Sigveil Match
Current prototypeCompare audience overlap across organizations using blinded identifiers and privacy-adjusted aggregate reports.
Private collaboration / Sigveil Match
Find common ground.
Keep your records private.
Understand the audience you share with a partner. Process matching identifiers in your browser and receive an aggregate overlap estimate with privacy controls.
See how a real match works
Your matching records stay local
Identifiers are normalized and blinded in your browser before token exchange.
Shared answers, with privacy controls
Minimum group sizes, statistical noise and a query budget constrain reporting.
A report you can understand
See estimated overlap and the remaining privacy budget in one place.
Your records
→ blinded tokens →
Shared insight
← blinded tokens ←
Partner records
01 / Your workspace
Inside the working app
Two lists. One shared answer.
A campus partnership, using synthetic records.
Actual app screens. Actual matching code.
- 01
Create a match room
Name the partnership, choose a privacy setting and copy the invite for your partner.
- 02
Prepare each list locally
Each organization selects its own file. The browser normalizes and blinds the matching identifiers.
- 03
Exchange blinded tokens
Both sides apply their session key to the other side's tokens. Matching records produce matching double-blinded tokens.
- 04
Read the aggregate report
The coordinator checks group sizes and privacy budget, adds statistical noise, then returns an overlap estimate.
The captured run returned an estimate of 83 (41.5%) after statistical noise, with 1.50 privacy budget remaining. This is demo data, not a customer result.
Try the matching engine
Run it. Watch the tokens change.
Two synthetic lists. Two fresh session keys. The same browser worker used in the workspace.
School / 200 records
student0@school.edu
student1@school.edu
+ 198 sample identifiersPartner / 200 records
student0@school.edu
external0@fintech.io
+ 198 sample identifiersNo account or file needed. Both parties run locally for this example; real partners use separate browsers.
- 01 / Prepare the sample lists
- 02 / Blind each list locally
- 03 / Apply the other session key
- 04 / Compare the token sets
Ready when you are. This is a real computation, not an animation.
School records—
Partner records—
Synthetic overlap—
Time this run—
This local example shows the synthetic intersection before privacy noise. The workspace applies group-size and privacy-budget controls before releasing its aggregate estimate.
One platform. Two complementary layers.
The product today, and where it is going.
Sigveil Verify
PlannedEvaluate behavioral evidence of human participation and validate signed attestations before records enter a match. This layer is not implemented yet.
Sigveil Match / Workspace
Your partnerships, in view.
Create a match room and invite a partner to explore your shared audience.
Recent match rooms (up to 25)
0
Completed Matches
0
Matching records processed
Locally
Minimum group size
50
Start a private match
Name the partnership and create a room for both organizations.
Lower epsilon adds more noise. Each report uses part of the shared privacy budget.
Your match rooms
Open a room to prepare local records, invite a partner or view the result.
| Partnership | Partner | Status | Estimate | Action |
|---|
Campaign Room
01 / Room ready
02 / Prepare locally
03 / Exchange tokens
04 / Aggregate report
Choose a local file, or drop it here
Simple CSV or text, one identifier per row. At least 50 unique records per organization. Matching records are processed locally.
0 records loaded locally
Report complete. Intermediate token rows have been deleted from the application's active store. Your aggregate report is retained.
Estimated overlap
—
Overlap / Org A
—
Org A outside overlap
—
Privacy budget left (ε)
—
Counts and percentages include statistical noise. They describe audience overlap, not causal conversion lift or newly acquired customers. Org A is the organization that created this room. Account information, room metadata, audit logs and privacy budgets are retained.