore three invented examples of what could leave a private research workspace.

Try the three cases Understand the scope

No sign-in or data entry. Medical services remain planned.

FICTIONAL LOCAL WORKSPACE

Source information
Local analysis
Proposed answer
Permission and output review
Only an eligible reviewed output
Required architecture, illustrated. No source information or answer is transmitted by this demo.

ONE BOUNDARY / THREE DIFFERENT REQUESTS

What is someone
trying to share?

Choose a fixed case, then change its release conditions. This page explains a decision pattern; it does not assess a real answer or run a model.

All three cases are readable below. Interactive controls require JavaScript.

FICTIONAL CONDITIONS MATCH

Share a reviewed research summary.

Does this manufactured resource contain the fields needed for recovery-pattern research?

This case assumes a completed review of this exact fictional summary for research group B. That assumption is supplied by the illustration; the page does not perform an anonymization assessment.

What stays local
Source rows, fictional identity fields and any linkage material stay inside the illustrated local workspace.
Example output
The manufactured resource describes procedure category, age band and recovery duration. The example grouping has nine aggregate cells across 120 invented records. Clinical validity has not been evaluated.

FICTIONAL RELEASE BLOCKED

Request one person’s private measurements.

Share an individual’s dated measurements with an outside research team.

This request seeks individual-level information outside the custody boundary. The illustration blocks it even when its other conditions match.

What stays local
The individual history stays inside the illustrated local workspace. Sign-in or participation consent does not override Discovery’s required sharing boundary.
Example output
Individual measurements are withheld in this illustration.

PRIVACY REVIEW REQUIRED

Share an answer that could reveal a person.

Share an individual’s detailed recovery timeline and rare-event pattern, with the name removed.

Removing a name does not establish that this pattern is anonymous. A less revealing output would need a separate assessment of the actual information, recipient and release context.

What stays local
The proposed individual-level answer remains inside the illustrated workspace while disclosure is unresolved.
Example output
The proposed detailed answer is held for privacy assessment.

BUILT TO EXPLAIN, BEFORE CONNECTING

A concrete demonstration.
A specific limit.

This is a working, client-side interaction over fixed fictional cases. It includes condition changes, direct case links and downloadable explanatory records. It sends no case selections to a service.

The positive case assumes a prior review; it does not perform one. The detailed single-person answer remains held because disclosure has not been assessed. A fixed rule cannot establish that arbitrary medical information is anonymous.

Vault: preserve custody

The proposed medical integration would keep identifiable originals and linkage material with their custodian and version the reviewed output.

Cortex: bound the execution

The proposed execution layer would limit tools, purpose and destinations. This page runs no AI and connects to no model provider.

Discovery: review the release

The intended workflow would require applicable authority, validated preparation and output assessment. Those medical capabilities remain planned.

AN OPEN RESEARCH QUESTION

What does an answer
reveal about a person?

An answer can reveal sensitive information even when the original record stays private. Exact dates, rare events and combinations of ordinary details can make a person recognisable.

Discovery’s planned approach keeps the purpose, recipient and proposed output in view. This synthetic demo illustrates why a reviewed summary, individual measurements and an unresolved disclosure risk need different decisions.

Explore Discovery’s privacy approach

Review what the answer reveals

  1. Choose one kind of answer and its intended recipient.
  2. Agree what the recipient must not be able to infer.
  3. Test useful and disallowed cases with manufactured data.
  4. Record failures, uncertainty and what would need independent review.
Explore the research workflow

EXPLORE THE WIDER WORKFLOW

The answer is one handoff.
Keep the whole path visible.

Follow the fictional feasibility workflow or inspect the required privacy boundary.

Follow question to answer Read the privacy requirements Check current availability